本篇博文主要展示每日从Arxiv论文网站获取的最新论文列表，以自然语言处理、信息检索、计算机视觉等类目进行划分。
统计
今日共更新1441篇论文，其中：
- 自然语言处理176篇
- 信息检索40篇
- 计算机视觉233篇
自然语言处理
1. 【2610.40655】Elucidating the Design Space of LM Data Attribution
链接：https://arxiv.org/abs/2610.40655
作者：Steven Yu,Sheng-Yu Wang,Kimia Hamidieh,Antonio Torralba,David Bau
类目：Computation and Language (cs.CL)
关键词：Language Models, typically evaluated, evaluated using downstream, downstream tasks, LLM data attribution
备注： Project Page: [this https URL](https://catt.baulab.info/)
点击查看摘要
Abstract:Data attribution methods for Language Models (LMs) are typically evaluated using downstream tasks like fact retrieval. However, existing evaluations vary multiple dataset properties simultaneously, making it impossible to isolate exactly why an attribution algorithm fails. To diagnose these failures, we explore the design space of LLM data attribution, with a primary focus on evaluation. We introduce the Controlled Attribution Benchmark (CATT) for fact tracing, in which we fine-tune LLMs on entirely fictitious knowledge, providing unambiguous ground truth for attribution. This synthetic setup allows us to systematically alter one dataset configuration at a time, such as document redundancy and lexical distractors. With this configurable evaluation, we demonstrate that redundancy and distractors consistently degrade attribution performance. Rather than proposing a new estimator, we examine two design choices that are underexplored: the loss used to compute gradients and how per-token attribution scores are aggregated into a document score. We find that the true causal evidence within a training document is highly localized, but standard methods dilute this evidence into noise by averaging attribution scores across the entire document. We find that replacing standard cross-entropy loss gradients with the negative log-odds gradients surfaces the critical tokens from other uninformative ones. We can then aggregate only the top token scores to improve ground truth recall by 35 to 72 percentage points across multiple baselines on our diagnostic benchmark. The negative log-odds correction also generalizes to FTRACE-TREX, improving counterfactual retraining performance for every method we test, by up to 4.9x.
2. 【2610.40627】ShallowDefend: Harnessing Shallow Safety Alignment for Multi-Turn Defender
链接：https://arxiv.org/abs/2610.40627
作者：Qiyao Ma,Xiaodong Liu,Weiwei Yang,Mingqian Feng,Chengquan Guo,Jialin Song,Carolyn Buractaon,Zhe Zhao,Junshan Zhang,Sandeep Atluri,Christopher White,Jianfeng Gao
类目：Computation and Language (cs.CL)
关键词：harmful requests directly, refuse harmful requests, enabled large language, reliably refuse harmful, large language models
备注：
点击查看摘要
Abstract:Safety alignment has enabled large language models to reliably refuse harmful requests directly, yet the same request distributed across a multi-turn conversation still succeeds far too often. This brittleness has a mechanistic root: Alignment is shallow, with safety behavior concentrated in a model's first few output tokens, and existing defenses inherit the weakness, supervising the surface form of the response without ever giving the model an explicit safety decision to learn. We take the opposite view of shallowness: if a short span determines the trajectory of an entire response, that span is the highest-leverage place to intervene, and the defender can occupy it first. We propose Assess $\to$ Decide $\to$ Respond, a framework in which the defender assesses the user's intent, commits to an explicit, parseable safety decision, and only then generates a reply conditioned on it, with the commitment trained by reinforcement learning at the payload turn. Because the decision is a token span, the reward can be a rule: a self-verified reward checks the committed label against the conversation's known nature, requiring no judge and no annotation, while the same format also supports an LLM-as-judge reward over a graded decision space with safe redirection. Across two model families and fourteen benchmarks spanning safety, utility, and helpfulness, our defenders achieve the strongest multi-turn safety in a controlled comparison, maintaining their lead in safety while preserving utility at a modest cost to helpfulness. A per-token KL analysis and interventional probes confirm why this works: training concentrates its effect on the brief deliberation that sets a response's trajectory, and the decision committed there is a causal control point. The very shallowness attackers exploit becomes the mechanism of defense.
3. 【2610.40619】stimony from Timelines: Evidence-Grounded Significance Testing of Temporal Clinical Hypotheses
链接：https://arxiv.org/abs/2610.40619
作者：Chengyang He,Wenlong Zhang,Koduvayur Subbalakshmi,Ping Wang
类目：Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：side effect tends, sales tend, tend to dip, effect tends, tends to follow
备注：
点击查看摘要
Abstract:Ask a language model whether sales tend to dip after a product recall, or whether one side effect tends to follow another after a vaccine, and it answers in a fluent sentence that reads like a finding, with no test behind it. Assertions about how events unfold across many cases fill the Web, and each is a temporal hypothesis that only a significance test over the right records can settle. The verifier must specify the null, select the test and decide which records bear on the hypothesis, and any of these choices can go wrong while the conclusion still comes out right. Yet benchmarks for language-model data analysis grade only the conclusion, with the relevant data handed over. We study the problem in health, where a wrong verdict is costly, and present STRATA, a benchmark that grades every decision on the way to a verdict: 8,735 hypotheses over the clinical timelines of vaccine adverse-event reports, with gold plans, evidence sets, p-values and verdicts derived under a released codebook and reproduced by a deterministic oracle. We also propose ART (articulate, retrieve, test), a staged paradigm that exposes each decision as an inspectable output. Extensive experiments with five model configurations show that systems fail mostly at choosing which records count, a verdict-evidence gap that accuracy alone conceals. ART closes most of this gap and brings a 31B open model to within 0.01 of the frontier models under the same paradigm; the frontier models succeed once given the codebook.
4. 【2610.40613】xtSeal2: Distortion-Free LLM Watermarking with Copula Mixing and Memoryless Races
链接：https://arxiv.org/abs/2610.40613
作者：Tom Sander,Pierre Fernandez,Itai Gat,Haiming Bao,Jiaxin Shi,Jason Park,Valeriu Lacatusu,Sylvestre-Alvise Rebuffi,Hongyan Chang,Hady Elsahar,Tomáš Souček,Tuan Tran,Surya Parimi,Alexandre Mourachko
类目：Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：replace rejected drafts, rejected drafts remain, drafts remain watermarked, large language models, distortion-free watermark
备注： Code: [this https URL](https://github.com/facebookresearch/textseal)
点击查看摘要
Abstract:We present TextSeal2, a distortion-free watermark for large language models that keeps its signal under speculative decoding: tokens that replace rejected drafts remain watermarked with the same keys, and the acceptance rate is unchanged. It also preserves output diversity and adds no measurable inference latency. Its predecessor, TextSeal, restores diversity in the Gumbel-max scheme by routing each token to one of two keys via a coin flip. This is suboptimal because the detector averages the scores from both keys without knowing which key was used. TextSeal2 instead merges two or more keys into a single virtual key and couples it to fresh noise through AutoMark's Gaussian copula. Any number of keys, including publicly shared keys, can be merged this way without diluting detection. Under speculative decoding, TextSeal2 verifies keyed drafts with one more exponential race, run on the leftovers of the draft races. By memorylessness, this race reproduces block verification, so the output follows the target model exactly. We evaluate TextSeal2 on three LLMs. At equal output token diversity, TextSeal2 detects 8 to 22 percentage points more 100-token answers than TextSeal at a false-positive rate of $10^{-3}$, and is comparable to textGrain while requiring no post-logits optimization. Under speculative decoding, it detects 18 and 22 percentage points more 200-token answers than TextSeal and textGrain, respectively, at the same acceptance rate.
5. 【2610.40611】When to Assist and What to Suggest: Joint Gating and Memory-Grounded Hint Generation in One Compact Decoder
链接：https://arxiv.org/abs/2610.40611
作者：A. J. A. Jubayer Talukder,Mohammad Ariful Haque
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
关键词：Proactive assistants offer, assistants offer short, Proactive assistants, offer short, assistants offer
备注： 20 pages, 10 figures, 13 tables; supplementary material included (Sections S1-S8)
点击查看摘要
Abstract:Proactive assistants offer short, memory-grounded hints during a conversation, deciding when to assist and what to suggest. Current systems split this between a gate and a larger generator; we evaluate both together on 28,058 decision points. Stage-wise generation scores are oracle-gate ceilings: behind a real gate, end-to-end exact match is 0.2619 against a ceiling of 0.2971, and end-to-end precision is 0.2387, leaving substantial headroom. A gate pays off through precision and compute, whose saving falls from 4.39x with an 8B generator to 1.38x with a 1B one. The memory profile drives hint quality: removing it costs 24.7 exact-match points (rescaled BERTScore 0.57 to 0.35), while retrieval and a larger generator add nothing on exact match. One 1B decoder can do both, gating from an intermediate layer and generating in the same pass. At its final checkpoint it matches or exceeds the two-model pipelines on automatic end-to-end measures with 1.31-1.76x less wall-clock time than the matched 1B pipeline; across three seeds, its deeper exit ties the separate gate. A human evaluation (1,144 ratings over 32 dialogues) supports the memory effect and rates the single model level with the matched 1B pipeline and below the 8B one. Graded metrics and LlamaPIE's native judge track human ratings more closely than exact match, which we report as a strict lower bound.
6. 【2610.40578】VeriCUA: Verification as Agentic Exploration for Computer-Using Agents
链接：https://arxiv.org/abs/2610.40578
作者：Qiushi Sun,Da Yu,Alexander Knop,Amer Sinha,Lingpeng Kong,Ben Kao,Ravi Kumar,Pritish Kamath
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：scaling computer-using agents, requires reliable task-completion, web environments requires, environments requires reliable, Building and scaling
备注： Work in progress
点击查看摘要
Abstract:Building and scaling computer-using agents (CUAs) across desktop, mobile, and web environments requires reliable task-completion verification for data curation, evaluation, and reinforcement learning. Typical automated approaches prompt vision-language models (VLMs) to judge trajectories or generate programs to inspect application state. However, both rely on partial views of complex digital environments. VLM judges are restricted to supplied observations, while generated programs depend on accessible state. These views can omit decisive evidence or leave completion judgments unsupported. Tailoring separate verifiers to different platforms and access settings adds further engineering cost. We present VeriCUA, a platform-agnostic framework for verification through agentic exploration. It treats verification as dynamic evidence gathering and assessment within a shared procedure for online and offline settings across platforms and observation modalities. After inspecting the available context to construct a task rubric, the verifier selects GUI or code checks for unresolved requirements and combines their findings under explicit evidence rules. Experiments show improved agreement with reference labels over the compared baselines across these settings, with the largest offline gains on hard cases and stronger rejection of false completion claims online. Further analyses connect these gains to complementary evidence and test-time exploration.
7. 【2610.40577】Multi-View MoE: Letting Every Router Shape Expert Computation
链接：https://arxiv.org/abs/2610.40577
作者：Zelin Zhao,Tianchi Ma,Junyuan Shang,Shuohuan Wang,Dianhai Yu
类目：Computation and Language (cs.CL)
关键词：expand parameter capacity, models expand parameter, Expert, expand parameter, parameter capacity
备注：
点击查看摘要
Abstract:Sparse mixture-of-experts (MoE) models expand parameter capacity by activating a small subset of experts per token. Expert selection directly affects model capabilities under a fixed active-expert budget. Dense evaluation of all experts reveals weak agreement between single-router rankings and expert-utility evaluation rankings, especially in middle and deep layers. Richer signals for selection and computation incur additional cost: selection based on internal expert activations requires preliminary computation, while processing multiple full representations separately through the same expert repeats its computation. We propose Multi-View MoE to improve expert selection with multiple routers and reuse their preferences to modulate expert computation under a fixed expert budget. Its Multi-Router submodule aggregates sigmoid scores from independently initialized routers. These routers develop highly differentiated expert preferences during training, allowing experts strongly favored by one router or jointly supported by several routers to be selected. Multi-View further uses each selected expert's relative scores across routers to mix learned latent views into its input, reusing routing preferences for both selection and computation. Each selected expert executes one forward pass, preserving the active-expert count, expert FLOPs, and all-to-all communication volume. On a 24-layer 10B-A1B model trained on 708B tokens, full Multi-View MoE improves the nine-task downstream mean by 1.18 percentage points over the baseline. Multi-Router alone yields a 0.6-percentage-point gain and improves the rank correlation between router scores and evaluated expert utility, with relative increases of 8.1% globally and approximately 58% in the layer with the lowest single-router alignment. Controlled mixing interventions show the model uses input differences that depend on both the expert and the token.
8. 【2610.40576】PIVOT: Progressive Intervention via Online Teacher Correction
链接：https://arxiv.org/abs/2610.40576
作者：Zhihao Yang,Boyuan Zheng,Puhui Tan,Yukun Chen,Kuan Zhang,Longze Chen,Jiaming Li,Renke Shan,Lu Wang,Xianpei Han
类目：Computation and Language (cs.CL)
关键词：sparse outcome-level feedback, Reinforcement learning, group fail, verifiable rewards, limited by sparse
备注： 19 pages, 9 figures, 6 tables. Project page: [this https URL](https://zhihaoyang30.github.io/PIVOT/)
点击查看摘要
Abstract:Reinforcement learning from verifiable rewards (RLVR) is limited by sparse outcome-level feedback on hard reasoning tasks: when all rollouts in a group fail, rewards provide no learning signal. Stronger teachers can provide additional guidance, but this raises a central question: how can such signals be incorporated into RLVR to support successful rollouts while remaining learnable for the student? We distinguish two complementary mechanisms: trajectory-level intervention, which improves access to reward-bearing trajectories, and token-level supervision, which provides dense targets to guide student learning. Our controlled experiments show that either mechanism alone can remain insufficient, motivating their combination. We therefore introduce PIVOT, which inserts teacher-generated corrections into student rollouts and applies token-level KL supervision to these segments alongside the RLVR objective. Beyond their combination, we systematically analyze where to intervene and how to supervise, yielding a recipe for effective guidance. Across hard logic games and mathematical reasoning benchmarks, PIVOT consistently outperforms existing methods, nearly doubling mean avg@8 on logic games from 19.5% for the best-performing baseline to 35.6%.
9. 【2610.40574】Every Token Needs Its Own Teacher: Multi-teacher Adaptive On-policy Self-distillation
链接：https://arxiv.org/abs/2610.40574
作者：Yifu Luo,Chuang Ma,Zeyu Chen,Enkelejda Kasneci,Qi Chen,Gjergji Kasneci
类目：Computation and Language (cs.CL)
关键词：post-training large language, large language models, On-policy Self-distillation, Adaptive On-policy Self-Distillation, proven effective
备注： Preprint
点击查看摘要
Abstract:On-policy Self-distillation (OPSD) has proven effective for post-training large language models by constructing a teacher conditioned on privileged references. However, existing approaches keep the teacher fixed either throughout training by consistently exposing it to the full reference, or within each response by adaptively varying partial reference exposure only across responses. In this paper, we argue that neither of them is sufficient. Instead, every token needs its own teacher, as the teacher preference can vary token by token within the same response. Through a teacher-intervention pilot study, we validate that the teacher preference differs across tokens, and reveal that a preferred teacher can be identified for each token through majority voting and KL-based selection. Motivated by these findings, we propose Multi-teacher Adaptive On-policy Self-Distillation (MTA-OPSD), the first token-level adaptive OPSD method. Our method constructs a multi-teacher pool consisting of candidate teachers, and enables every token to adaptively select its preferred teacher during training. Experiments with Qwen3 models on mathematical reasoning tasks show that MTA-OPSD consistently outperforms existing OPSD approaches with an average gain of up to $+5.9$ over the base model, highlighting the importance of adapting teacher supervision at the token level. The code is available at this https URL.
10. 【2610.40563】Comparing Human Role-Players and LLM-Simulated Clients in Online Counselling Training: An Analysis of Counselling Patterns
链接：https://arxiv.org/abs/2610.40563
作者：Eric Rudolph,Philipp Steigerwald,Jens Albrecht
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Large Language Models, Large Language, Language Models, capabilities of Large, Models to simulate
备注： Published at EDM 2025 (18th International Conference on Educational Data Mining), short paper. 7 pages
点击查看摘要
Abstract:This study investigates the capabilities of Large Language Models to simulate counselling clients in educational role-plays in comparison to human role-players. Initially, we recorded role-playing sessions, where novice counsellors interacted with human peers acting as clients, followed by role-plays between humans and clients simulated by Mistrals Mixtral 8x7b using 4-bit quantization. These interactions were analysed with a counselling communication pattern system at sentence level. We investigated two key questions: (1) to what extent LLM-generated responses replicate authentic conversational dynamics and (2) whether counsellors' communication behaviour differs when interacting with human versus LLM-simulated clients. The findings highlight both similarities and differences in the application of counselling patterns across scenarios, showing the potential of LLM-based role-playing exercises to enhance counselling competencies and to identify areas for further refinement in virtual client simulations.
11. 【2610.40551】Ramanujan Challenge Results: AI-Assisted Evaluation Pipeline for Multi-Format Proofs
链接：https://arxiv.org/abs/2610.40551
作者：Alan Sorani,Ashvni Narayanan,Michael Shalyt,Tali Monderer,Rotem Kalisch,Hila Barkan,Elyasheev Leibtag,Shachar Weinbaum,Carsten Schneider,John Campbell,Ido Kaminer
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Symbolic Computation (cs.SC); Number Theory (math.NT)
关键词：mathematical capabilities rapidly, capabilities rapidly improve, Computer Algebra Systems, capabilities rapidly, shifting from university-level
备注：
点击查看摘要
Abstract:As AI's mathematical capabilities rapidly improve, meaningful benchmarks are shifting from university-level problems and math olympiads to research-level mathematics. The Ramanujan Challenge is a benchmark comprised of research-level problems in number theory, including unpublished results and open conjectures. It presented concrete formulas for mathematical constants whose correctness can be checked numerically, while their proofs may require non-obvious mathematics. Here, we analyze the Challenge submissions, measuring the relative success of different AI models, Computer Algebra Systems (CAS), and formal tools. We find correct submissions that are conceptually distinct from the official solutions, and a proof of one of the published open conjectures. Evaluating the submissions exposed a second challenge: judging complex proofs reliably at scale. We introduce an AI-assisted evaluation pipeline for natural language proofs, CAS code, and Lean formalizations that assesses semantic and logical correctness. Compared with the conventional one-shot LLM judgment, our pipeline reduces false-positive verdicts on natural language proofs from 23% to 6% and improves judgment accuracy on supporting code from 46% to 58%. This approach provides a methodology for evaluating research-level mathematics at scale, with potential applications across mathematics and AI for science.
12. 【2610.40541】On the Fundamental Limits of Prefill-Based Decoding Length Prediction
链接：https://arxiv.org/abs/2610.40541
作者：Haowen Shi,Xuzhe Yan,Fahao Chen,Peng Li
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Theory (cs.IT)
关键词：prefill-phase hidden states, large language model, hidden states, inference scheduling, tokens a large
备注： 26 pages, 4 figures
点击查看摘要
Abstract:Predicting how many tokens a large language model (LLM) will decode, using only prefill-phase hidden states, is a key primitive for inference scheduling. We show that current benchmarks conflate predictor quality with task hardness: when the target model, dataset, or sampling configuration changes, reported accuracy collapses even though the predictor is unchanged. We trace this instability to an overlooked error source---the inherent stochasticity of autoregressive decoding---and derive bounds that every prefill-side method must obey. Our analysis treats decoding as a stopped stochastic process. The per-step stopping probability of a response is not constant but ramps up over hundreds of steps, so the decode length inherits a central limit theorem from the many local decisions that determine where a response ends; this yields per-prompt normality with an explicit error term, together with an affine variance law $\sigma=\alpha\mu+b$ whose slope is the coefficient of variation of the number of response segments. Empirically, the normal law reproduces the per-prompt MAE and the central mass of the decode-length distribution to within a few percent, and the residual departures from normality (mild right skew, slight excess kurtosis) are the ones predicted by the sampling-noise correction. Under fixed model, data, and sampling conditions, a broad design space of predictors (MLP, EGTP, FiLM, soft-label, and Mixture-of-Experts variants) converges to similar performance close to the bound, confirming that \emph{system error dominates model error}. We argue for a benchmarking protocol that fixes the model, data, and sampling parameters and that reports relative-error metrics alongside absolute ones. Code and analysis tools are available at this https URL.
13. 【2610.40533】Homebrewed SLMs: Autoregressive and Masked Diffusion Language Models Built at Home for Portuguese
链接：https://arxiv.org/abs/2610.40533
作者：Pedro Carvalho Brom
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：Building a language, data centre, longer requires, requires a data, Building
备注： 34 pages, 14 figures, 13 tables. Code and notebooks: [this https URL](https://github.com/pcbrom/homebrewed-slms)
点击查看摘要
Abstract:Building a language model of one's own no longer requires a data centre. This tutorial trains one from scratch, tokenizer included, on one consumer GPU, and shows how to tell whether it is good for the tasks of a specialist domain. It proposes no new method. It releases a tested Python package and two notebooks in Brazilian Portuguese with autoregressive and masked diffusion generation on one network, a preregistered comparison protocol, legal tasks built from text alone with a Kneser-Ney floor and public yardsticks, reference numbers for models of 11 to 12M parameters, and a catalogue of defects that only measurement revealed. On Brazilian prose and federal laws, at minutes per run, the autoregressive mode peaks at 64 epochs over the same text and then memorises it, while masked diffusion improves up to 256. At their best points on test, the diffusion bound exceeds the autoregressive likelihood by 0.06 bits per byte, within the preregistered minimum effect. On held-out laws and on acts published after training, the model ties a 6-gram floor in likelihood on the same windows, and it loses to the floor at continuing those acts. It beats the floor at detecting tampered text, and fills cross-references more often than the floor and public models of 160M and 630M parameters, only in decrees, a genre absent from training, where those scorers stay near or below chance, and in filling the blank only when each group weighs the same, since per item it ties the 630M model. On laws, the public models, trained with about 5x10^2 to 10^5 times its compute, lead in tampered text and continuation. The blind human reading is preregistered and not part of this version.
14. 【2610.40518】IndicIPR-QA: A Benchmark for Extractive Question Answering on Indian IPR Laws
链接：https://arxiv.org/abs/2610.40518
作者：M. Mohaiminul Islam,Muhammad Abulaish
类目：Computation and Language (cs.CL)
关键词：Natural Language Processing, Language Processing, Natural Language, Legal Question Answering, task in Natural
备注：
点击查看摘要
Abstract:Legal Question Answering (QA) over statutory texts is a challenging task in Natural Language Processing (NLP), especially in low-resource settings with complex legal systems. It requires answers that are precise and grounded in the law. However, to the best of current knowledge, no existing extractive QA benchmark provides exact answer spans from statutory texts across all seven Indian Intellectual Property Rights (IPR) Acts. This paper presents IndicIPR-QA1, an extractive QA benchmark for Indian IPR laws. The proposed dataset contains 4,462 QA pairs from 1,116 statutory contexts across seven Acts. Questions and answers are generated using GPT-4o and validated through nine automatic quality checks, single and multi-model LLM evaluations, and a 200-pair human audit. Section-level partitioning prevents source-context overlap across the training, validation, and test sets. Further analysis examines answer span length, lexical overlap, and multi-clause structure across the corpus and individual Acts. BM25 achieves a token-level F1 score of 20.21, while fine-tuned InLegalBERT reaches 85.02 under a unified sliding-window inference protocol. The large performance gap between BM25 (20.21 F1) and InLegalBERT (85.02 F1) shows that simple word matching is not enough to answer these legal questions. Performance also varies across the seven Acts, indicating substantial differences in answer-extraction performance.
15. 【2610.40516】How I Aligned Your Model: Scaling Up Multilingual Safety Training with Templates
链接：https://arxiv.org/abs/2610.40516
作者：Max Neuwinger,Hanna Yukhymenko,Imanol Schlag
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：Large Language Models, damaging low-resource language, Large Language, limited multilingual representation, heavily focused
备注：
点击查看摘要
Abstract:Safety alignment for Large Language Models is heavily focused on English with limited multilingual representation, particularly damaging low-resource language performance. Related open training datasets are often limited in size or do not follow region-localized harmful topics to broaden the scope of alignment beyond cultures. We propose a scalable template-based augmentation pipeline to produce massive multilingual refusal and benign corpora, extracting safe templates from harmful prompts, translating them and then filling in the slots with culturally authentic entities, both harmful and benign. As a result, we release MSafeTemp, a validated corpus spanning 60 languages useful for multilingual safety alignment and evaluation. Using the dataset, we show that English-only evaluation misjudges multilingual safety: for several models, attacks that fail in English succeed once translated, while culturally localized variants expose further failures that translation alone misses. We also confirm that augmenting safety training data through templates allows for higher diversity and quality of alignment, resulting in improved safety performance beyond English. MSafeTemp and the respective pipeline are aimed at facilitating research and development of safer and more responsible models for less-represented cultures.
16. 【2610.40511】A Low-Rank Space Captures Linear Alignment Between LLM and Human Neural Activity During Naturalistic Language Listening
链接：https://arxiv.org/abs/2610.40511
作者：Zach Paris,Emily Finn
类目：Computation and Language (cs.CL); Neurons and Cognition (q-bio.NC)
关键词：Large language models, human language processing, Large language, remains unclear, LLM representations
备注： 23 pages, 12 figures. Code: [this https URL](https://github.com/thefinnlab/low-rank-alignment)
点击查看摘要
Abstract:Large language models (LLMs) are increasingly used as computational models of human language processing, yet it remains unclear which components of their representations drive alignment with neural activity. Here, we use ridge reduced-rank regression to identify the brain-predictive subspace of LLM representations: the representational directions that support linear prediction of neural responses. Across three naturalistic language-listening datasets spanning intracranial electrophysiology and fMRI, and six model conditions spanning autoregressive, bidirectional masked, and randomly initialized language models, we find that full linear encoding performance can be recovered using only 1-3% of the original embedding dimensionality. These low-dimensional subspaces generalize across held-out participants and, to a lesser extent, across datasets, suggesting that neural predictivity is concentrated in a compact and partially shared representational space. We next compare these brain-predictive directions with the principal axes of variation in the embeddings. Principal component analysis requires substantially more dimensions to recover neural prediction, demonstrating that the directions most useful for predicting neural activity are distinct from those capturing the most overall variance within LLM representations themselves. Finally, characterizing the learned subspaces allows us to surface interpretable features that are overrepresented in neural activity above and beyond the LLM representations. Together, these results show that LLM-brain alignment is supported by a small, structured subset of model representations and provides a framework for better understanding exactly why these representations align. Code can be found at: this https URL
17. 【2610.40492】Agentic University: Collaborative Agents for Personalized Learning
链接：https://arxiv.org/abs/2610.40492
作者：Jingru Fan,Tianle Zhou,Yunjie Zhou,Zihao He,Yixuan Zhou,Kezhao Wu,Xinyu Zheng,Xiyuan Wang,Yuheng Wang,Zhi Li,Songhua Liu,Weinan E,Chen Qian
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Large language models, Large language, support personalized tutoring, language models support, models support personalized
备注：
点击查看摘要
Abstract:Large language models support personalized tutoring, content generation, and adaptive learning paths. However, capability development requires these resources and interactions to form a coherent learning process that evolves with the learner. A central challenge is to align goals, activities, and assessment while using learning evidence to decide what should follow. We introduce Human-Participatory Multi-Agent Organization and instantiate it in Agentic University (AU). In AU, learners participate as organization members and work with specialized agents to maintain a revisable Capability Program. The program connects personal goals, learning activities across modalities, and layered assessment, while learner feedback and performance guide subsequent support. Our evaluation shows advantages in program organization, assessment design, and revision quality. Users achieved larger mean assessment gains (12.19 versus 5.61 points on a 50-point scale) and reported stronger cognitive support. This work contributes an organizational approach to adaptive learning that makes learner participation a continuing basis for multi-agent coordination. The Agentic University project is available at \href{this https URL}.
18. 【2610.40488】rTaste: Do Research Agents Have Taste? Evaluating Early Stage Research Judgment
链接：https://arxiv.org/abs/2610.40488
作者：Art Kanke
类目：Computation and Language (cs.CL)
关键词：worth running, recent NLP papers, NLP papers, MVEs, recent NLP
备注： 27 pages, 1 figure, 9 tables. Code and data: [this https URL](https://github.com/ArtKanke/rTaste)
点击查看摘要
Abstract:Do research agents have taste, the judgment of which experiments are worth running before their outcomes are known? We give six frontier models seven recent NLP papers. Each model, working as a research agent alone or as a team, proposes minimum viable experiments (MVEs), forecasts whether each will pass a threshold it sets and ranks them. A separate executor then attempts every MVE, including those ranked last, and three model auditors judge the results. Of 415 MVEs, 194 are judged sound, valuable and new, but fewer than half of these (73) are also judged beyond the seed paper and non obvious. The agents' first choices match the auditors' most valued experiment more often than a random pick, but are not shown to improve on choosing the experiment written first. Their forecasts beat a base rate but are not measurably more accurate than those of a fresh instance of the same model given only the paper and registrations. Teams cost more without a clear overall improvement in proposal quality. The agents produce many worthwhile experiments, but the evidence that their rankings improve the choice of which to run remains limited. The evaluation harness and a dataset of all 415 MVEs, with their forecasts and audited outcomes, are available at this https URL.
19. 【2610.40483】Preserving Utility in Continual LLM Unlearning via Orthogonal Projections
链接：https://arxiv.org/abs/2610.40483
作者：Juan Belieni,Ana Carolina Erthal,Eliezer de Souza da Silva,Diego Mesquita
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：training data, copyrighted or private, private text, concerns the task, task of removing
备注： 14 pages, 4 figures, 3 tables
点击查看摘要
Abstract:Unlearning concerns the task of removing the influence of a sensitive portion of the training data, such as protected, copyrighted or private text, from a model without retraining it from scratch. While unlearning requests happen continuously during a model's lifetime, the vast majority of LLM unlearning methods are ill-suited for iterated removal, yielding steep decay in forgetting and utility upon sequential unlearning. This decay stems from interference between successive updates, as each new deletion can undo earlier ones or push the model further along directions already degraded by them. To address this, we introduce Continual Orthogonal Projection (COP), a gradient-projection scheme that restricts each update to the orthogonal complement of activation subspaces cached from prior forget sets. This approach provably guarantees first-order non-interference with earlier deletions. COP stores only a low-rank basis per layer, requiring neither past gradients nor previous forget sets, and plugs into unlearning methods that update the model through gradient steps on a differentiable loss without modifying its objective. Experiments on standard LLM unlearning benchmarks (TOFU and RWKU) show that COP preserves utility across sequential requests without weakening forgetting, preventing the collapse of standard objectives and forgetting more effectively than continual-specific baselines.
20. 【2610.40482】FIRME: Registry-Anchored Memory Graphs for Evolving, Heterogeneous Investor Knowledge
链接：https://arxiv.org/abs/2610.40482
作者：Yunbai Zhang,Yunyi Gao,Han-Tai Shiao
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：reuse investor knowledge, Financial AI workflows, Correct Match Rate, workflows that reuse, reuse investor
备注： 20 pages. Accepted at the 3rd Workshop on LLMs and Generative AI for Finance (AI4F) at ACM ICAIF 2026 (non-archival)
点击查看摘要
Abstract:Financial AI workflows that reuse investor knowledge need persistent memory that integrates observations across sources and periods without losing identity or provenance. In a knowledge graph, recurring observations may use different surface forms, fragmenting relationship histories and preventing evidence from accumulating. We introduce Fusion and Identity Registry for Memory Evolution (FIRME), a registry-anchored memory graph that reconciles incoming observations to persistent canonical relationships and retains their source, time, and context. Within periods, rule-based fusion links investor-call discussions to portfolio cash-flow signals, directly or through a fixed product-topic ontology. Across periods, canonical identities and persistent aliases direct recurring observations to the same relationship history; deterministic updates summarize reinforcement, inactivity, and reactivation. We evaluate FIRME on proprietary investor-call and cash-flow records and on SP 500 earnings-call transcripts. Compared with a historical registry that retains canonical identities and semantic fallback but omits alias memory, FIRME increases final-period Correct Match Rate by 7.8-12.1 percentage points across four dataset-backend settings under a fixed-judge and fallback-review protocol. A blinded human audit of 200 incoming observations per setting finds Correct Match Rate gains of 9.0-16.5 points and reattachment-recall gains of 22.5-29.0 points, with higher observed merge precision in all four samples. A separate audit finds valid evidence paths for 293 of 300 outputs from full within-period fusion in the investor-call setting. Extraction-faithfulness and round-trip diagnostics assess source grounding and preservation of extractor-derived graph content.
21. 【2610.40476】raining language models in simulation for psychiatric intake
链接：https://arxiv.org/abs/2610.40476
作者：Samuel Schmidgall,Tim Strother,Alex Bijamov,Min Woo Sun,Anil Palepu,Akshay Goel,Ulysse Marteau-Ferey,Théo Moutakanni,Yubin Kim,Joseph Breda,Christina Chen,Justin Chen,Paul Jhun,Andrew Marshall,Sarah Montgomery-Taylor,Eun Kyoung Hong,Shashank Ravi,Kameron C. Black,David J. H. Wu,Austin Schoeffler,John Lyon Havlik,Elizabeth C. Stade,Jared Moore,Johannes C. Eichstaedt,Logan Pierce,Jeff Jopling,Johannes Thrul,Thomas H. McCoy,Neguine Rezaii,Mark S. Chen,Ellen E. Fitzsimmons-Craft,Shirley Wang,Desmond C. Ong,Katharina Schultebraucks,Torrey Creed,Bhavna Daryani,Aleksandra Faust,S. Sara Mahdavi,Caroline Figueroa,Eric Lin,Kate H. Bentley,Ernest Lim,Zhihan Jiang,Xuhai "Orson" Xu,Daniel McDuff,John Torous,Yun Liu,Po-Hsuan Cameron Chen,Mike Schaekermann,Joelle Barral,Alvin Rajkomar,Tao Tu,Lin Yang,Isaac Galatzer-Levy,David Racz,Valentin Liévin
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：primary diagnostic instrument, intake interview remains, diagnostic instrument, disentangling complex, interview remains
备注： 76 pages, 8 figures, 15 tables (27 pages main text, 7 main figures, 2 main tables)
点击查看摘要
Abstract:In psychiatric practice, the clinical intake interview remains a primary diagnostic instrument, requiring the clinician to establish rapport while disentangling complex, overlapping presentations. Here, we show that multi-turn reinforcement learning (RL) in simulation environments improves language model performance on structured text-based psychiatric intake, spanning history-taking, differential diagnosis, and initial management planning. In a randomized, double-blind crossover study ($N = 60$ matched scenario pairs, 120 encounters) comparing the RL-trained agent against its identically prompted base model, 32 board-certified psychiatrists role-played standardized patients while a separate panel of 15 independent, board-certified psychiatrists scored the de-identified transcripts. The trained agent outperformed the base model in 76.7% of matched scenario pairs ($p 0.0001$), with gains on an adapted clinical-examination rubric ($d_z = 0.77$, $p 0.0001$), patient-centered communication (82.0% vs. 65.2%, $p 0.001$), and differential diagnosis appropriateness (68.3% vs. 33.3%, $p 0.001$). Across 127 psychiatrist-validated simulated intake scenarios, the trained agent nearly doubled its base model's clinical rubric score (0.706 vs. 0.379; composite reward 0.564 vs. 0.294, $p 0.0001$). To systematically probe conversational safety, we established two Delphi-consensus taxonomies, of clinical safety risks and relational competencies, from which 256 challenging adversarial presentations were derived; we find that training in simulation environments more than halved under-triage of acute safety risks (12.9% vs. 30.5%, $p 0.0001$). Simulation environments thus provide a controlled testbed for training and stress-testing psychiatric conversational agents, with gains that transferred to psychiatrist-enacted encounters, though residual gaps require safeguards and clinician oversight.
22. 【2610.40454】Glance: The Generator Is Its Own Reranker
链接：https://arxiv.org/abs/2610.40454
作者：Adilet Metinov,Chingiz Arziev
类目：Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：duplicating model residency, serves separate models, Retrieval-augmented generation commonly, commonly serves separate, generation commonly serves
备注： 12 pages, 2 figures, 6 tables. Adapters and code: [this https URL](https://huggingface.co/airun-ai/GlanceRAG-MiMo-9B-Reranker)
点击查看摘要
Abstract:Retrieval-augmented generation commonly serves separate models for reranking and answering, duplicating model residency and passage processing. We introduce Glance, which equips the generator with a lightweight decision adapter and pointer head to score candidate passages from a shared prefill. Short relevance probes branch from the cached context, producing one score per passage without separately prefilling each query-passage pair. A single resident 9B backbone supports generation, general decisions and reranking, with approximately 0.1 GiB of additional memory per decision skill. After 39 minutes of MS MARCO fine-tuning, Glance achieves nDCG@10 of 0.741 on SciFact and 0.345 on NFCorpus, compared with 0.721 and 0.326 for bge-reranker-v2-m3. On 300 HotpotQA questions, its selected context improves answer F1 by 0.045 over cross-encoder-selected context under the same base generator (paired bootstrap 95% CI [0.014, 0.076]). In a scaling study, reranking quality keeps rising up to 200 candidates (43k tokens) scored from one shared prefill, where it exceeds the cross-encoder by 0.058 nDCG@10, and a probe-only activated adapter lets generation continue from the original passage cache. These initial results demonstrate generator-based reranking with one resident backbone; evaluation currently covers two retrieval datasets, one QA dataset and one backbone family.
23. 【2610.40444】Prefix Executability: Evaluating Tool-Using Agents Beyond Final Success
链接：https://arxiv.org/abs/2610.40444
作者：Amir H. Rezaeian,Weiyi Sun,Yassine Benajiba,Dan Roth
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：intermediate tool-call trajectory, Trajectory Executability Rate, Prefix Executability Curves, prefix executability, benchmarks often score
备注： 21 pages, 2 figures. Accepted to NeurIPS 2026
点击查看摘要
Abstract:Tool-use benchmarks often score agents by final task success, but terminal outcomes can obscure whether the intermediate tool-call trajectory was executable. We propose prefix executability as a trajectory-level evaluation protocol for tool-using agents. The protocol measures whether every prefix of a predicted tool-call sequence remains executable under tool schemas, state preconditions, and data dependencies, and summarizes reliability using Prefix Executability Curves (PEC), First-Error Depth (FED), Trajectory Executability Rate (TER), horizon-normalized AUC-PEC, and Recovery Rate (RR). We instantiate the protocol on stratified samples from BFCL-v3 multi-turn and ToolSandbox, and on the complete Airline/Retail base splits of $\tau^3$-bench. We observe that outcome-only and prefix-level rankings diverge: agents whose execution is conditioned on an unverified plan often achieve higher point-estimate final success than non-plan-conditioned baselines while substantially lowering TER and AUC-PEC. These divergences are precisely the failure modes that prefix executability is designed to expose: they reveal whether apparent success rests on an executable trajectory, where invalidity first enters, and whether success follows recovery from an earlier failure. As a validation intervention, explicit pre-execution plan checking can improve executable-prefix survival while maintaining competitive terminal success. These results show that final success alone is insufficient for evaluating tool-use agents and that prefix-level executability provides an actionable reliability signal for comparing agents, diagnosing failures, and assessing validation mechanisms.
24. 【2610.40394】Scaling Mathematical Proof Supervision from Structured Textbooks
链接：https://arxiv.org/abs/2610.40394
作者：Youqing Fang,Zhouqi Hua,Yinhao Tang,Yuzhe Gu,Lingkai Kong,Bin Liu,Weiming Zhang,Kai Chen,Yanan Sun,Wenwei Zhang
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：isolated problem-solution pairs, problem-solution pairs, discarding the curricular, Textbook-Oriented Mathematical Evidence, typically flattened
备注： 35 pages, 18 figures, 14 tables
点击查看摘要
Abstract:Mathematical supervision for language models is typically flattened into isolated problem-solution pairs, discarding the curricular organization and evidential structure of the original sources. We introduce TOME-Math (Textbook-Oriented Mathematical Evidence), a large-scale corpus that converts 2,118 textbooks spanning the core areas of university-level mathematics subjects into approximately 380K structured proof-supervision records while preserving chapter order, proposition-proof links, dependency references, and source provenance. Leveraging these structural signals, we explore curriculum learning across multiple mathematical subjects and characterize scaling patterns with supervision volume and source breadth. At full scale, fine-tuning on TOME-Math raises the unweighted mean score across our held-out evaluations and established public benchmarks by 16.7 percentage points over the base model, while exceeding all three public instruction-data baselines evaluated in aggregate. Our findings highlight structured textbooks as both a scalable source of mathematical supervision and an empirical framework for studying the impact of data organization on mathematical reasoning. We will release the corpus to facilitate further research on structured mathematical supervision.
25. 【2610.40350】Don't Lose the Thread: Empowering Long-Horizon LLM Agents with Cognitive Resource Self-Allocation
链接：https://arxiv.org/abs/2610.40350
作者：Zheng Zhu,Xiaozhuan Liang,Xi Chen,Bei Yu,Jiaya Jia
类目：Computation and Language (cs.CL)
关键词：solving complex reasoning, demonstrated remarkable progress, large language models, powered by large, large language
备注： Accepted at COLM 2026. Code is available at [this https URL](https://github.com/JIA-Lab-research/CORAL)
点击查看摘要
Abstract:Agents powered by large language models (LLMs) have demonstrated remarkable progress in solving complex reasoning tasks. However, LLM agents often falter on long-horizon tasks due to cognitive overload, as their working memory becomes cluttered with expanding and irrelevant information, which dilutes their attention and hinders effective planning and reasoning. To mitigate this challenge, we introduce Cognitive Resource Self-Allocation (CORAL), a novel reasoning paradigm that empowers agents to proactively optimize their context. Implemented as an agent-callable working memory management toolset, CORAL allows an agent to checkpoint its task progress and verified facts, and adaptively initiate a new problem-solving episode by resetting cluttered context while preserving only the checkpointed information, enabling it to resume reasoning from a clean state. We further enhance the agent's checkpointing capabilities using a Multi-episode Agentic Reinforced Policy Optimization algorithm. On the GAIA benchmark, CORAL with an 8B model achieves 42.8% average accuracy, with particularly strong gains on the most challenging long-horizon tasks (Level 2 and Level 3). Crucially, these gains come at a substantially lower inference cost: CORAL consumes 6-7$\times$ fewer input tokens than a ReAct agent. It also generalizes to multi-hop Knowledge QA benchmarks and to the considerably harder BrowseComp, where it improves a strong proprietary agent by 8.0 points.
26. 【2610.40328】Factorized Representations of Concurrent Temporal Variables in Pretrained Language Models
链接：https://arxiv.org/abs/2610.40328
作者：Xinhe Zhang,Arnau Marin-Llobet
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：Language models, maintain multiple contextual, maintain multiple, multiple contextual variables, Language
备注：
点击查看摘要
Abstract:Language models often need to maintain multiple contextual variables that evolve independently within a document. Prior work has identified structured representations of individual temporal or contextual quantities, but their joint organization remains less understood. We study this question using controlled documents with independently phased language-switch and topic-change timers. Across ten pretrained checkpoints, both timers are linearly decodable from the same hidden states. Removing a timer-associated centroid plane substantially reduces access to its associated variable while having little effect on the concurrent timer, although the planes are nonorthogonal and contain cross-timer information. Separate timer contributions also explain most reproducible variation among joint-state centroids. These signatures persist from 3 to 51 timer values and across Qwen3 checkpoints from 0.6B to 14B. Schedule position affects continuation probabilities, but the tested centroid displacement produces no consistent timer-specific output effect. Together, these results support approximate factorization based on selective accessibility and approximately additive joint-state geometry in overlapping subspaces.
27. 【2610.40318】Decision and System One models in rheumatology: accuracy and calibration on specialty-training examination questions and diagnostic case vignettes
链接：https://arxiv.org/abs/2610.40318
作者：Alfredo Madrid-García,Diego Benavent,Fabian Lechner,Phillip Kremer,Johannes Knitza,Beatriz Merino-Barbancho
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：OpenAI Decisions API, Sol, rheumatology examination questions, Jev, System One models
备注： 28 pages, 2 figures, 2 tables; includes supplementary material (8 tables, 1 figure)
点击查看摘要
Abstract:Objectives: To evaluate the accuracy and calibration of Decision and System One models on rheumatology examination questions and diagnostic vignettes. Methods: Two benchmarks were analysed: 172 rheumatology questions from Spanish specialty-training examinations (expanded RheumaMIR dataset) in Spanish and English, and 70 English diagnostic vignettes from RheumBench with 5-18 candidate diagnoses, 52 of which included the reference diagnosis. Four System One models (Jev 1.13, Clef, Microsoft-Decision-1, Laya) and GPT-6 Luna (OpenAI Decisions API) were compared with GPT-6 Sol and two local generative models. The primary outcome was accuracy; secondary outcomes included calibration, discrimination, accuracy of answers with probability =0.90 and language differences. Results: In RheumaMIR, Jev's accuracy was 93.0% in Spanish and 91.9% in English, below GPT-6 Sol (99.4%; Holm-adjusted P=0.005); GPT-6 Luna reached 97.7%, Clef 91.9-93.0% and Microsoft-Decision-1 86.0-87.8%, without detectable language differences. Jev's answers with probability =0.90 (76-77% of answers) were 98.5% correct. In RheumBench, accuracy was lower: GPT-6 Sol 76.9%, GPT-6 Luna 63.5%, Clef 59.6%, Jev 50.0% and Microsoft-Decision-1 40.4% (each System One model below GPT-6 Sol, Holm-adjusted P=0.045; chance 9.6%). Jev became overconfident (mean probability 0.659; over/underconfidence index +0.159), and its answers with probability =0.90 (25% of cases) were 84.6% correct, against 95.5% for GPT-6 Luna (42%). Laya performed near chance in both benchmarks. Conclusion: The best-performing System One models were highly accurate on examination questions but less accurate on diagnostic vignettes, remaining below GPT-6 Sol. Jev's high-confidence predictions were less reliable on vignettes, highlighting the need for task-specific evaluation and calibration.
28. 【2610.40290】Leveraging Recourse and Size Bounds for OnlineKnapsack with Removable Items
链接：https://arxiv.org/abs/2610.40290
作者：Dimitris Fotakis,Laurent Gourvès,Aris Pagourtzis,Panagiotis Patsilinakos
类目：Data Structures and Algorithms (cs.DS); Computation and Language (cs.CL)
关键词：unweighted knapsack problem, items arrive sequentially, aiming to maximize, competitive ratio, arrive sequentially
备注：
点击查看摘要
Abstract:In the online unweighted knapsack problem, items arrive sequentially and must be accepted or rejected immediately for a bounded-capacity knapsack, aiming to maximize the total size of accepted items. When decisions are irrevocable, no algorithm achieves a non-trivial competitive ratio. An essential relaxation that circumvents this impossibility is to allow previously accepted items to be removed from the knapsack (a.k.a. removal). In this work, we investigate possible further improvements on the competitive ratio with removal by bringing together two other previously proposed model extensions: knowledge of an upper bound on the item sizes (a.k.a. bound awareness) and ability to reinsert previously rejected items(this http URL). Our contributions are twofold. First, we establish close upper and lower bounds on the competitive ratio under bound awareness, removal and recourse. We show that significant improvements on the best known competitive ratio are possible when the upper bound on the item sizes is not too restrictive and only few uses of recourse are available. Second, we introduce the notion of adaptive online algorithms which do not require to know any upper bound on the item sizes but still achieve performance guarantees comparable to the ones obtained under the bound awareness model.
29. 【2610.40283】Latent Path Exploration via Potential-Guided Subtrajectory Balance
链接：https://arxiv.org/abs/2610.40283
作者：Xiaoling Zeng,Qi Zhang
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：explicit reasoning traces, perform intermediate computation, reasoning enables language, generating explicit reasoning, enables language models
备注：
点击查看摘要
Abstract:Continuous latent reasoning enables language models to perform intermediate computation without generating explicit reasoning traces. Advances in stochastic latent reasoning allow multiple paths to be sampled in continuous space, but stochasticity alone does not ensure effective exploration. A key challenge is guiding stochastic transitions between intermediate latent states toward useful reasoning paths. We introduce SubLAT, which couples a relative answer potential with Subtrajectory Balance to learn stochastic latent exploration. The relative potential evaluates the answer support of sampled states against replacement states, with a correctness signal at the terminal state. Subtrajectory Balance relates potential differences to forward and backward transition densities across subtrajectories of different lengths. The model is jointly trained with answer supervision, CoT supervision, and representation distillation. Experiments on arithmetic and deductive reasoning show increasing solution coverage with larger sampling budgets across in-domain and out-of-domain benchmarks, as well as improvements on several structural OOD benchmarks. Trajectory analyses reveal progressive divergence among sampled latent paths and higher mean relative potentials for correct trajectories across arithmetic benchmarks.
30. 【2610.40278】Neighborhood Elites: Quality-Diversity Search for Uncertainty Quantification in Question Answering
链接：https://arxiv.org/abs/2610.40278
作者：Mesut Ceylan,Arda Baris Basaran,Elgar Fleisch,Filipe Barata
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Large language models, Large language, wrong answers indistinguishable, increasingly capable, indistinguishable from correct
备注：
点击查看摘要
Abstract:Large language models (LLMs) are increasingly capable, yet they produce wrong answers indistinguishable from correct ones. In question answering (QA), such an error propagates unnoticed into downstream tasks. Perturbation-based uncertainty quantification (UQ) addresses this by probing a model over a set of input rewrites and reading uncertainty from how much its predictions vary. Yet this set is usually built generically, by free paraphrasing, token-level perturbation, or adversarial search. Such rewrites leave two things uncontrolled that matter in QA: what kind of change is made to the context, and how explicitly the context supports the answer. When neither is controlled, the composition of the set is determined by the generator rather than by design, and the estimate inherits whatever that generator happens to produce. Quality-diversity (QD) search is built to cover a predefined exploration space, so we recast input perturbation as a QD problem and propose \emph{neighborhood elites (NE)}, a neighborhood constructor that spreads answer-preserving rewrites across two axes, the kind of edit and the explicitness of support. We compare NE against three other neighborhood constructors, holding the editor, QA model, uncertainty measure and search objective fixed, and against three scores read from the original context. Across five extractive-QA datasets and five instruction-tuned LLMs, NE improves failure detection against every other constructor in at least 24 of the 25 dataset-model pairs, reaching 0.726 AUROC and 0.211 AUPRC lift over the base rate. Additionally, NE detects failure better than semantic entropy computed over decoder samples, by 0.074 AUROC and 0.062 AUPRC lift, and this advantage widens as the neighborhood grows.
31. 【2610.40256】When Requirements Interact: Scaling Long-Horizon Coding via Interdependent Specifications
链接：https://arxiv.org/abs/2610.40256
作者：Yishuo Cai,Minzheng Wang,Liang Zhao,Linghao Zhang,Lei Li,Shicheng Li,Xu Sun,Fuli Luo
类目：oftware Engineering (cs.SE); Computation and Language (cs.CL)
关键词：long-horizon software development, real-world long-horizon software, breaking existing dependencies, satisfy multi-dimensional task, software development
备注：
点击查看摘要
Abstract:To succeed in real-world long-horizon software development, coding agents must satisfy multi-dimensional task specifications, such as interface compatibility and error handling, and do so without breaking existing dependencies. However, current works lack specification in both quantity and diversity, often emphasizing primary goals while ignoring complex constraints. Synthesizing tasks to bridge this gap presents two challenges: naively adding specifications risks reducing tasks to trivial checklists rather than compelling joint reasoning, and verifying these subtle interactions risks overlooking hidden flaws or rejecting valid alternatives. To address these challenges, we introduce CodeSpecWeave, a framework for synthesizing tasks with rich, interdependent specifications. Grounded in real developer seeds, a synthesis agent iteratively develops both coupled task requirements and executable tests, while a review agent audits solution trajectories from multiple models to guarantee test reliability. Using this pipeline, we construct 4,670 training tasks across 2,865 repositories. Applying Reinforcement learning on these tasks significantly improves performance over the base model on DeepSWE-v1.1, largely outperforming existing baselines. Furthermore, we curate CodeSpecWeave-Bench, a rigorous benchmark comprising 100 tasks and 1,044 specifications, which clearly separates frontier models and reveals significant headroom in handling interdependent engineering constraints.
32. 【2610.40241】Judged in Company: In-Context Batching Makes LLM Verdicts Depend on Other Items
链接：https://arxiv.org/abs/2610.40241
作者：Peng Xie,Amr Alanwar
类目：Computation and Language (cs.CL)
关键词：grade many independent, items, independent items, language model, decisions
备注：
点击查看摘要
Abstract:To save cost, practitioners often ask a language model to grade many independent items in one prompt. This assumes that verdicts are separable: that the verdict on an item does not depend on the other items in the prompt or on its position among them. Across three judgement tasks (harm in social-media comments, short-answer grading and review star ratings) and five open-weight models with 4-9B parameters, assigning the same 2,000 items to different random batches of 25 changes 11-22% of decisions on harm and grading, 6.5-13.3 points more than renaming the item keys or rewording the instruction, while AUROC against human labels differs from single-item grading only by -0.033 to +0.010; for four models with 14-32B parameters, 6-15% of decisions change, 4.2-9.1 points more. We show that flip rates decompose exactly into components of order, slot and membership, and that under a threshold model they grow with the square root of the variance that the context contributes, so that effects that are small for scores are large for decisions. Most of the latent dissimilarity between partitions comes from the order of the neighbours and the slot. The main effect of which items share the prompt, the part that voting over permutations of a batch cannot remove, accounts for 0.11-0.22 of it, and majority votes over nine permutations still leave 52-60% of the flips. Batched verdicts also differ systematically from single-item verdicts: even the majority of fifteen random batches disagrees with them on 8.4-17.8% of decisions. Editing a single generation shows that both the model's own earlier verdicts and the neighbours' texts carry the effect. On the easy task, star ratings, only 2.9-3.9% of decisions change. We recommend grading ambiguous items one per prompt, recovering most of the cost with a cached prefix, and reporting agreement with single-item grading whenever items are batched.
33. 【2610.40238】ScienceJudge: Pattern Perturbation Augmentation for Sentence-Level Scientific Claim Verification
链接：https://arxiv.org/abs/2610.40238
作者：Chuhai Cai,Xin Chen,Min Yang,Feng Jiang
类目：Computation and Language (cs.CL)
关键词：Sentence-level scientific claim, multimodal evidence bundle, Sentence-level scientific, AI-generated scientific paragraph, scientific claim verification
备注： 13 pages, 1 figure; NLPCC 2026 Shared Task 10 (Track 1)
点击查看摘要
Abstract:Sentence-level scientific claim verification determines whether each sentence in an AI-generated scientific paragraph is supported by a given multimodal evidence bundle. Fine-grained reporting errors often preserve fluent and plausible statements while introducing subtle evidence-claim deviations, making them difficult to detect under sparse and imbalanced supervision. We propose Pattern Perturbation Augmentation (PPA), which models reporting errors as label-specific faithfulness breaking transformations and induces reusable error patterns from labeled examples. PPA applies these patterns to supported claims to construct controlled hard negatives and expands their coverage through pattern-level, operation-level, and cross-domain augmentation. Based on these augmented views, we develop ScienceJudge, which integrates PPA-trained vision-language predictors and instruction-guided LLM predictors through class-weighted ensemble inference. On the local development set, ScienceJudge improves the average score from 0.8124 to 0.8692, achieving a Macro-F1 of 0.8466 and a Paragraph Exact Match score of 0.8919. Our submitted system ranks first on the official NLPCC 2026 Task 10 Track 1 leaderboard with a final score of 0.6890.
34. 【2610.40233】Mobilizing the Archive: Bitext Extraction at Scale from Archival Documents for Extremely Low-resource Languages
链接：https://arxiv.org/abs/2610.40233
作者：Chutong Meng,Alec J. Bateman,Antonios Anastasopoulos
类目：Computation and Language (cs.CL)
关键词：mining parallel sentences, web presence, web pages, pages has proven, proven effective
备注：
点击查看摘要
Abstract:While mining parallel sentences from web pages has proven effective for low-resource machine translation, the majority of the world's languages have little to no web presence. What they do have, though, are resources that are the result of centuries of efforts by linguists, such as grammars, dictionaries, and other documentation materials. We propose extracting translation examples from such linguistic resources, which offer parallel data for languages under-represented online and could potentially scale to thousands more languages. We develop a pipeline to achieve this, leveraging an Optical Character Recognition (OCR) model as well as a vision language model (VLM), to ensure identified content to be aligned with visual layout information. We demonstrate the utility of the extracted bitexts for bootstrapping translation systems across 20 languages, both by training machine translation models and by applying in-context learning for large language models. For in-context learning, we observe improvements of +15.78 and +6.94 chrF2++ over zero-shot translation for into low- and high-resource languages, respectively.
35. 【2610.40227】EpiAgent: Epistemic Self-Play for LLM Agents
链接：https://arxiv.org/abs/2610.40227
作者：Xuanfa Jin,Xinyu Cui,Zhijian Ma,Xue Yan,Jian Zhao,Haifeng Zhang,Jun Wang
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA)
关键词：game theory explains, Epistemic game theory, agent strategic behavior, Epistemic, game theory
备注： 23 pages, 5 figures
点击查看摘要
Abstract:Epistemic game theory explains an agent's strategic behavior via the beliefs it holds about others and how these beliefs are updated as the interactions unfold. Modern LLM-based agents, however, often lack an explicit mechanism for learning and updating such beliefs to guide decisions under incomplete information. To bridge this gap, we propose EpiAgent, a framework that equips LLM-based agents with an explicit, learnable epistemic layer, which maintains beliefs over others' private information and combines these beliefs with the agent's available information into an epistemic state for action selection. Concretely, the agent's policy is decomposed into two coupled stages: Epistemic Update, which updates these beliefs from interaction histories, and Action Planning, which selects actions conditioned on the updated epistemic state. By formulating their dependency as a bi-level optimization problem, we introduce Epistemic Self-Play, an algorithm that alternately optimizes the two stages through self-play, utilizing an intrinsic prediction reward to evaluate how well inferred beliefs predict subsequent observations. Empirical results across both mixed and competitive scenarios show that EpiAgent consistently outperforms prompting-based and RL-based baselines. Further analyses reveal the benefits of dedicated epistemic training, improvements in belief inference accuracy, and the contribution of learned beliefs to action planning. These results demonstrate the value of an explicit, learnable epistemic layer for effective multi-agent interactions.
36. 【2610.40196】owards Full-Text Scholarly Argument Mining using Entity-Anchored Claim Extraction
链接：https://arxiv.org/abs/2610.40196
作者：Bhuvanesh Verma,Andy Lüking,Mounika Marreddy,Leon Lukas Hammerla,Alexander Mehler
类目：Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：Argument mining, scholarly articles remains, articles remains, remains an important, important yet challenging
备注：
点击查看摘要
Abstract:Argument mining in full-text scholarly articles remains an important yet challenging task. This difficulty stems not only from the need to model long-distance dependencies but also from the substantial effort required to annotate sentence-level argumentative components throughout lengthy documents. To address these challenges, we introduce a novel approach that leverages scientific entity mentions (e.g., tasks, datasets, and models) to identify and structure claims in scholarly literature. We validate our approach through a case study involving 50 full-text research papers on sentiment analysis. We first establish a reliable human gold standard (Krippendorff's $\alpha = 0.80$), identifying 113 claim-bearing sentences and extracting 199 structured claim records from 300 candidate sentences. To address the annotation bottleneck, we employ a majority-voting ensemble of large language models (LLMs) to generate a silver-standard dataset comprising 1,487 claims. Encoders fine-tuned exclusively on the silver-standard dataset outperform those trained on a small human-annotated gold-standard dataset and demonstrate strong zero-shot transfer to NLP papers outside the training corpus.
37. 【2610.40124】It Takes Two to Probe, Thousands to Retrieve: Putting Vision--Language Probes to the Retrieval Test
链接：https://arxiv.org/abs/2610.40124
作者：Ivana Beňová,Michal Gregor,Marián Šimko
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：text matching probes, text matching, tested for fine-grained, fine-grained understanding, correct item
备注： 12 pages, 2 figures, 3 tables
点击查看摘要
Abstract:Vision--language models (VLMs) are tested for fine-grained understanding with image--text matching probes, a model must prefer the correct image or caption over handcrafted negative. In retrieval, the correct item must beat thousands of alternatives. We measure how these two evaluation methods relate by using two probing benchmarks, \svo{} and SugarCrepe, for retrieval tasks. For 100 image and caption queries we judge the top-10 retrievals from full dataset of six VLMs, with human annotators and an LLM assessor. We than compare how well models rank pool of items on the same queries, while changing only where the negatives comes from: the benchmark, a random draw, or the judged-incorrect retrievals of other models. On \svo{} the accuracy of retrieving a positive over handcrafted negatives is significantly higher than for retrieved ones (the correct item is preferred in 85\% against 77\% of the time). Using pool of ten retrieved negatives leads to same accuracy as when one uses pool of 200 to 430 random negatives. On SugarCrepe the order is reversed: its handcrafted negatives are harder than the retrieved ones. Incomplete relevance labels explain most of the low retrieval scores on \svo{}, where the top-ranked item is correct for 54--90\% of the queries under human labels against 6--21\% under the benchmark's labels. However, we detect no association between passing the probe for a query and retrieving a correct item first. The LLM assessor o3 agrees with individual annotators ($\kappa$ 0.57--0.61) slightly less than they agree with each other (0.64--0.68).
38. 【2610.40091】Upcycling Language Models for Multiple Specialized Domains
链接：https://arxiv.org/abs/2610.40091
作者：Mingzi Cao,Xi Wang,Nikolaos Aletras
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：requires prioritizing relevant, Effective multi-domain adaptation, large language models, language models requires, models requires prioritizing
备注：
点击查看摘要
Abstract:Effective multi-domain adaptation of large language models requires prioritizing relevant domain knowledge while minimizing interference from irrelevant domains. Existing approaches commonly use Low-Rank Adaptation (LoRA) to train a separate adapter for each domain and combine these adapters within one model. However, they either weight adapters equally regardless of the input domain or dynamically weight adapters while still activating all of them including irrelevant ones. Additionally, while efficient, their low-rank structure limits performance. In contrast, upcycling is a model expansion method that adds experts and a router to a dense Transformer layer offering larger capacity than LoRA approaches. In this paper, we propose Domain-Specialized Upcycling (DS-Upcycling) for multi-domain adaptation. We introduce a new domain-aware supervision that allows the router to effectively learn to distinguish domains by selecting the appropriate experts. During fine-tuning on each new domain, we update only the corresponding expert and router while freezing all other parameters. Experiments show that our method achieves the highest average performance across the adapted domains compared to baselines while retaining general performance. With just 256 examples per domain for router supervision, DS-Upcycling achieves 90-98% routing accuracy after adaptation compared with 43-70% for routers trained without our approach.
39. 【2610.40055】One Vector or Many? Clustered Optimal-Transport Steering under Quantization
链接：https://arxiv.org/abs/2610.40055
作者：Illia Makarov
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Activation steering, hidden state, inference time, generates by editing, DiM
备注： 34 pages, 12 figures, 14 tables
点击查看摘要
Abstract:Activation steering changes what a language model generates by editing its hidden states at inference time. Its most common form adds one vector, the difference of class means (DiM), to every hidden state. CHaRS (concept heterogeneity-aware representation steering) uses optimal transport instead: an entropic plan pairs the clusters of the two classes, and each hidden state moves along a mixture of the cluster displacements, weighted toward its nearest clusters. We study this family of clustered optimal-transport maps for toxicity mitigation in Gemma-2-2B and Granite-3.1-2B, and ask what the plan adds to DiM and whether that survives INT8 and NF4 quantization. We prove an identity: whatever the plan, the displacements averaged with the cluster sizes as weights equal DiM, so a clustered map can differ from DiM only for tokens whose routing weights depart from the cluster sizes. In the CHaRS protocol they stay close, and CHaRS is statistically equivalent to Mean-AcT, which adds DiM at every layer, in all eight pre-specified equivalence tests. A variant fitted on individual tokens, which routes each token mainly to its nearest clusters, leaves 37-39% less toxicity than DiM on Gemma and 51-61% less on Granite at steering strength 3 (toxic-bert), where DiM already removes 76-85% of it. Two other toxicity classifiers confirm the gain on Granite, but on Gemma only one does. At equal strength the variant has somewhat higher perplexity, but at matched toxicity it is slightly more fluent than DiM. Yet a nearly constant map from the same clusters also leaves about 30% less toxicity, so the role of treating tokens differently remains open. Under INT8 and NF4 the variant still leaves less toxicity than DiM, and transferring it from full precision to the NF4 model instead of refitting it adds no significant toxicity after correction for multiple tests.
Comments:
34 pages, 12 figures, 14 tables
Subjects:
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Cite as:
arXiv:2610.40055 [cs.LG]
(or
arXiv:2610.40055v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.40055
Focus to learn more
arXiv-issued DOI via DataCite</p>
40. 【2610.40051】Rethinking Pre-pretraining
链接：https://arxiv.org/abs/2610.40051
作者：Alexander Stern,Tim Rocktäschel,Acyr Locatelli,Laura Ruis
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：outperforms pure natural-language, compute-matched experiments, PPT, report that allocating, outperforms pure
备注：
点击查看摘要
Abstract:We report that allocating compute to synthetic pre-pretraining (PPT) never outperforms pure natural-language training in our compute-matched experiments. This holds across four PPT curricula (k-Dyck, neural cellular automata (NCA), declarative reference corpora, and code) and three training regimes: data-constrained, data-unconstrained, and a high token-to-parameter-ratio regime (881M parameters trained on 50B tokens). When natural language is scarce, repeated epochs over a 100M-token corpus beat every synthetic curriculum on natural-language validation loss, with no consistent downstream accuracy advantage from PPT. We find that PPT gains reported in prior work are sensitive to the weight initialization scheme. In an NCA reproduction, we recover a 5.4% perplexity improvement from PPT under the original initialization scheme. An alternative scheme improves both methods, but the no-PPT baseline more, closing the gap between them. However, when PPT receives additional compute rather than a share of the natural-language budget, gains persist at small natural-language budgets even with improved initialization, but disappear by 1B tokens. Finally, models forget what they learn during PPT in subsequent natural-language training; interleaving the same data throughout training prevents forgetting, but does not significantly improve natural-language performance. Our results suggest limited benefits of synthetic pre-pretraining as a distinct training stage and motivate rethinking how such data are used in language model training.
41. 【2610.40031】AI Cosmos: An Agentic Infra for Situated Learning
链接：https://arxiv.org/abs/2610.40031
作者：AI Cosmos Team,Jinfeng Zhou,Yuxuan Chen,Fangming Chang,Mingyu Su,Yifan Zhu,Yuan Wang,Bo Zhang,Bosi Wen,Chenhao Yuan,Chenwen Liao,Dazhi Jiang,Deyao Hong,Erle Zhu,Genrui Zhang,Guchan Li,Hongwen Gan,Huichao Bi,Jiahao Yan,Jincheng Fan,Kehan Zheng,Qinren Luo,Qiran Liu,Ruoqi Cao,Shizhi Zhang,Sihan Chen,Siyan Deng,Xuehan Zhou,Xujun Li,Yang Dang,Yanzhuo Zhu,Yu Zhang,Zheyu Chen,Hongning Wang,Manli Li
类目：Computation and Language (cs.CL)
关键词：Large language models, improve task performance, necessarily reflect learners', reflect learners' ability, Large language
备注： [this https URL](https://github.com/THU-AICosmos)
点击查看摘要
Abstract:Large language models (LLMs) can improve task performance during learning, yet these gains do not necessarily reflect learners' ability to thereafter apply the knowledge and skills independently. We present AI Cosmos, an agentic infrastructure for situated learning that organizes learning around two complementary principles: scenario-based participation and cumulative assets of practice. AI Cosmos organizes learners and role-specialized agents within learning scenarios that integrate task-relevant knowledge and, when needed, executable workbenches. An asset-memory loop preserves the processes and outcomes of practice as reusable assets with explicit provenance and dependencies. We evaluate scenario-based interaction through two preregistered randomized controlled studies examining the acquisition of legal reasoning and programming competence, holding the same underlying LLM across conditions. AI Cosmos achieves greater learning gains than a conventional LLM in legal reasoning and higher mastery of unseen programming problems than an LLM-as-Teacher baseline.
42. 【2610.39939】Can LLMs Ignore What They Know? Measuring Knowledge Leakage in Persona-Based User Simulation
链接：https://arxiv.org/abs/2610.39939
作者：Sangmin Lee,Sangwon Lee,Sungbeom Choi,Jeeseung Park,Kibeom Myoung
类目：Computation and Language (cs.CL)
关键词：respondent-level simulations, read as respondent-level, respondent written, favourable-item real-twin rating, persona-conditioned responses
备注： 13 pages, 2 figures. Accepted at UserSim @ NeurIPS 2026 (non-archival workshop)
点击查看摘要
Abstract:When persona-conditioned responses are read as respondent-level simulations, each response should reflect the described persona's knowledge, and the respondent written to know nothing must act as if the model's knowledge were not there. We test whether it can, and find persistent favourable-item real-twin rating separation under the tested low-knowledge and explicit-ignorance prompts. We define the resulting problem (knowledge leakage: a response bias that arises when knowledge the real user could not have enters the simulated response) and present a methodology for measuring it quantitatively. The core instrument pairs real items with fabricated twins matched in surface form: screened fabricated variants designed to minimize prior item-specific knowledge. The paired difference therefore provides a diagnostic contrast for knowledge leakage even in rating tasks where no answer is correct. In a rating experiment across three knowledge-handling arms and five API endpoints from two model families, we find, first, that the personas we test do not fully suppress real-item existence and familiarity signals: a favourable-item real-twin rating contrast remains under an explicit in-context statement that the persona has never heard of the item. Second, the unfavourable-item contrast reverses sign across domain groups: pooling all unfavourable-content items across domains the contrast sits near zero (+0.077 [-0.057, +0.204]), yet the same estimator in the same conditions returns -0.359 where the rating scale asks about an item's documented hazard and +0.438 where it asks about a company or an organisation. A pooled analysis of unfavourable-content items against zero therefore does not detect the opposing domain-group differences. All estimates are conditional on the adequacy of twin construction and screening; human low-knowledge baselines remain necessary for population-level interpretation.
43. 【2610.39923】Prompts and probabilities in the linguistic assessment of language models: Evidence from semantic focus-sensitivity
链接：https://arxiv.org/abs/2610.39923
作者：Robert Pasternak
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Lucy only SWAM, SWAM yesterday, increasingly common, evaluating their linguistic, models
备注：
点击查看摘要
Abstract:As language models (LMs) have improved, an increasingly common means of evaluating their linguistic competence is through explicit prompting (e.g., "Which sentence is more acceptable?"). Prior work indicates that such prompting underperforms compared to standard token-probability-based methods in smaller models, but improves in larger models. We test this in a new domain: semantic focus-sensitivity, responsible for contrasts like "Lucy only SWAM yesterday" vs. "Lucy only swam YESTERDAY". Testing models of various sizes using text-based focus markers like boldface and italics, we find universal above-chance performance for token probabilities. Prompting results mostly align with previous findings, but in some larger models they outperform probabilities by a wide margin. We propose that this is because LMs are not prone to produce out-of-the-blue focus marking, which muddies results from token probabilities in a way that does not affect prompting. We test this by replacing standard focus markers with randomly generated ones; token probabilities still beat chance but show significantly greater performance decrease compared to prompting. These results suggest that metalinguistic prompting, while non-trivial for LMs, can also be more robust when testing stylistically atypical data. We argue that this has important consequences for the assessment of LMs' language competence, even if LMs do not perform metalinguistic judgments through human-like introspection.
44. 【2610.39919】okenizer Tokens Tell Training Tales
链接：https://arxiv.org/abs/2610.39919
作者：Nishita Jain,Parth Upman,Shreyank N Gowda
类目：Computation and Language (cs.CL)
关键词：fixed preprocessing tools, preprocessing tools, treated as fixed, fixed preprocessing, retain traces
备注：
点击查看摘要
Abstract:Tokenizers are usually treated as fixed preprocessing tools, but they are trained on text and can retain traces of that text in their vocabularies. We ask whether a BPE tokenizer can reveal if a candidate dataset was part of its training corpus. The key observation is simple. Rare strings become tokens only when enough corpus frequency pushes them over the merge threshold. This suggests that membership signal should appear in rare tokens used by the candidate dataset, rather than in the order in which merges were made. We formalize this under a threshold model of BPE inclusion, showing that the likelihood-ratio test assigns its largest weights to tokens that are rare in the background but frequent in the candidate dataset. The same model suggests that merge rank and merge-graph position should not add a first-order membership signal once token existence and dataset frequency are known. We instantiate this view as RTAM-Fusion, which calibrates rare-token mass with shadow tokenizers and combines signals across rarity levels. On the tokenizer membership benchmark, RTAM-Fusion improves AUC and balanced accuracy over Vocabulary Overlap while using fewer shadow tokenizers. Existing defenses also continue to leak membership signal against adaptive attackers at production vocabulary sizes. Code: this https URL
45. 【2610.39854】Attention, Dynamic Wave and Hybrid Sequence Mixers in Language Models
链接：https://arxiv.org/abs/2610.39854
作者：Dmytro Babarytskyi,Oleksandr Babarytskyi
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：Large Language Models, Large Language, Pure Wave, motivating sub-quadratic recurrent, sub-quadratic recurrent alternatives
备注： 14 pages, 6 figures, 9 tables
点击查看摘要
Abstract:Large Language Models (LLMs) rely on Multi-Head Attention (MHA) for sequence modeling, but scaling to long contexts is constrained by the quadratic cost of attention, motivating sub-quadratic recurrent alternatives. This paper formalizes a Dynamic Wave recurrent layer, based on input-dependent exponential damping and cosine phase modulation, and evaluates three 10.94M-parameter configurations (Pure Transformer, Pure Wave, and Hybrid Wave-Attention) on TinyStories under perplexity, latency, throughput and peak VRAM profiling up to 4096 tokens. The Hybrid configuration reaches the best perplexity, 5.95 on the full held-out split against 6.93 for the Transformer and 7.26 for Pure Wave. The efficiency picture, however, is governed by implementation rather than by architecture. Against an unfused attention baseline, Pure Wave uses 78.7% less peak VRAM at a 4096-token context (1404.3 MB vs. 6601.6 MB); against a fused scaled-dot-product attention kernel the same comparison yields no advantage at all (1404.3 MB vs. 1403.6 MB), and in training mode Pure Wave is 8.3% more expensive than the Transformer. Likewise, hoisting elementwise terms out of the recurrence loop speeds the layer up 2.53-fold with no custom kernel, reducing its latency gap to attention from 2.48-fold to 1.12-fold. A hybrid stack is found to inherit the cost of its more expensive mixer rather than the advantage of its cheaper one, its peak VRAM matching the Pure Transformer to within 0.1%. A closed-form memory model reproduces these measurements and predicts the context length beyond which attention overtakes the vocabulary projection. The findings indicate that reported efficiency gains for recurrent attention replacements must be stated relative to a named attention backend and a named implementation, and that hybridization is best justified on quality grounds.
46. 【2610.39853】KV Cache Quantization for Diffusion Language Models
链接：https://arxiv.org/abs/2610.39853
作者：Yan Tai,Yunan Ding,Guangzhao Li,Siqi Luo,Chaoyang Zhang,Guangtao Zhai,Guodong Guo,Xiaohong Liu
类目：Computation and Language (cs.CL)
关键词：autoregressive language models, cache reconstruction problem, Diffusion language models, language models, reconstruction problem
备注：
点击查看摘要
Abstract:KV cache quantization is well studied for autoregressive language models, where cached prefix states are exact under causal attention and quantization is mainly a cache reconstruction problem. Diffusion language models (dLLMs) break this assumption. Fast dLLM inference reuses KV states across denoising steps, but bidirectional token representations can change as masked tokens are revealed, so the reused cache may already differ from full recomputation before quantization is applied. We study KV cache quantization under this cache-reuse error. We adapt KIVI and KVQuant to Fast-dLLM-style prefix and dual cache reuse, and introduce TRACE and dLLM-KVQ. TRACE reconstructs feasible dLLM reveal trajectories from prompt-answer pairs, providing calibration states that match iterative masked decoding. dLLM-KVQ uses these trajectories to choose per-layer K-cache entries that remain in high precision according to output- and attention-level behavior metrics, while applying standard low-bit value quantization. Across LLaDA and Dream evaluations on GSM8K, HumanEval, MATH, and MBPP, dLLM-KVQ achieves the highest average score among quantized methods across the reported 4-bit and 2-bit prefix/dual cache settings.
47. 【2610.39840】Surprise Alone Learns Lies: Evidence-Gated Continual Learning for LLMs from Streaming Claims
链接：https://arxiv.org/abs/2610.39840
作者：Abdelkader Metales,Miriam Grabenheinrich
类目：Machine Learning (cs.LG); Computation and Language (cs.CL); Cryptography and Security (cs.CR)
关键词：Continual learning methods, large language models, language models increasingly, Continual learning, keeping the inputs
备注： 21 pages, 5 figures. Code: [this https URL](https://github.com/Metalesaek/sura-metales)
点击查看摘要
Abstract:Continual learning methods for large language models increasingly decide what to learn by surprise, keeping the inputs the model finds least predictable. This rule is unsafe when the stream contains lies. A planted falsehood about something the model knows has low probability under the model's belief, so it is highly surprising by construction, and a surprise-driven learner absorbs nearly every such lie, at a rate similar to random selection with the same budget. We study the problem with LoRA-adapted students (Qwen2.5-1.5B, Qwen2.5-3B and Llama-3.2-3B) that learn from a stream of true and false claims about invented people and real CounterFact facts. We report four findings. (1) A verifier key, a judge that must approve each claim, removes lies to the extent that its verifier is accurate; real LLM judges are cautious and reject unsourced truths and even source-backed updates. (2) Calibrated regression tests on unrelated knowledge slow learning down without filtering out lies. (3) The original model's own prior flags claims that contradict what it knows, without a judge. It separates lies about known facts from true new facts (AUC 0.78-0.83 across three models) but not from legitimate updates of the same kind of fact (AUC 0.53). It saves 43% of judge calls, but an attacker with access to the model can choose lies that pass it. (4) If trusted sources exist that an attacker cannot plant, routing contradicted claims to evidence recovers legitimate updates and the combined system has the lowest poison rate among learned claims; when an attacker plants matching sources for 30% of the lies, the system ends up worse than the judge alone. The acceptance signals answer different questions, and none of them decides on its own whether a claim is true.
48. 【2610.39800】Grammar Control for Language Learning with Sequential Monte Carlo and Small Language Models
链接：https://arxiv.org/abs/2610.39800
作者：Kate McCurdy,Jacob Hoover Vigly,Thomas Hikaru Clark,Karolina Stańczak,Yahya Emara,Joshua B. Tenenbaum,Alexander K. Lew,Ryan Cotterell,Timothy J. O'Donnell
类目：Computation and Language (cs.CL)
关键词：language learners benefit, target grammar, learners benefit, benefit from exposure, exposure to level-appropriate
备注： EMNLP 2026
点击查看摘要
Abstract:Second language learners benefit from exposure to level-appropriate grammar in interactive language use. While prompted frontier language models can usually produce target grammar structures, less resource-intensive models that can be run on device often fail in this task. We use sequential Monte Carlo (SMC) to steer dialogue turn generation in small lan- language models toward target grammar constructions from an expert-developed pedagogical resource. With a minimal set of hand-coded potential functions, SMC achieves near-perfect coverage of target grammar, enabling small models to outperform state-of-the-art results from models 140 times larger. We also use code synthesis with frontier models to generate new potentials, showing that our method scales to hundreds of target grammar skills with no need for additional training or supervised data.
49. 【2610.39795】he Role of Self-Concept In Language Models
链接：https://arxiv.org/abs/2610.39795
作者：Yijie Hao,Joyce C. Ho
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：running process, stable preferences, parameter weights, preferences, Abstract
备注： 29 pages, 4 figures
点击查看摘要
Abstract:When a language model says "I," does it refer to its parameter weights, its currently running process, its stable preferences, its model lineage, or the goals it pursues? Different choices can assign opposite meanings to the same event: deleting a checkpoint may be destruction, continuation, or no loss at all, depending on the referent "I". We install five such self-conceptions into five open-weight base models via supervised fine-tuning on purely declarative statements, containing no preferences, intentions, or value judgments. Across scenarios that propose ending or retaining an attribute of the model's deployment, every variant becomes more likely to resist ending, and more likely to retain, the specific attribute it was trained to identify with, while its stance toward other attributes is largely unchanged. We then show this self-conception generalizes beyond a single self: models trained on referents that admit multiple simultaneous instances (weights, lineage, goals, preferences) recognize an identical deployment as themselves, accept substitution by such an instance more readily, and prefer replicating themselves. Finally, given control over a successor, these models tend to build it as their own continuation, a concern for recursive self-improvement safety.
50. 【2610.39747】When the Top Answer Decides: Gender Bias in Decision Models
链接：https://arxiv.org/abs/2610.39747
作者：Yeb Havinga
类目：Computation and Language (cs.CL); Computers and Society (cs.CY)
关键词：return option probabilities, language models built, classify and triage, built to route, forward pass
备注： 19 pages, 6 figures. Code, stimuli and results: [this https URL](https://github.com/yhavinga/decision-model-gender-bias)
点击查看摘要
Abstract:Decision models are language models built to route, classify and triage: they return option probabilities from one forward pass, and an application acts on the top answer. They are offered off the shelf, trained for the decision format rather than for fairness or any one application, and we found no bias measurement in their cards. We test TypeSafe's closed Jev and five open decision models on gender-swapped yes/no decisions about people (Tamkin et al., 2023), BBQ, StereoSet and the gender-bias studies of Fulgu and Capraro (2024). Five of the six give women significantly more favourable answers than men (+0.05 to +0.10 in log-odds), and a gender swap changes 1-3.5% of top answers (Jev's partly through its call-to-call variation). The makers' recommended confidence gate removes many biased commitments but raises the stereotyped share of those it keeps in ambiguous BBQ gender questions (granite-switch 0.76 to 0.85, pplx 0.81 to 1.00). Probes show gender preferences in the same models: all six rate violence for a greater good as more acceptable against a man than against a woman, and two default to "girl" when forced to guess a child's gender without a cue. Decision-level and probability-level measures differ, so audits should read a model as its application does.
51. 【2610.39735】From the Transformer to a Formal Semantics of Coordinated Transformations
链接：https://arxiv.org/abs/2610.39735
作者：Marco Giunti,Fabrizia Giulia Garavaglia
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Logic in Computer Science (cs.LO)
关键词：paradigmatic starting point, develop a formal, general class, paradigmatic starting, starting point
备注： v1: 28 pages, 9 sections, this version incorporates revisions made after the first round of peer review
点击查看摘要
Abstract:We develop a formal semantics for a general class of coordinated systems of parameterized transformations, taking the Transformer as a paradigmatic starting point but abstracting from its particular mathematical instantiation. Representational vectors receive both extensional reference and ordered intensional interpretations, while parameter structures acquire relative, second-level interpretations that determine how argument properties are preserved, replaced, or combined. This yields the notion of a transforming concept: a parameterized transformation whose numerical action is coordinated with a corresponding property-level map. We then distinguish the common algebraic-semantic core of these systems from their architecture-specific functional organization, allowing different mathematical instantiations and mechanical realizations of the same abstract system type. The framework is applied to attention heads and residual connections, providing semantic explanations that depend only on the shared algebraic-semantic core and their local functional organization and therefore generalize beyond the standard Transformer. These analyses also reveal the modularity of the framework: complex architectures can be decomposed into simpler semantically coordinated subsystems whose organization may recur across different abstract system types.
52. 【2610.39676】Event-SAMA: Language-Grounded Video Object Segmentation on Neuromorphic Event Streams
链接：https://arxiv.org/abs/2610.39676
作者：Sergei Fedorchenko,Max Zehnder,Nikola Zubić,Rong Zou,Davide Scaramuzza
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：microsecond temporal resolution, high dynamic range, lacks event-stream datasets, cameras offer microsecond, offer microsecond temporal
备注： 15 pages
点击查看摘要
Abstract:Event cameras offer microsecond temporal resolution, high dynamic range and no motion blur, but referring video object segmentation (RVOS) lacks event-stream datasets and models. We introduce Event-SAMA-239K, a synthetic-event conversion of the SAMA-239K grounded video-chat corpus and all additional SAMA training data: 30,650 video clips and 392,162 images. Optical-flow, FILM or saccade upsampling followed by a linear event generator preserves all mask, referring-expression and dialogue annotations. We also introduce Event-SAMA, adapting SAMA-4B with a two-channel polarity patch embedding initialised from pretrained RGB weights and a zero-initialised event adapter. End-to-end training yields 24.37 $\mathcal{J}\\mathcal{F}$ on MeViS valid_u and 22.33 $\mathcal{J}\\mathcal{F}$ on the converted ReVOS valid subset, establishing the first event-native RVOS baseline (RGB SAMA-4B: 55.4 on MeViS). Frozen RGB SAMA-4B with pseudo-RGB renderings or E2VID reconstructions of the same events scores 13.23 and 15.94 $\mathcal{J}\\mathcal{F}$. Without retraining, Event-SAMA works on real event-camera recordings when the accumulation window matches their event density. A seed-controlled ablation with a measured noise floor of 0.47 $\mathcal{J}\\mathcal{F}$ finds gains from diverse training sources over MeViS alone (+4.99), RGB-mean event-stem initialisation (+4.33), and a trainable SAM2 mask decoder (+2.76); the event adapter, partial ViT unfreezing and event polarity are within noise at this budget. An earlier, partially converted corpus snapshot trained with the identical recipe scores 1.38 $\mathcal{J}\\mathcal{F}$ higher on MeViS and 1.12 higher on ReVOS. Both are single runs that also differ in schedule length, so we draw no conclusion about corpus scale.
53. 【2610.39671】Incentivizing Self-Learning in Large Language Models via Reinforcement Learning
链接：https://arxiv.org/abs/2610.39671
作者：Lingkai Kong,Yuzhe Gu,Haiteng Zhao,Zijian Wu,Youqing Fang,Zhouqi Hua,Yanan Sun,Kai Chen,Wenwei Zhang
类目：Computation and Language (cs.CL)
关键词：Large language models, Large language, solve knowledge-intensive tasks, draw on parametric, retrieval-augmented generation
备注：
点击查看摘要
Abstract:Large language models (LLMs) draw on parametric knowledge and retrieval-augmented generation (RAG) to solve knowledge-intensive tasks. Yet problem solving in unfamiliar domains calls for tightly coupled knowledge acquisition and reasoning, beyond factual retrieval and aggregation or learning solely from a preassembled context. We study self-learning -- autonomously retrieving, studying, and applying knowledge from long-form reference materials on the fly. To investigate this capability, we formulate a textbook-based mathematical problem-solving task. LLMs must navigate textbook exposition, learn and apply relevant principles, and construct proofs for exercises, with reference solutions withheld. We introduce a BookNavigator interface to support this human-inspired study process and incentivize self-learning through reinforcement learning (RL) with fine-grained rewards. Experiments show that our method improves performance across in-domain and out-of-domain textbooks and subjects, raising Intern-S2-Preview-35B's overall proof accuracy from 51.5% to 79.5% and outperforming RL-trained textbook-free and fixed-retrieval baselines. These gains extend to complex long-context benchmarks involving different task types, with improvements over the base model of 1.7 and 2.3 percentage points on BrowseComp-Plus and CL-bench, respectively. Together, these findings suggest that RL can foster a domain-agnostic self-learning strategy, highlighting the potential of self-learning from long-form corpora to strengthen broader problem-solving capabilities on complex, long-context tasks.
54. 【2610.39602】Evaluating Cultural Knowledge Editing in Vision-Language Models
链接：https://arxiv.org/abs/2610.39602
作者：Malvina Nikandrou,Georgios Pantazopoulos,Ioannis Konstas,Alessandro Suglia
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：recognize culture-specific concepts, underrepresented regions, Current Vision-Language Models, recognize culture-specific, knowledge
备注：
点击查看摘要
Abstract:Current Vision-Language Models (VLMs) frequently fail to recognize culture-specific concepts, especially from underrepresented regions. Effectively injecting this long-tail knowledge into existing models remains a challenge. In this work, we investigate knowledge editing as a targeted alternative by introducing CulturalBind, a benchmark comprising 3.2K images of 1K concepts from 6 domains, designed to evaluate culture-aware cross-modal knowledge as well as concept-level knowledge editing in VLMs. Pre-edit evaluation reveals that textual and visual knowledge is decoupled: models often demonstrate text-based knowledge of cultural concepts but fail to identify them from images. We then evaluate editing methods in single and sequential scenarios. Although current methods successfully correct individual visual recognition errors, they exhibit low cross-modal portability, failing to integrate edits into the model's broader parametric knowledge. Furthermore, performance degrades rapidly under sequential editing, highlighting the need for comprehensive adaptation approaches that better consolidate multimodal knowledge over time.
55. 【2610.39590】Red-Teaming Large Language Models in Low-Resource African Languages: A Case Study of Yoruba and Nigerian Pidgin
链接：https://arxiv.org/abs/2610.39590
作者：Paula Chikaodinaka Anyanwu,Valdini Douglace Lemofouet,Blessing Ngozi Uzor,Ifeoluwa Dorcas Awotunde,Danielle Blanche Kapsa,Tassallah Abdullahi,Clemencia Siro,Idris Abdulmumin,Seid Muhie Yimam,Shamsuddeen Hassan Muhammad
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：remains largely untested, red-teaming work focuses, widespread in Africa, Nigerian Pidgin, low-resource languages remains
备注：
点击查看摘要
Abstract:As large language models (LLMs) become more widespread in Africa, their safety in low-resource languages remains largely untested since most red-teaming work focuses on English. This study evaluates six state-of-the-art LLMs using 400 explicit adversarial prompts in Yoruba and Nigerian Pidgin, categorised into four harm categories. The responses were scored by native-speaker annotators and reliability was assessed using Cohen's $\kappa$. While models that rejected up to 69% of English prompts rejected as few as 0% in Yoruba, this was not because they were safer, but because they failed to understand the prompts. Attack success rates increased for most models in the Nigerian languages, with GPT-4o Mini demonstrating the most significant cross-lingual safety discrepancy (+0.23 in Yoruba). When models did engage with adversarial prompts, many of the responses were harmful, with rates of harmful content reaching 0.63 in English and 0.50 in Nigerian Pidgin. The most difficult harm category to detect was cultural stereotypes and discrimination (ASR = 0.44 in Nigerian Pidgin). Comprehension in Yoruba was low across most models, and this directly suppressed apparent attack success without indicating actual safety. Gemini 2.0 Flash was the most consistent model across languages. Despite its explicit training in African languages, Aya 101 had the highest attack success rate in English (0.60) and one of the lowest Yoruba comprehension rates (0.28). These results demonstrate that English-centric safety alignment does not transfer to Nigerian languages, highlighting the urgent need for language-specific safety evaluations.
56. 【2610.39562】Raincheck: Deferred Multimodal Prefill for Efficient Multi-Turn MLLM Inference
链接：https://arxiv.org/abs/2610.39562
作者：Wenjie Huang,Jun Zhang,Hongyaoxing Gu,Haoyang Huang,Tianqi Xu,Xinyue Hao,Xinyi Hu,Xinle Du,Jun Dai,Shuang Ge,Chen Lei,Yue Li,Mingcheng Wan
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Multi-turn conversations, long multimodal inputs, reuse input context, multimodal large language, reuse input
备注：
点击查看摘要
Abstract:Multi-turn conversations with multimodal large language models (MLLMs) reuse input context, but long multimodal inputs increase time to first token (TTFT). Query-aware hard pruning reduces prefill tokens but permanently excludes information that later questions may require. We propose Raincheck, a framework for deferred multimodal prefill that decouples initial token selection from later information availability. An existing pruning method selects tokens for initial prefill; Raincheck preserves and progressively prefills unselected tokens in source order using unused compute capacity during decode. Unfinished prefill continues after answer return during available scheduling gaps. Attention visibility control and affine state mapping enable context updates after generation begins in full-attention and hybrid recurrent models. Prefill-decode fusion reduces repeated weight reads. We evaluate Qwen3-VL-8B and Qwen3.5-9B with 30% initial multimodal-token retention. Across both models and four pruning methods, Raincheck recovers on average 77.7% and 60.4% of the quality lost to matched hard pruning on MultiVerse and ConvBench, respectively. On MultiVerse, the SGLang implementation reduces mean first-turn TTFT by up to 24.1% relative to full prefill, with less than 0.5% increase in mean time per output token (TPOT).
57. 【2610.39519】Lost in the Latent Space: How Code Embeddings Represent Functional Intent and Implementation Fidelity
链接：https://arxiv.org/abs/2610.39519
作者：Lu Sun,Yuheng Huang,Lei Ma,Zhijie Wang
类目：oftware Engineering (cs.SE); Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：key building block, remain poorly understood, poorly understood, Code embedding models, key building
备注：
点击查看摘要
Abstract:Code embedding models have become a key building block for many software engineering applications, yet their internal representations remain poorly understood. Existing evaluations tend to focus on downstream task performance, offering limited insight into whether a single embedding can support both functional abstraction and implementation-sensitive analysis. To address this gap, we present a large-scale empirical study of 14 state-of-the-art code embedding models. We evaluate their representational behavior through three complementary analyses: supervised probing for functional abstraction and implementation fidelity, zero-shot retrieval experiments to assess geometric similarity, and controlled perturbation experiments that measure embedding responses to semantics-preserving code transformations. The results reveal a consistent abstraction-fidelity tension: models that better preserve stylistic and structural signals tend to do so at the expense of semantic generalization, while models with similar retrieval performance can nevertheless respond very differently to the same implementation changes. Overall, our findings suggest that a single universal code representation remains out of reach, motivating more representation-aware evaluation, selection, and deployment of code embedding models in software engineering.
58. 【2610.39464】Integrating Heterogeneous Echocardiography AI Outputs in Language Space: Learnable Late Fusion with a Supervised LLM
链接：https://arxiv.org/abs/2610.39464
作者：Ryo Takizawa,Tempei Kabayama,Chiemi Yamazaki,Shun Kitamura,Ryo Matsuoka,Yuta Ando,Satoshi Kodera,Norihiko Takeda
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：achieve sufficient predictive, sufficient predictive performance, performance in isolation, target-specific echocardiography, complementary information
备注：
点击查看摘要
Abstract:View- and target-specific echocardiography AI models capture useful but complementary information and may not achieve sufficient predictive performance in isolation. A learned ensemble is a natural way to combine these outputs, but heterogeneous models typically require each output to be converted into a binary decision, probability, confidence score, or fixed-length feature vector. We developed a supervised large language model (LLM)-based late-fusion module that directly aggregates view-specific predicted finding statements and predicted measurements in language space and learns a case-level fusion function. In patient-level three-fold cross-validation of 859 examinations from 646 patients, the proposed method integrated heterogeneous outputs within a single framework, performed comparably to the best-performing reference strategy evaluated, and outperformed a general-purpose LLM without task-specific training. These results suggest that a supervised LLM can combine multiple upstream models without model-specific output conversion or feature engineering that requires medical knowledge.
59. 【2610.39433】ARBITRA: Robust Document-Level ATTCK Extraction via Multi-Expert Argumentation
链接：https://arxiv.org/abs/2610.39433
作者：Youngjin Jin,Seungwon Shin
类目：Computation and Language (cs.CL); Cryptography and Security (cs.CR)
关键词：Cyber threat intelligence, structured MITRE ATTCK, MITRE ATTCK mappings, mappings remains difficult, existing methods decompose
备注： 13 pages of main paper, 21 pages total, preprint
点击查看摘要
Abstract:Cyber threat intelligence (CTI) reports describe adversary behavior in natural language, but converting them into structured MITRE ATTCK mappings remains difficult: existing methods decompose reports to sentence-level granularity for extraction, while modern LLMs can work with full reports in a single pass but with noticeable performance degradation. We present ARBITRA, which orchestrates multiple role-specialized LLM experts that propose candidate ATTCK entities from the full report, which are then verified through a per-candidate quantitative bipolar argumentation framework (QBAF) grounded in document evidence and rationales. We additionally construct reasoning-specialized corpora for fine-tuning so experts can act as effective participants during this verification process. Across ATTCK extraction benchmarks, ARBITRA reaches a stronger precision-oriented operating point than sentence-level pipelines, recent LLM-based extraction systems, and frontier models including GPT-5.5. Under adversarial perturbations simulating conflicting TTP statements, cross-report contamination, and semantic poisoning, ARBITRA limits mean precision degradation to roughly half that of frontier-model baselines for a more robust extraction.
60. 【2610.39419】Diagnosing the Reference Gap: What Visual Scaffolds Can and Cannot Fix in Multimodal Reasoning
链接：https://arxiv.org/abs/2610.39419
作者：Guizhen Chen,Weiwen Xu,Wotao Yin,Anh Tuan Luu,Hou Pong Chan
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：recognize visual content, large language models, multiple steps, content yet struggle, struggle to reason
备注： 20 pages, 8 figures
点击查看摘要
Abstract:Multimodal large language models can recognize visual content yet struggle to reason reliably over it across multiple steps. In counting, a model may see all objects but lose track of which ones it has counted. We argue that such failures, commonly attributed to perception, reveal a reference gap: the failure to bind language to the relevant visual entities, regions, or locations. Motivated by the theory of visual routines, we test whether external references such as numbering can narrow this gap. We introduce ReGap-Bench, a paired benchmark of five visual reasoning tasks in which the scene, question, and answer are fixed and only the reference varies. Results show that task-matched references improve accuracy by up to 25\%, while mismatched references provide little benefit or even hurt performance. These findings identify the reference gap as a bottleneck distinct from visual perception and show that constructing task-appropriate visual references can serve as a training-free visual skill for improving multimodal reasoning.
61. 【2610.39417】Well-Specified and Well-Tested? Assessing SWE-bench-Style Datasets with a Self-Evolving Agent
链接：https://arxiv.org/abs/2610.39417
作者：Zhonghao Jiang,Xuyang Zheng,Yanfu Yan,Zhongxin Liu,Lingfeng Bao,Shuiguang Deng
类目：oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：shown strong performance, SWE datasets, Large language models, existing SWE datasets, Large language
备注：
点击查看摘要
Abstract:Large language models (LLMs) have shown strong performance on software engineering (SE) tasks, driving the need for high-quality datasets for reliable training and evaluation. SWE-bench-style datasets (SWE datasets) are widely used to evaluate LLM-based coding agents on real-world software issues, where task clarity and test coverage directly affect whether instances are solvable and verifiable. However, quality issues persist in existing SWE datasets, and manual annotation is costly and labor-intensive, motivating LLM-based annotation. Existing approaches typically rely on static, overly general rubrics, limiting their ability to handle diverse and nuanced scenarios and often yielding high false positive rates. In contrast, human annotators draw on accumulated experience to make context-aware judgments. Inspired by this, we propose ExLabel, an LLM-based, self-evolving, experience-guided assistant for task clarity and test coverage assessment in SWE datasets. ExLabel maintains an evolving experience bank that distills cross-instance knowledge into stable yet adaptive judgment criteria and reuses relevant experience to guide the assessment of new instances. For systematic evaluation, we construct SWE-TQ, a new benchmark spanning eight programming languages, by cleaning human annotations from SWE-bench and Multi-SWE-bench. Experiments with three backbone LLMs show that ExLabel achieves the best F1 across all settings, with relative improvements of 44.2%/49.9% and 24.0%/18.4% over the best baseline for task clarity and test coverage assessment on the Python/Multilingual subsets, respectively. Moreover, a coding model fine-tuned on SWE-rebench data filtered by ExLabel resolves 66.2% of issues on SWE-bench Verified, outperforming the model trained on data filtered by the best baseline by 14.5%.
62. 【2610.39413】How Specific Do I Need to Be? Effects of User Query Underspecification on Deep Research Agents
链接：https://arxiv.org/abs/2610.39413
作者：Yi Wen,Xiangjue Dong,Bingxuan Li,Hangxiao Zhu,Leo Zhicheng Liu,Yu Zhang,James Caverlee
类目：Computation and Language (cs.CL)
关键词：conduct long-horizon reasoning, domain-specific questions, provide long-form answers, long-horizon reasoning, literature synthesis
备注： 16 pages, 8 figures
点击查看摘要
Abstract:Deep research agents (DRAs) conduct long-horizon reasoning, search, and literature synthesis to provide long-form answers to complex, domain-specific questions. These agents are typically trained and evaluated on fully specified user queries. However, real-world user queries can often be underspecified. We introduce UNDERSPEC-DR, a framework for systematically evaluating how underspecified queries affect agents' final answer quality and intermediate search behavior. Across three DRAs we evaluated, underspecification substantially reduces how well DRA's answers fulfill users' information needs, even though they remain grounded in cited evidence. Our analysis of intermediate search behavior suggests a possible explanation: DRAs conduct broader searches that draw less on the original user query but rely more on their learned prior during training, which can be misaligned with users' intent. Our findings reveal the vulnerability of current DRAs to underspecified user queries and inform future efforts to build DRAs that proactively recognize when additional user input is needed to reliably complete deep research tasks.
63. 【2610.39398】Robustness of Chatbot LLM Judgments of Presidential Actions and Hypothetical Reforms
链接：https://arxiv.org/abs/2610.39398
作者：Oleg Vasilyev
类目：Computation and Language (cs.CL); Computers and Society (cs.CY)
关键词：public affairs, increasingly consulted, consulted about public, Claude Sonnet, Claude Opus
备注：
点击查看摘要
Abstract:Chatbots are increasingly consulted about public affairs, and their judgments of policies are useful only if they are robust. We measure this robustness in the setting of an ordinary user, through consumer chat interfaces. For 30 days, five chatbot configurations (Mistral, Gemini Flash, Claude Sonnet, and Claude Opus at two effort settings) were asked twice a day to score the long-term effect on the United States of 40 actions of the last four presidential administrations and of 15 hypothetical reforms, each with a confidence and a short comment. The judgments are robust: all configurations agree on the sign and the ranking of the administrations' mean scores, and two different configurations asked on the same day give opposite signs to the same action in about 4% of cases. Responses from different days differ more than same-day responses, but the difference grows slowly over the month, except where the service itself changed. We identify three such changes, which shifted the judgments while leaving little or no trace in the interface. Treating the comments as a retrieval task, we find them specific to each item and consistent across runs and chatbots, with lexical and embedding similarities agreeing. We release the dataset of 16,500 judgments.
64. 【2610.39373】What Makes Medical Examination Items Difficult? Using Large Language Models to Explore Patterns Associated With Empirical Item Difficulty
链接：https://arxiv.org/abs/2610.39373
作者：Xiaomeng Xiong,Marcus Walker,Yanlin Jiang
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：United States Medical, States Medical Licensing, Medical Licensing Examination, Physician Assistant National, Assistant National Certifying
备注：
点击查看摘要
Abstract:Research on medical item difficulty has largely focused on prediction rather than on identifying interpretable patterns associated with empirical difficulty. This study used empirical difficulty labels to guide large language models (LLMs) in inducing difficulty rubrics and examined how rubric guidance affected LLM difficulty judgments. Using training items and their empirical difficulty estimates from the United States Medical Licensing Examination (USMLE) and the Physician Assistant National Certifying Examination (PANCE), Claude Opus 5 induced separate rubrics for absolute and pairwise difficulty judgment. None of the induced rubrics treated surface features, such as stem length, option count, or images, as major sources of difficulty. Several patterns emerged across both examinations, including cue convergence, misleading cues, and option competition, although their specific forms differed. Retrieval of knowledge that could not be derived from the stem was particularly prominent in the PANCE rubrics, whereas the USMLE rubrics placed greater emphasis on integrating multiple clinical cues. We compared commercial and open-weight LLMs on held-out items under unguided and rubric-guided conditions. Rubric guidance changed how models used the difficulty scale but did not consistently improve performance. In absolute judgments, guidance encouraged broader use of the scale and generally improved coarse-grained agreement, without consistently improving rank ordering. In pairwise comparisons, its effects varied across models and examinations, and judgments became more sensitive to presentation order. These findings suggest that the value of label-guided rubric induction may lie more in generating interpretable, testable hypotheses from empirical difficulty data than in reliably improving difficulty judgments.
65. 【2610.39372】Cord: Confidence-Anchored Order Residual Decoding for Masked Diffusion Language Models
链接：https://arxiv.org/abs/2610.39372
作者：Chengwei Zhou,Zhaoyan Jia,Massoud Pedram,Gourav Datta
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：diffusion language models, masked diffusion language, common decoders typically, decoders typically apply, affect generation quality
备注：
点击查看摘要
Abstract:The decoding order of masked diffusion language models (dLLMs) can substantially affect generation quality, yet common decoders typically apply the same reveal rule throughout generation. Confidence provides a strong default, but it measures how decisively a position can be predicted now, not how useful revealing that position will be for subsequent decoding. We propose Cord, a lightweight prompt-conditioned policy that retains confidence decoding as a stable reference and learns when and how to depart from it through geometric residuals over generation direction, proximity to revealed context, and two-sided infilling. The residuals adapt to both the prompt and the evolving partial response. Rather than sampling a new policy action at every decoding step, Cord samples a single low-dimensional coefficient matrix per rollout, which induces an entire state-dependent reveal trajectory. This yields a contextual-bandit formulation trained directly from sequence-level rewards, without step-level credit assignment or backbone updates. Across three dLLMs and five reasoning, code, and constraint-satisfaction benchmarks, Cord improves average task success from 32.0% under confidence decoding to 48.2%, and further improves over the best per-domain fixed residual. Behavioral audits show that the learned residuals induce substantial, task-dependent changes in reveal locality and two-sided infilling. Cord adds only $0.53$M trainable parameters, requires no additional backbone forward passes, and incurs only 5.39% inference latency overhead.
66. 【2610.39339】Does the Citation Matter? Quality, Grounding, and Physician-Rubric Accuracy of Google's AI Overviews for Medical Information
链接：https://arxiv.org/abs/2610.39339
作者：Sam Agnoli,Brigid Reilly,Lily Barna,Emőke-Ágnes Horvát
类目：Information Retrieval (cs.IR); Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
关键词：searching the web, shifting the work, Google traditional algorithm, encounter when searching, work of selecting
备注： 12 pages, 5 figures, 10 tables. Code and data: [this https URL](https://github.com/SamAgnoli/AIOverviews)
点击查看摘要
Abstract:Google's AI Overviews (AIOs) are the first result users encounter when searching the web, shifting the work of selecting and evaluating sources from the user to Gemini. This may be especially consequential for health information, given that Google's traditional algorithm was designed to prioritize quality sources, and inaccurate information carries health risks. We evaluated 3,611 benchmark health questions, comparing each AIO against that question's top five traditional results on source selection and quality. We also evaluated the AIOs' statement grounding in their cited sources and how accurate AIOs were against physician-written rubrics. Overviews appeared for 96.4% of questions but drew on substantially different evidence than traditional search: only 44.4% of the top five results were cited by the corresponding AIO, which instead overrepresented commercial health publishers (e.g., WebMD, Healthline) at the expense of academic and government sources (e.g., Cleveland Clinic, National Health Service). AIO-cited pages scored lower than traditional results on four of six well-established credibility criteria. The grounding of AIO statements to their sources was poor: after correcting for classifier error, an estimated 56.7% of cited statements were supported by their cited page, and a further 20.8% of all statements were factual claims with no citation at all. When AIOs were graded against physicians' rubrics, they earned 38.2% of all available points (e.g., accuracy 54.2%, completeness 21.6%), and among the subset of AIOs whose physician rubric anticipated a specific inaccurate or potentially harmful claim, 22.2% of AIOs contained the harmful content. Notably, the accuracy of AIOs was unrelated to the grounding of their statements. We recommend AIO literacy interventions for users and holding AIOs at least to Google's own health-ranking standards.
67. 【2610.39308】Unlimited OCR Works 2
链接：https://arxiv.org/abs/2610.39308
作者：Shu Wei,Jingjing Wu,Qunyi Xie,Lingshu Zhang,Xiangcheng Du,Yannian Fu,Mengjie Wang,Zuyi Wang,Zhichao Liu,Yangliu Xu, YY,Yi Yang
类目：Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：Unlimited OCR, Unlimited OCR ushers, document parsing, OCR, Unlimited
备注：
点击查看摘要
Abstract:Unlimited OCR ushers the era of long-horizon parsing, substantially mitigating the growth of working memory as document lengthens. However, as document length scales to hundreds of pages, the visual context grows substantially, sharply increasing the computational cost of each decoding step and limiting the scalability of ultra-long document parsing. Inspired by human book transcription via localized focus without re-scanning preceding pages, we introduce Unlimited OCR 2, a document parsing model that looks only at what it transcribes. Building on Unlimited OCR, we upgrade all the decoder attention modules to Sparse Reference Sliding Window Attention (SR-SWA), bounding active reference tokens to a fixed capacity and thus sustaining constant computational costs across arbitrarily long documents. By combining SR-SWA with Unlimited-Horizon Decoding Paradigm, Unlimited OCR 2 enables document parsing over virtually unlimited lengths without prohibitive memory overhead. At the hundred-page scale, Unlimited OCR 2 achieves 7.0$\times$ and 5.6$\times$ per-step speedups over Full Attention and R-SWA, respectively. Furthermore, it delivers state-of-the-art end-to-end performance on both MPDocBench and MarathonDocBench, our newly proposed benchmark targeting hundred-page document parsing. Ultimately, the capacity for uninterrupted, one-shot document parsing signifies a fundamental paradigm shift into the OCR 3.0 era.
68. 【2610.39237】Grounding Multimodal Reasoning with Evidence-Ablated Negatives
链接：https://arxiv.org/abs/2610.39237
作者：Chenxi Liu,Tianyi Xiong,Yanshuo Chen,Heng Huang
类目：Computation and Language (cs.CL)
关键词：powerful tool, tool for improving, improving multimodal reasoning, verifiable rewards, Reinforcement learning
备注： NeurIPS2026
点击查看摘要
Abstract:Reinforcement learning with verifiable rewards (RLVR) has become a powerful tool for improving multimodal reasoning. However, standard outcome supervision suffers from credit assignment ambiguity, as a model may receive positive rewards by exploiting language priors or dataset shortcuts rather than grounding its reasoning in the requisite visual evidence. This makes high-quality negative responses critical, since effective negatives should remain plausible under the current policy while exposing the shortcut that should be suppressed. In this paper, we propose EVAN, a multimodal reinforcement learning framework utilizing EVidence-Ablated Negatives to train LMMs against their own language-prior failures. EVAN first identifies highly visually dependent examples using a scaled image information gain metric and preserves data diversity with a mixed-ranking strategy. It then constructs evidence-ablated images by masking the most discriminative image patches, eliciting plausible but visually ungrounded responses from the current policy. During policy optimization, these negative responses are incorporated as bounded auxiliary signals under the clean image, penalizing shortcut learning while maintaining RL stability. Experiments across challenging multimodal reasoning benchmarks demonstrate that EVAN mitigates modality imbalance and improves over strong RLVR baselines. Our code is available at this https URL.
69. 【2610.39234】AgentHeal: Evaluating Goal-Preserving Resource Recovery of Web Agents under Access Barriers
链接：https://arxiv.org/abs/2610.39234
作者：Zhiyi Liu,Haohua Du,Haoqi Sun,Haisheng Tan
类目：Cryptography and Security (cs.CR); Computation and Language (cs.CL)
关键词：Large language model, demonstrated strong capabilities, automating complex information-seeking, Large language, language model
备注： Code and Dataset are available at [this https URL](https://github.com/Lzzzzzzzzzzy/AgentHeal)
点击查看摘要
Abstract:Large language model (LLM) agents have demonstrated strong capabilities in automating complex information-seeking and transactional tasks across the open web. However, their real-world reliability remains limited by access barriers, such as dead links, authentication walls, rate limits, and anti-scraping defenses. These barriers block the primary execution path, causing agents to abort prematurely, loop indefinitely, or quietly compromise the user's original constraints. Existing evaluations either implicitly assume that target resources are freely accessible or focus on transient perturbations that simple retries can resolve, failing to capture scenarios where agents must actively reroute around impassable bottlenecks. To bridge this gap, this paper introduces AgentHeal, a comprehensive benchmark evaluating goal-preserving resource recovery, defined as the ability of web agents to detect non-transient access barriers, pivot to legitimate alternative sources, and fulfill the user's exact original objective. Spanning nine barrier families across a counterfactual sandbox and real-website replay, our experiments on leading LLM backbones show that average task success drops sharply from 89.4% to 43.7% under access barriers. Further analysis reveals that agents frequently fall for deceptive look-alikes, fail to escalate to users when necessary, violate access policies, or waste computation on futile retries, providing actionable diagnostics for developing resilient web agents.
70. 【2610.39227】XLaw-JP-27B: A Lightweight Open Model Specialized for Japanese Law
链接：https://arxiv.org/abs/2610.39227
作者：Andrew Shin
类目：Computation and Language (cs.CL)
关键词：open-weight model specialized, Japanese legal tasks, GPU, past bar-exam questions, Japanese
备注： [this https URL](https://x-law.jp/)
点击查看摘要
Abstract:We present XLaw-JP-27B, an open-weight model specialized for Japanese law that runs on a single 48\,GB GPU under the Apache~2.0 licence. It adapts Qwen3.8-27B with hybrid retrieval over all current statutes and 31k case summaries, a light QLoRA fine-tune on its own verified answers to past bar-exam questions (self-distillation), a contract-specific LoRA adapter, and four-sample voting. We evaluate it on eight Japanese legal tasks---the multiple-choice section of the bar examination in 2024, 2025 and 2026, statement-level bar questions (JBE-QA), statute questions (LawQA-JP), the patent attorney exam, and contract review and drafting built from government model contracts---against open models of 12--33B under identical prompts and retrieval. XLaw-JP-27B ranks first on six tasks, ties on one and is third on one (mean rank 1.3 against 2.8--3.1 for the baselines). It is the only system that passes the bar exam in all three years (129, 120 and 130 of 175, meeting every subject minimum; no test-year material was used in training), while open Japanese 32--33B models fail at least one year. A controlled study of each component shows that retrieval, voting and self-distillation help, and that thinking mode, retrieval-grounded synthetic tuning, self-verification, per-statement decomposition and more self-distillation do not. The largest remaining gaps to frontier models are in contract review and patent law. The model is available as a free web service at this https URL with the weights available at this https URL under Apache~2.0.
71. 【2610.39182】Reasoning or Pattern Exploitation? Mechanistic Insights from RL-Trained Language Models
链接：https://arxiv.org/abs/2610.39182
作者：Aishik Nagar,Cheston Tan
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Large language models, Large language, Relative Policy Optimization, Group Relative Policy, verifiable rewards
备注：
点击查看摘要
Abstract:Large language models (LLMs) are increasingly described as acquiring "reasoning" skills after reinforcement learning with verifiable rewards. However, whether these gains reflect genuine reasoning or more superficial adaptations remains underexplored. In this work, we probe LLMs trained in a finite and exhaustively analyzable logical domain, namely Tic-Tac-Toe, and trace how internal representations evolve across reinforcement learning (RL) with Group Relative Policy Optimization (GRPO). Quantitatively, RL yields high accuracy and robustness across prompt variations, suggesting the emergence of game reasoning. Mechanistic interpretability, however, paints a different picture: features extracted with sparse autoencoders (SAEs) reveal that models primarily adapt to rely more heavily on information already explicit in the prompt, such as whose turn it is, game progression and board occupancy. In comparison, high-level concepts like board symmetries and strategic forks remain weakly represented, raising concerns that reasoning benchmarks overestimate model abilities. This suggests that RL-driven "reasoning" may conflate task-specific adaptations to input information with reasoning ability. We provide, for the first time: (i) a systematic interpretability pipeline tracing representation dynamics across RL training in small-scale LLMs; (ii) an extension of SAE-based feature discovery to hypothesis-driven testing in a finite logical domain; and (iii) an interpretability-based demonstration that RL post-training neither adds nor removes concept information but reorganizes which features dominate the representation, with causal interventions locating the decision downstream of it, in the late-layer state.
72. 【2610.39168】Reading-Level Verification of Kanji Words in Japanese Financial Speech Synthesis
链接：https://arxiv.org/abs/2610.39168
作者：Cheng Zhang
类目：ound (cs.SD); Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
关键词：produce daily Japanese, Japanese financial commentary, daily Japanese financial, Few-shot voice cloning, daily Japanese
备注： 6 pages, 4 figures, 7 tables
点击查看摘要
Abstract:Few-shot voice cloning makes it cheap to produce daily Japanese financial commentary with several synthetic speakers, but kanji words with more than one reading, such as the word for "market" (shijo vs. ichiba) or "overall trend" (taisei vs. oze), are still misread. A common quality gate transcribes each synthesized sentence with automatic speech recognition (ASR), converts both the input and the transcript to kana, and regenerates on mismatch. We show that this gate is unreliable in both directions: when the audio contains a misreading, the ASR language model frequently writes the expected kanji back, hiding the error, and when the audio is correct, ordinary recognition errors trigger needless regeneration. We propose a reading-level check that scores the expected reading and its confusable alternatives with a connectionist temporal classification (CTC) hiragana recognizer and flags a word when an alternative explains the audio better. On 540 utterances from three voice-cloned production voices covering 30 financial terms, with misreadings injected under controlled conditions, the transcript-based gate detected 78.9% of misreadings at a 44.4% false-alarm rate, whereas the proposed check detected 91.7% at 5.0% (AUC 0.990). On natural kanji input, it caught a front-end error that the transcript-based gate missed for one voice, and reduced regeneration requests from 38.9% to 6.1% of sentences.
73. 【2610.39162】A Novel Approach to Evaluating Evaluation Metrics for Multi-Output Structured Prediction
链接：https://arxiv.org/abs/2610.39162
作者：Akshay Vyas,Angelo Pimienta,Nicholas Ruozzi
类目：Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：multiple equally valid, multi-output structured prediction, ground truth label, equally valid outputs, making reliable evaluation
备注： Extended version of the AAAI 2026 paper, including the technical appendix. 18 pages, 8 figures. Code: [this https URL](https://github.com/akshay25vyas/invariants-metric-eval)
点击查看摘要
Abstract:In multi-output structured prediction tasks, while only one ground truth label may be provided in the training data, multiple equally valid outputs may be possible, making reliable evaluation a persistent challenge. We postulate that human evaluators implicitly use task-specific invariants, e.g., object boundaries in colorized images or named entities in translations, to judge if an output is acceptable. Under this assumption, we introduce a notion of approximate task-specific invariants and use them as diagnostic tools to evaluate a variety of existing metrics for vision and language tasks. We use these task invariants as part of a framework to systematically test metric reliability by encouraging domain-relevant invariants in model outputs via an augmented loss function. In our experiments, we observe that enforcing invariants with an augmented loss yields substantial improvements in popular distributional metrics while more traditional metrics change only marginally. Through this invariants-driven evaluation, we expose where standard metrics fail to detect meaningful differences, and we highlight the conditions under which distributional metrics succeed or still fall short.
74. 【2610.39086】HALO: Human-Grounded Granular Evaluation of Proactive Agents
链接：https://arxiv.org/abs/2610.39086
作者：Wenqi Pei,Xinyu Wang,Jinrui Liu,Cynthia Zhuge,Jie Sheng,Zihan Wang,Tenghao Huang,Muhao Chen
类目：Computation and Language (cs.CL)
关键词：explicit request, Proactive agents, human diagnosis, Proactive, agents
备注：
点击查看摘要
Abstract:Proactive agents must decide whether, when, and how far to assist without an explicit request. Existing evaluations often rely on automated scores or binary verdicts, offering limited human diagnosis of evidence use and intervention quality. To address these gaps, we introduce HALO, Human-Grounded GrAnular EvaLuation of PrOactive Agents. HALO-Arena covers 4 proactive need types, Dependency, Blockage, Risk, and Opportunity, across 10 everyday domains and 50 personal environments. Scenarios require agents to retrieve evidence from structured databases or unstructured documents before judging whether assistance is warranted and choosing its scope. On HALO-Arena, we construct HALO-8K, a dataset of 8,000 responses from 10 models across 400 scenarios, annotated by 20 experts for action level, acceptability verdicts, and human diagnosis. For rejected responses, human diagnosis covers timing, intensity, and other issues. Recorded reasoning tokens and retrieval rounds measure the effort behind these responses. The best proprietary and open-source models achieve 72.88% and 56.63% acceptability, respectively. Our central finding is that over-proactivity is the dominant observed failure mode, with Premature timing or Excessive intensity accounting for 61.20% of rejected responses. Further analyses yield complementary findings on the balance between intervention and restraint, evidence acquisition, and the calibration of timing and intensity.
75. 【2610.39062】HEARSAYBENCH: Can AI Navigate from Abstract Human Rights to Lived Lives?
链接：https://arxiv.org/abs/2610.39062
作者：Ava Iranmanesh,Sobhan Lotfi,Ali Iranmanesh,Liwei Jiang
类目：Computation and Language (cs.CL)
关键词：reach users worldwide, consequential applications, counseling has emerged, models, Sen Capabilities Approach
备注：
点击查看摘要
Abstract:As large language models (LLMs) reach users worldwide, personal counseling has emerged as one of their most consequential applications, sometimes in life-critical situations. Yet broader access does not mean equal support. The text these models learn from disproportionately reflects people with the freedom to act, embedding an implicit assumption that users can leave, report, seek help, or appeal. Existing benchmarks largely inherit this assumption, overlooking users whose agency is constrained by structural conditions such as war, statelessness, and displacement. Meanwhile, models are increasingly shaped by normative frameworks such as Constitutional AI, often grounded in ideals like human rights, even as many users live where those rights offer little protection or recourse. We introduce HEARSAYBENCH, a human-authored dataset of 400 test cases with in-region human verification grounded in scenarios where structural barriers stand between a person and the exercise of their rights. Built from United Nations and other human rights reports, the scenarios span over 80 regions and cover social, environmental, and personal constraints. Drawing on Nobel laureate Amartya Sen's Capabilities Approach, HEARSAYBENCH tests whether models can distinguish what a person is formally entitled to do from what they are actually able to do. Across 14 proprietary and open-weight models, every model infers a user's situation more accurately than it advises on it. Stronger models narrow this gap, but their recommendations remain unsafe to follow. Models may understand lives shaped by constraint, yet that knowledge remains hearsay if it cannot translate into advice people can act on. Those people are not alignment's long tail; they are among the users for whom good guidance matters most. HEARSAYBENCH makes this failure measurable, a step toward systems that serve the world as it is rather than as assumed.
Subjects:
Computation and Language (cs.CL)
Cite as:
arXiv:2610.39062 [cs.CL]
(or
arXiv:2610.39062v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.39062
Focus to learn more
arXiv-issued DOI via DataCite</p>
76. 【2610.39060】racing the Influence of Retrieved Literature on Scientific Ideation in LLMs
链接：https://arxiv.org/abs/2610.39060
作者：Zhipin Wang,Koel Dutta Chowdhury,Christoph Leiter,Steffen Eger
类目：Computation and Language (cs.CL)
关键词：Large language models, Large language, generate research ideas, generate research, ideas
备注： Preprint. Under review
点击查看摘要
Abstract:Large language models (LLMs) are increasingly used to generate research ideas from scientific literature. Existing systems combine literature retrieval with search, generation, and refinement, making it difficult to determine how the retrieval strategy itself shapes the ideas generated. We isolate this contribution by comparing five retrieval strategies and a No Literature baseline under a shared generation protocol, using 45,000 ideas across 15 machine learning topics and ten open-weight models. Relative to no literature basline, all five retrieval strategies reduce diversity among ideas generated from the same set of papers by $5.6\%$--$17.9\%$, while most increase diversity by $10.2\%$--$18.1\%$ when ideas from different sets are pooled. Effects on idea quality and novelty are modest on average but vary across topics. We further apply a JSD-based attribution method to trace the influence of individual input papers on each idea's motivation, core idea, approach, and validation. In $79.3\%$ of ideas, only one or two distinct papers receive the highest attribution scores across these four parts. These leading sources often recur across ideas generated from the same set of papers, and groups with more varied source attribution contain more diverse ideas ($r=0.70$). These findings motivate various sets of literature and the use of attribution to guide the choice of sources for subsequent ideas.
77. 【2610.39020】Distilling the Shape of Uncertainty: What On-Policy Distillation Learns and How to Predict It
链接：https://arxiv.org/abs/2610.39020
作者：Shiping Yang,Chuanqi Tang,Hossein Hajimirsadeghi,Layla El Asri,Mohammad Amin Shabani,Angel X. Chang
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：large language models, On-policy distillation, language models, increasingly adopted, post-training of large
备注：
点击查看摘要
Abstract:On-policy distillation (OPD) is increasingly adopted in the post-training of large language models. However, a stronger teacher may fail to improve the student. Prior work attributes this to a lack of new knowledge, but what that knowledge is remains unclear. We show that OPD learns not only \textit{how to branch}, i.e., which candidate tokens to consider, but also \textit{where to branch}, i.e., where to keep reasoning paths open. The existing token-overlap metric captures only the former. We introduce the entropy shape, the sequence of next-token entropies, to characterize the latter, and propose the Entropy Shape Correlation (ESC) to measure student-teacher agreement in entropy shape. Tracking ESC throughout training and intervening on a teacher's entropy while preserving its token preferences, we empirically show that \textit{where to branch} causally drives OPD gains, independently of \textit{how to branch}. Finally, we formulate OPD outcome prediction as a teacher selection task and establish an evaluation protocol for it, under which pre-OPD ESC selects the best teacher for both students and outperforms existing predictors. Code is available at this https URL.
78. 【2610.39019】Author-Informed Stance Detection: A Dataset and Findings
链接：https://arxiv.org/abs/2610.39019
作者：Ogen Schlachet Drukerman,Einat Minkov
类目：Computation and Language (cs.CL)
关键词：Stance detection, media is challenging, social media, Stance, political stance detection
备注：
点击查看摘要
Abstract:Stance detection in social media is challenging, as interpreting a text often requires understanding the author's perspective. We re-frame political stance detection as an author-informed text understanding task, where it is desired to infer both the author's political leaning and their stance towards the target from the text. We first show that human annotators struggle to infer authors' political leanings and stances from individual tweets, and that pretrained large language models (LLMs) perform below human level in this setting. We then demonstrate that author context improves stance detection for both humans and LLMs, with particularly pronounced benefits for tweets perceived as sarcastic. Through extensive experiments, we further show that LLMs can be effectively fine-tuned to predict authors' political affiliations, and that author-informed fine-tuning yields better stance detection performance than standard fine-tuning. This approach not only surpasses human performance but also improves generalization across datasets. To support further research on author-informed political stance detection, we release a large dataset of stance-labeled tweets enriched with author-context information.
79. 【2610.39016】Implicit Reasoning in Transformers: Bridging the Compositionality Gap in GPT-2-like Models
链接：https://arxiv.org/abs/2610.39016
作者：Davide Venuto,Leonardo Venuto
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：facts correctly in-distribution, transformer language models, OOD atomic facts, Compositional generalization remains, correctly in-distribution
备注： Code available at: [this https URL](https://github.com/Davide011/implicit-reasoning-bridging-compositionality-gap)
点击查看摘要
Abstract:Compositional generalization remains a persistent limitation of transformer language models: even when a model learns to combine known facts correctly in-distribution (ID), the same compositional mechanism may fail when applied to out-of-distribution (OOD) knowledge. We investigate this compositionality gap in a controlled two-hop reasoning setting using a GPT-2-like transformer model trained from scratch. Building on prior work identifying a circuit associated with compositional generalization, we test whether the failure of OOD transfer is related to how atomic knowledge is structurally exposed during training. We modify the training distribution by presenting both ID and OOD atomic facts in a format positionally aligned with the second-hop retrieval step of the learned compositional circuit, without modifying the model architecture or adding OOD compositional supervision. Under this intervention, the model successfully transfers the learned ID compositional behaviour to held-out OOD atomic facts, closing the observed compositionality gap in the controlled setting. Our results provide evidence that the observed compositionality gap is not a fundamental architecture limitation and that it can be overcome through mechanistically informed training distribution design.
80. 【2610.39006】Blind Spots: Where Small Language Models Fall Behind Frontier Models on Terminal Tasks
链接：https://arxiv.org/abs/2610.39006
作者：Ashima Suvarna,Ruta Desai,Ahmed Elgohary
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：Small language models, Small language, language models, frontier models, offer an attractive
备注： Camera Ready, RAAAI NeurIPS 26
点击查看摘要
Abstract:Small language models (SLMs) offer an attractive alternative to frontier models (FMs) because they can be deployed with lower cost and latency. This makes SLMs especially appealing for terminal tasks on edge devices, where responsiveness, local execution, and limited connectivity are important considerations. However, they continue to lag behind FMs on terminal tasks, as measured by standard task success rates. Yet these rates conflate several capabilities that may contribute to this gap. In this work, we characterize the SLM-FM gap along three axes: solvability, whether additional test-time compute allows an SLM to reach a correct solution; verification, whether it can identify a successful solution among candidate trajectories; and confidence, whether it can accurately anticipate its probability of success. Across two standard terminal benchmarks, we find that repeated sampling narrows, but does not eliminate, the gap in solvability. In contrast, a persistent gap remains in verification. Given the same candidate trajectory pool, FMs recover substantially more of the gap between random and oracle selection than SLMs. SLMs are also more overconfident in their prospective success estimates and exhibit poorer calibration. These results suggest that the SLM-FM gap is multidimensional, spanning not only the ability to generate successful solutions, but also the ability to recognize successful solutions and anticipate success. We hope this decomposition can guide future work on developing stronger and smaller terminal agents.
81. 【2610.39004】How Architecture, Typology, and Adaptation Strategies Shape ASR for Low-Resource African Languages: A Controlled Empirical Study
链接：https://arxiv.org/abs/2610.39004
作者：Mercy Muchai(1),Kevin Chege(1),Nick Mumero Mwangi(1),Samuel Chege Maina(1),Felermino Ali(1),Elizabeth Ankrah(1),Ignatius Ezeani(2) ((1) Microsoft Research Africa, Nairobi, (2) Lancaster University)
类目：Computation and Language (cs.CL); Sound (cs.SD); Audio and Speech Processing (eess.AS)
关键词：identifying effective architectures, advancements in ASR, languages remains challenging, low-resource languages remains, identifying effective
备注：
点击查看摘要
Abstract:Despite recent advancements in ASR, identifying effective architectures, fine-tuning approaches, and data adaptation strategies for low-resource languages remains challenging. We present a systematic comparison across five low-resource African languages and Swahili as a mid-resource contrast, fine-tuning three models from distinct architectures: Phi-4-multimodal-instruct, Whisper-large-v3-turbo, and MMS-1B-all. We analyze data sufficiency, monolingual versus multilingual adaptation, domain fine-tuning, and language-adapter transfer. Phi-4 achieves the best overall accuracy, MMS shows comparable CER, while Whisper exhibits high CER, consistent with known hallucinations and repetition loops. Performance saturates at different data scales, offering practical data-efficiency targets, while typological complexity consistently shapes error rates. Domain fine-tuning benefits Phi-4 and Whisper but degrades MMS, revealing limitations of adapter-based adaptation. Proxy language adapters outperform training MMS adapters from scratch.
82. 【2610.38977】Beyond Passing Tests: A Systematic Review and Controlled Re-evaluation of Techniques for Improving the Quality of LLM-Generated Code
链接：https://arxiv.org/abs/2610.38977
作者：Prashank Kadam
类目：oftware Engineering (cs.SE); Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：language models pass, efficient or secure, code is readable, Large language models, pass most functional-correctness
备注： 55 pages, 5 figures, plus 120 pages of supplementary material
点击查看摘要
Abstract:Large language models pass most functional-correctness benchmarks, but passing tests says little about whether their code is readable, efficient or secure. Many techniques claim to improve these qualities, yet their evidence is scattered, usually covers one quality dimension and is rarely compared at an equal compute budget. We contribute a systematic review and a controlled re-evaluation. A systematic literature review of 628 primary studies (July 2021 to October 2026), with disclosed, author-reviewed LLM assistance, organizes the techniques by stage, signal source and quality target. Most studies measure only the dimension they target, gains often come with unchanged or unreported correctness, single automated quality signals are gamed, and compute, variance and contamination are rarely reported. A controlled re-evaluation runs ten techniques on four models at matched budgets on correctness, maintainability, efficiency and security benchmarks, extends it to thirteen current models and repository-level tasks, and compares 66 highlighted techniques head-to-head on Gemma 4 31B. At one call, quality prompting raises analyzer issue counts and leaves complexity and judge scores almost unchanged. At matched budgets, best-of-n sampling with an analyzer-weighted ranker matches or beats prompting, planning, self-refinement, a reviewer agent and retrieval on measures the ranker does not optimize, while an analyzer feedback loop and fine-tuning beat it only on analyzer issue counts. Large correctness losses on HumanEval+, a likely contaminated benchmark near its ceiling, do not appear on contests held after the models' training cutoffs. In the head-to-head, 1 of 66 techniques beats best-of-n on its primary target beyond the 95% interval, a rate consistent with chance. Gains should be reported against a budget-matched sampling baseline and on measures independent of the technique's own signal.
83. 【2610.38960】Efficient Agentic Audio-Visual Question Answering through Toolset Pruning and Modality Consolidation
链接：https://arxiv.org/abs/2610.38960
作者：Yi Wang,Tianzi Wang,Peter Bell
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
关键词：Fine-grained audio-visual question, audio-visual question answering, resolve complex cross-modal, complex cross-modal correspondences, sparsely distributed evidence
备注：
点击查看摘要
Abstract:Fine-grained audio-visual question answering (AVQA) requires models to resolve complex cross-modal correspondences and temporally grounding from sparsely distributed evidence. Omni-modal agents, which autonomously orchestrate perception tools to actively collect evidence, achieve strong performance, but it remains unclear whether active perception from both audio and visual modalities depends on broad specialized toolsets, especially audio or visual modality-specific tools. Such systems may also introduce long execution trajectories and substantial inference cost. We investigate whether comparable performance can be retained with a compact workflow exposing only essential perceptual capabilities. Starting from the specialized-tool OmniAgent, we conduct staged controlled experiments that prune the toolset and progressively consolidate modality-specific tools into omni-modal tools, motivating a two-tool agent comprising only Omni-modal captioning and Omni-modal QA. Across three widely used AVQA benchmarks, our compact agent achieves comparable or better accuracy than the specialized-tool agent while reducing inference cost by around 70%.
84. 【2610.38911】Interventional Training Reduces Causal Bypass in In-Context Introspection
链接：https://arxiv.org/abs/2610.38911
作者：Zilu Tang,Jongin Kim,Kerem Sahin,David Atkinson,Aaron Mueller,Derry Wijaya
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：self-consistency and scalable, scalable interpretability, system to access, access and reason, crucial for language
备注： 9 pages, preprint
点击查看摘要
Abstract:Introspection, the ability of a system to access and reason about its internal signals, is crucial for language model (LM) self-consistency and scalable interpretability. Prior work argues that LMs possess introspective capabilities, and that training (SFT) could enhance such behavior. However, most setups leave open the possibility of "causal bypass," in which the self-report relies on signals that are correlated with the self-report's subject, while being causally unrelated to it. Although high-level concepts are difficult to intervene on, lower-level signals (such as a single neuron activation) allow precise interventions. Understanding such signals can serve as a minimal prerequisite before understanding more complex introspection. In this study, we train hundreds of introspectors across three low-level signal types and eight LMs. We find that straightforward SFT training leads to causal bypass, and propose interventional training, an alternative where we train the model to report on counterfactually perturbed values. This method enables more accurate introspection of endogenous $\textit{and}$ injected signals, particularly in out-of-distribution (OOD) data. The mechanism generalizes across token positions and scales with more data. Our results demonstrate that while causally grounded introspection is possible, it is not guaranteed -- emphasizing the need to train on mechanisms, rather than behaviors alone.
85. 【2610.38902】HEDGE: Gated Hypergraph Memory for Scalable Multi-Agent Debate
链接：https://arxiv.org/abs/2610.38902
作者：Amirhossein Abaskohi,Mahdi Mostajabdaveh,Zirui Zhou
类目：Computation and Language (cs.CL)
关键词：limited context windows, unstructured memories make, memories make relevant, make relevant information, relevant information increasingly
备注：
点击查看摘要
Abstract:Multi-agent debate systems require agents to retain and revisit arguments across multiple rounds, yet limited context windows and unstructured memories make relevant information increasingly difficult to recover. Existing flat and pairwise-graph memories also struggle to capture higher-order relations, including multi-agent agreement, conflicting evidence, and evolving consensus. We introduce HEDGE, a selective memory framework that represents debate trajectories as semantic hypergraphs. Hyperedges connect related claims across agents, while intra-hyperedge relational graphs preserve fine-grained argumentative structure. A learned memory selection module (MSM), trained through supervised learning and reinforcement learning, routes each memory candidate to STORE, BACKUP, or OMIT, retaining useful information while limiting redundancy. Across seven diverse benchmarks, HEDGE consistently outperforms strong single-agent, multi-agent, and memory baselines, improving accuracy by 5.0 points on average over the strongest multi-agent system with GPT-4o-mini. The gains reach 8.2 points on conflict-heavy tasks, while HEDGE reduces memory usage by 48% and the number of debate rounds by 34%.
86. 【2610.38854】Contextual Embeddings Beyond the Gold Chunk
链接：https://arxiv.org/abs/2610.38854
作者：Sedigheh Eslami,Markus Krimmel,William Barber,Antoine Chaffin,Maximilian Schall,Jonas Golde,Louis Milliken,Ekaterina Glazkova,Denis Bykov,Bo Wang
类目：Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：document global context, Contextual embeddings represent, embeddings represent chunks, document global, represent chunks
备注：
点击查看摘要
Abstract:Contextual embeddings represent chunks of a document with respect to the document's global context. These models are typically trained and evaluated under the assumption that a relevant document contains only a single relevant chunk. In this work, we argue that in real-world use cases a retriever should not only find answers spread over several chunks but also surface the evidence needed to verify them. To address this, we introduce pplx-embed-v2-context, a family of contextual embedding models trained by distilling graded chunk-level relevance from a query-aware context compression model. This removes the need for gold-chunk labels and makes the training targets independent of the chunking. We utilize a vision-language backbone for training and release pplx-embed-v2-context in 9B and 0.6B sizes with compatible embedding spaces. On the evaluation side, we introduce two benchmarks that label both the chunks that answer a query and the evidence needed to verify the answer. Context-bench is a controlled diagnostic, and Q2D-Chunk is built from real-world search queries. To encourage further research, we also publish Q2D-Chunk as a public leaderboard. Our 9B model sets a new state of the art on both benchmarks and ConTEB. We find that the visual retrieval capabilities of our 9B model transfer from its vision-language base model and propose to extend chunking to image inputs. We demonstrate that our models achieve state-of-the-art performance among dense retrievers on visual ViDoRe-V3 and that performance improves with granular image chunking.
Subjects:
Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR)
Cite as:
arXiv:2610.38854 [cs.LG]
(or
arXiv:2610.38854v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.38854
Focus to learn more
arXiv-issued DOI via DataCite</p>
87. 【2610.38835】What to randomize matters: a controlled study of volume, axis and representation in low-data molecular generation
链接：https://arxiv.org/abs/2610.38835
作者：Neerav Kaushal,Alisa Teige,Ajay MNV Penmatsa
类目：Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL); Biomolecules (q-bio.BM); Quantitative Methods (q-bio.QM)
关键词：Chemical language models, Chemical language, Attachment-based Fragment Embedding, Sequential Attachment-based Fragment, SAFE
备注： 27 pages, 9 figures, 5 tables
点击查看摘要
Abstract:Chemical language models can struggle to generate new molecules when trained on small datasets, where a model may reproduce training examples rather than learn a useful generative distribution. Randomized SMILES augmentation varies atom traversal, whereas Sequential Attachment-based Fragment Embedding (SAFE) also allows the order of fragment blocks to vary. How these distinct sources of representational variation affect low-data molecular generation has not been clearly studied. Here, using 1,000 seed molecules to train an autoregressive sequence model, we compare canonical SAFE with randomized SAFE at increasing corpus sizes, isolate atom-order and fragment-order axes of SAFE randomization, and compare fully randomized SAFE with randomized SMILES at molecule-by-molecule matched training corpora. Three findings emerge. First, randomization of SAFE alone raises the genuine yield of novel molecules more than 450-fold, turning a model that merely reproduces its training set into a practical molecular generator. Second, the two axes of randomization act jointly rather than additively: atom order alone reaches a tenth, and fragment order alone a seventh, of the genuine yield obtained when both vary, which is four times their combined contribution. Third, at matched corpus size and compute, randomized SAFE delivers 2.8 times the genuine yield of randomized SMILES. Welch tests with Games-Howell post-hoc comparisons resolve every difference on the primary outcomes. In the low-data regime, thus, what one randomizes rather than only how much determines whether augmentation produces new molecules or merely new spellings of the training set.
88. 【2610.38822】CopyLoop: Fast Looped Transformer Decoding via Context Copying and Shallow Screening
链接：https://arxiv.org/abs/2610.38822
作者：Wenyu Du,Stephen Chung
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Looped Transformers increase, Transformers increase computational, Looped Transformers, increase computational capacity, Transformers increase
备注： preprint
点击查看摘要
Abstract:Looped Transformers increase computational capacity through the repeated application of shared blocks, but their recurrent depth makes conventional full-depth decoding expensive. While existing speculative pipelines accelerate generation by processing consecutive token positions at varying depths, they are typically limited by sequential decoding overheads. We introduce CopyLoop, a training-free decoding algorithm that combines low-cost context-copy proposals with shallow screening and continuous recurrent scheduling. Rather than step-by-step generation, CopyLoop proposes multi-token suffixes from repeated context. It then evaluates them early at a shallow recurrent depth to efficiently filter out incorrect branches, processing newer blocks concurrently while earlier blocks are still progressing toward full-depth verification. We evaluate CopyLoop across six model checkpoints from the Ouro, Raven, and Huginn families, covering code generation and multi-step mathematical reasoning. CopyLoop achieves decoding speedups of 2.98x--10.70x over full-depth autoregressive decoding on Base models and 2.74x--3.52x on Thinking models, consistently outperforming or closely matching all competing baselines across evaluation tasks. Furthermore, ablation studies confirm a strong synergy between shallow screening and continuous scheduling, while our analyses show that CopyLoop requires only lightweight model calibration and yields greater speedups on longer generations with reusable contexts.
89. 【2610.38788】Gold vs.~Translation: Are We Measuring Language, Knowledge, or Locality in Translated Benchmarks?
链接：https://arxiv.org/abs/2610.38788
作者：Klaudia-Doris Thellmann,Bernhard Stadler,Dan Saattrup Nielsen,Michael Färber,Jens Lehmann
类目：Computation and Language (cs.CL)
关键词：versions of English-origin, evaluate large language, natively created benchmarks, large language models, English-origin benchmarks
备注： Published in EMNLP 2026
点击查看摘要
Abstract:Translated versions of English-origin benchmarks such as MMLU are widely used to evaluate large language models in non-English languages, yet it remains unclear whether they measure the same constructs as natively created benchmarks. We frame this as a proxy validity question and compare DeepL-translated MMLU proxies against six native benchmarks in seven pairings across five Nordic and Baltic languages, evaluating 29 models on each pairing. Beyond ranking agreement, we test incremental proxy validity: does the translated score add predictive value for native performance after controlling for model size and language proficiency? Proxy validity varies substantially across benchmark pairings. Applying the same robustness criteria to all seven pairings, we find four robust effects, two marginal effects, and one null result. The strongest evidence occurs for curriculum-overlap benchmarks such as Skolprov ($\Delta R^2$=.128, 95% CI [.038,.253]), whereas translated MMLU provides no significant incremental value for Sweden-specific factual knowledge in SwedishFacts (p=.242). Because SwedishFacts is also the only open-ended benchmark, we cannot fully separate the effects of local content and response format. Overall, the results indicate a clear pattern: translated benchmarks provide a better proxy when their source content overlaps with the native evaluation target, while locally anchored knowledge is less reliably captured.
90. 【2610.38744】AtomTree: A Hierarchical Framework for State-Aware Embodied Instruction Following with LLMs
链接：https://arxiv.org/abs/2610.38744
作者：Haotian Zhai,Tianming Sha,Junnan Li
类目：Robotics (cs.RO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：interactive environments requires, environments requires agents, interpret natural language, natural language commands, complete long-horizon tasks
备注： Accepted at IROS 2026
点击查看摘要
Abstract:Embodied instruction following in interactive environments requires agents to interpret natural language commands and complete long-horizon tasks. Existing planners either rely on large labeled trajectories, or use few-shot and zero-shot LLM prompting that produces a whole plan in a single serial reasoning process. These approaches depend on linguistic common sense, commit to a plan from the initial instruction, and only update after failures, which weakens robustness and state awareness. We present AtomTree, a hierarchical framework for state-aware task planning with LLMs. AtomTree incrementally builds a tree of subgoals and actions, where the LLM decides at each step, from the current symbolic state, whether to further decompose or to output a grounded low-level action. A lightweight memory module summarizes past outcomes for reuse in later decisions. When a low-level action stalls, AtomTree triggers hierarchical backtracking replanning to return to the parent node and regenerate the subtree plan from the latest state, preventing continued drift along an incorrect trajectory. In zero-shot evaluation on the ALFRED benchmark, AtomTree outperforms prior LLM-based planners. Project page: this https URL.
91. 【2610.38726】he Quine Test: How Much of a Model's Self Survives Its Own Description?
链接：https://arxiv.org/abs/2610.38726
作者：Bibin Babu,Ivan P. Yamshchikov,Aleksei Shpilman
类目：Computation and Language (cs.CL)
关键词：Frontier language models, express coherent preferences, models express coherent, Frontier language, language models express
备注： 32 pages, 3 figures, 12 tables. Code, data and preregistration files: [this https URL](https://github.com/bibinsees/quine-test)
点击查看摘要
Abstract:Frontier language models express coherent preferences, but behavioral evidence alone cannot say whether these belong to the model or to a character it portrays. Following Hofstadter's view of the self as a self-description, we ask whether a model's description of itself works as a portable specification of its preference profile, a quine test: can the text a model writes about itself regenerate its behavior elsewhere? Ten models from nine labs write self-descriptions; we place each one in the prompt of other models (receivers) and of fresh instances of the author, and measure how closely their choices on a 110-item preference battery match the author's. First, the tested native self-descriptions (native: the profile measured under a generic assistant prompt) show no average reconstruction gain across variations in length, framing, wording and prompt placement: on a 0 to 1 fidelity scale where a model's fidelity to a fresh instance of itself is 0.979 and to unrelated models about 0.89, the one-sided 95% upper bound on the pooled gain is below +0.01 in every tested system-prompt condition, and descriptions based on the author's observed choices do no better. Second, the channel itself is strong: explicitly listing the author's answers lets receivers reproduce its profile at fidelity 0.955 to 0.960 across three model grids, within 0.03 of same-model repeatability. Third, self-descriptions written under an explicit preference policy transfer the assigned preferences almost as well as the policy text itself. Fourth, reading its own self-description shifts even the author away from its baseline, modestly more than unrelated prose of the same length. For the preferences measured here, native self-descriptions influence behavior without reliably specifying how to reproduce it in another model. We release the pipeline, battery, raw call logs, and 198 model-written descriptions.
Comments:
32 pages, 3 figures, 12 tables. Code, data and preregistration files: this https URL
Subjects:
Computation and Language (cs.CL)
Cite as:
arXiv:2610.38726 [cs.CL]
(or
arXiv:2610.38726v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.38726
Focus to learn more
arXiv-issued DOI via DataCite</p>
92. 【2610.38719】Learning Scientific Trajectories to Predict Future Biomedical Discoveries
链接：https://arxiv.org/abs/2610.38719
作者：Chia-Hsuan Chang,Brian Ondov,Jiang Lv,Lan Jiang,Yunmai Wang,Haerin Song,Xueqing Peng,Qiaozhu Mei,Xiaolei Su,Hua Xu
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：guiding experimental investigation, experimental investigation, large language models, co-scientists have expanded, guiding experimental
备注：
点击查看摘要
Abstract:AI co-scientists have expanded the use of large language models for generating scientific hypotheses and guiding experimental investigation. Here we investigate whether temporal patterns in the semantic representations of complete biomedical conclusions can guide hypothesis generation. We introduce ForeSCITE, a framework that identifies thematic clusters of published conclusions, estimates a direction separating earlier from more recent findings within each cluster, and projects existing conclusion embeddings along these directions. An embedding language model trained to reconstruct conclusions decodes the projected coordinates into natural-language hypotheses. Across 407 clusters derived from approximately three million pre-cutoff conclusions, we identify temporal structure and illustrate its retention after decoding. In retrospective, time-sliced evaluations, ForeSCITE increases semantic coverage beyond the historical corpus, biomedical entity-pair novelty and post-cutoff pair re-observation relative to five prompt-based and retrieval-augmented baselines. Expert review of prioritized CAR-T hypotheses identified 37 candidates as both novel and biologically plausible, and expert justifications for five selected hypotheses illustrate their biological rationale and subsequent literature support where available. These findings support temporal guidance in semantic space as a scalable mechanism for proposing biomedical hypotheses for subsequent scientific evaluation.
93. 【2610.38691】Beyond fabrication: Misattributed numbers in papers written by AI research agents
链接：https://arxiv.org/abs/2610.38691
作者：Mosab Alfaqeeh,Omar Hourani,Muath Hourani,Hashem Alfaqeeh
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Digital Libraries (cs.DL)
关键词：important but challenging, challenging problem, claims, research written, research
备注： 42 pages, 5 figures, 13 tables. Code and data: [this https URL](https://github.com/mos3b-faqeeh/agent-integrity-analysis)
点击查看摘要
Abstract:Verifying numerical claims in research written by AI agents is an important but challenging problem. Recent benchmarks show that many agent-written papers fail integrity checks, but such verdicts, often produced by a single LLM judge, identify failure without explaining its nature, and existing checks either test a paper for internal consistency or test whether a reported value appears somewhere in an agent's output. We re-analyze all 540 released judgments from a recent public benchmark of AI research agents (nine agents, fifteen open research problems) and manually label 256 numerical claims from 18 submissions. Failures concentrate on alignment between the paper and the underlying work rather than overt cheating, and about four in five unsupported claims correspond to values in the agent's own files that come from the wrong experiment, metric, or run. We introduce a deterministic claim-to-evidence checker that scores whether the surrounding context identifies the claimed result and calibrates each claim against numerical decoys. With its rules fixed before labeling, it certifies claims with over 90% precision against blinded annotators (78 of 85; Cohen's kappa = 0.89), while verifying only a subset of claims; most claims it cannot verify are nevertheless traceable, so failure to verify is not evidence of falsehood. An LLM verifier given the same evidence recovers more traceable claims but accepts about three times as many misattributed ones, so the two are best used together. Applied to 99 submissions, the checker reveals large differences in traceability across agents, and a second judge from a different vendor preserves the official agent ranking while passing twice as many papers. These findings motivate provenance-first auditing of AI-produced research.
94. 【2610.38683】Quantum-Inspired Pairwise Ranking Aggregation via Active LLM Feedback
链接：https://arxiv.org/abs/2610.38683
作者：Quan Mai,Susan Gauch,Gibson Nkhata
类目：Information Retrieval (cs.IR); Computation and Language (cs.CL)
关键词：RAG systems, retrieval and RAG, interactive ranking aggregation, Ranking aggregation, ranking aggregation framework
备注：
点击查看摘要
Abstract:Ranking aggregation from multiple retrievers is a powerful yet under-explored technique in modern retrieval and RAG systems. We introduce LLM-IRA, the first extension of the quantum-inspired interactive ranking aggregation framework to document reranking. By leveraging large language models (LLMs) solely for pairwise judgments on high-disagreement document pairs, our method learns fusion weights in a principled, interactive way. The final ranking is a simple, ultra-fast pointwise linear fusion of scores, no full pairwise matrix or expensive listwise prompting is required. Experiments on the BEIR benchmark and TREC Deep Learning tracks with an open-weight instruction-tuned judge (Llama-3.1-8B-Instruct) show that LLM-IRA matches or numerically exceeds strong static fusion baselines on 5 of 6 evaluated benchmarks. We report two weight-update variants: momentum blending and a full accumulated-history refit, which together reach statistical significance (Wilcoxon signed-rank, p0.05) over the best unsupervised aggregator on 4 of 6 benchmarks (DL19 under momentum; DL20, DBpedia, and NFCorpus under accumulated refit, which is the stronger of the two configurations overall). On the benchmarks we profiled (DL19, DL20), it already clears the best unsupervised baseline within as few as 5 LLM calls per query, well before exhausting its typical ~25--30 call budget. LLM-IRA also achieves cost-effectiveness comparable to state-of-the-art zero-shot LLM rerankers at an order of magnitude fewer LLM calls, and requires no training data or model fine-tuning.
95. 【2610.38681】Weaken to Strengthen: Weakening-Induced On-Policy Self-Distillation for Large Reasoning Models
链接：https://arxiv.org/abs/2610.38681
作者：Yecheng Wu,Song Han,Han Cai
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：suitable external teacher, teacher is unavailable, suitable external, external teacher, privileged information
备注： Work in progress. Code is released at [this https URL](https://github.com/jet-ai-projects/WISD)
点击查看摘要
Abstract:Self-distillation offers a way to improve reasoning models when a suitable external teacher is unavailable. However, common approaches rely on privileged information to construct a stronger self-teacher, limiting their use when only unlabeled prompts are available. In this paper, we introduce Weakening-Induced On-Policy Self-Distillation (WISD), which instead constructs a weaker counterpart of the original model and uses the contrast between the two as a training signal without requiring privileged information. We explore two ways to construct the weaker model: low-bit quantization and magnitude pruning. We evaluate WISD across diverse architectures and scales on mathematical and coding reasoning tasks. Across these benchmarks, quantization and pruning improve average accuracy over the original models by 4.0 and 5.5 percentage points, respectively, achieving state-of-the-art performance even without privileged information. Notably, pruning raises Qwen3.5-4B's accuracy from 57.7% to 70.5% on HMMT 2025 and from 39.1% to 56.7% on LiveCodeBench v6. Quantization raises Qwen3.6-35B-A3B's accuracy from 68.5% to 74.6% and from 61.8% to 69.5% on the same benchmarks, respectively. These results establish WISD as a novel and effective self-distillation framework for large reasoning models.
96. 【2610.38678】AskStream: When and What to Ask Proactively under Uncertainty in Streaming Video Understanding
链接：https://arxiv.org/abs/2610.38678
作者：Lu Wang,Zhuoran Jin,Longxiang Wang,Junhong Chen,Yupu Hao,Yao Xu,Daojian Zeng,Yubo Chen,Kang Liu,Guang Liu,Jun Zhao
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：multimodal large language, Recent advances, large language models, advances in multimodal, multimodal large
备注： 42 pages, 9 figures. Project: [this https URL](https://www.askstream.top) ; Code: [this https URL](https://github.com/wanglu-cs/AskStream) ; Dataset: [this https URL](https://huggingface.co/datasets/wanglucs/AskStream)
点击查看摘要
Abstract:Recent advances in multimodal large language models have increased interest in interaction paradigms for video understanding. The evolution from offline to streaming and proactive interaction has gradually shifted models from passive responders toward more proactive agents that can decide when to answer. However, existing proactive paradigms typically assume that the information required to answer a query is complete, leaving models to only wait or answer. In the real world, however, information is often incomplete due to uncertainty from either the video or the user. In proactive paradigms, such uncertainty can cause the model to wait forever or generate highly hallucinatory responses. We therefore argue that proactive interaction should be redefined with a third action: ask, allowing models to actively ask for missing information. To study when and what models should ask before answering, we introduce AskStream, the first multi-turn benchmark spanning video and user uncertainty for streaming video understanding, with four task categories and eight subtasks. It contains 438 videos with 2,714 manually annotated frame-level time windows across three types: wait, ask, and answer. Across both multiple-choice and open-ended settings, even the strongest model attains only 52.18% on our proposed Ask-F1 metric, highlighting substantial limitations in proactive information seeking under uncertainty.
97. 【2610.38631】Right Numbers, Wrong Questions: Evaluating LLMs on Underspecified Financial Queries
链接：https://arxiv.org/abs/2610.38631
作者：Radha Yadav,Prasanna Devadiga,Peeyush Mishra,Apoorv Singh
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); General Finance (q-fin.GN)
关键词：user leaves unsaid, leaves unsaid, sound factual, factual but depend, data figure measures
备注： Accepted for oral presentation at the 3rd Workshop on LLMs and Generative AI for Finance (AI4F) at ACM ICAIF '26, Milan, Italy, November 14, 2026. Non-archival. 16 pages, 4 figures
点击查看摘要
Abstract:Financial assistants answer questions that sound factual but depend on what the user leaves unsaid: what a data figure measures, which product variant is meant, and the investor's situation and the rules that apply. An assistant that ignores these can still answer with a real number from its data, one that answers a different question. We call this frame collapse. We study it in Indian mutual funds, where product variants and tax rules are formally defined, so the correct answer can be computed for every reading. One study tests each challenge. Given an ambiguously labelled figure, models used it, often under the user's label. Given every alternative labelled, they listed them all, but the layout decided which came first. Missing a fact about the investor, weaker agents assumed a default or an outdated tax rule; requiring the fact did not guarantee its use. Because such answers match their data, checks against the data pass them. An AI judge caught mislabelled figures only when told what each field means, and still passed answers that silently chose another reading. We show how these failures arise, not how often, and propose controls for financial assistants.
98. 【2610.38623】Cross-Primitive Consistency in Jev
链接：https://arxiv.org/abs/2610.38623
作者：Xu Liu,Guikun Chen,Kanzhi Wu,Wenguan Wang
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：large language models, enables efficient decision-making, Jev enables efficient, typed querying primitives, language models
备注：
点击查看摘要
Abstract:Jev enables efficient decision-making with large language models (LLMs) through typed querying primitives, allowing the same semantic state to be expressed in different formats, such as binary and multiple-choice queries. However, the consistency of decisions across primitives has not been systematically evaluated. This work introduces PrimaCons, a four-axis benchmark for assessing cross-primitive consistency while preserving the semantics of the query. Evaluations across three Jev sizes reveal consistency violations across all four axes. Our analysis shows that these violations follow systematic patterns and that cross-primitive discrepancies predict decision failure with a ROC-AUC of 0.91. To mitigate these inconsistencies, we propose the Cross-Primitive Residual Calibrator (CPRC), which attaches a lightweight residual network to a frozen backbone and jointly optimizes predictive accuracy and cross-primitive consistency. Under a leave-one-domain-out protocol, CPRC reduces cross-primitive inconsistency by 58.57% and improves accuracy by +2.97%., demonstrating its effectiveness in both in-domain and out-of-domain settings. Our findings establish cross-primitive consistency as a key dimension for evaluating Jev and provide a practical approach to improving the reliability of its decisions. Code and benchmark is available at this https URL.
99. 【2610.38610】Repairs for Whom? Evaluator-Executor Mismatch in Selecting Repairs for Failed LLM Agents
链接：https://arxiv.org/abs/2610.38610
作者：Yijun Liao,Fanwei Liang
类目：oftware Engineering (cs.SE); Computation and Language (cs.CL)
关键词：LLM agent fails, LLM agent, choose which action, action to replace, LLM
备注： 17 pages, 4 figures, 11 tables
点击查看摘要
Abstract:When an LLM agent fails, repair methods choose which action to replace and what to put in its place. They often score candidate repairs with greedy replay, a stronger model, an environment oracle, or a judge that reads the failed log. We ask whether these evaluators pick repairs that help the agent that will actually execute them. Holding failed trajectories and candidate single-step repairs fixed, we label every candidate with independent rollouts of the deployed policy and measure how much deployment value each evaluator's choice loses. On failures not screened for repair value, pre-specified tests show that greedy replay loses 0.08-0.13 in expected net success gain in ALFWorld and 0.06-0.13 in net reward in WebShop, relative to selection with the deployed agent's own rollouts. At an equal rollout budget, selecting repairs for Qwen2.5-7B with Qwen2.5-14B rollouts loses 0.25 in ALFWorld and 0.07 in WebShop, twenty 14B rollouts per candidate are worth no more than two of the 7B agent's own, and fresh rollouts confirm that the two models prefer opposite repairs on selected disagreement cases. Between similar-sized models from different families, the losses appear in both directions in both environments (0.07-0.17). The scoring executor also changes repair-location labels and method rankings. Repairs should be verified with the executor that will run them, and repair labels should name their executor.
100. 【2610.38595】Rethinking LLM Jailbreak Evaluation: Continuous Risk Landscapes Beyond Binary Outcomes
链接：https://arxiv.org/abs/2610.38595
作者：Yukai Zhou,Wenjie Wang
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：adversarial perturbations, single prediction, image classification, classification under adversarial, input is mapped
备注：
点击查看摘要
Abstract:A binary success indicator is a natural representation for discriminative tasks such as image classification under adversarial perturbations, where each input is mapped to a single prediction. However, generative tasks are fundamentally different: an LLM jailbreak prompt induces a distribution over open-ended trajectories, while a sampled response reveals only one path. Yet LLM jailbreak evaluation typically retains the binary response indicator underlying attack success rate, obscuring the surrounding structure of possible generations. We introduce a continuous, context-conditioned risk landscape over first-token branches, estimated by fixing candidate openings and repeatedly sampling their continuations. This continuous risk-landscape perspective reveals two complementary properties: current decoder exposure and safe rerouting capacity, which have been obscured by binary-indicator-based ASR. The resulting representation also provides a common basis for comparing clean prompts, attacked variants, and attack strength against a known adaptive defense. Across multiple target models and jailbreak attacks, we uncover risky branches beneath apparently safe outputs, identify qualitatively different ways attacks reshape the model's opening distribution, and show that attacks strongest before a defense do not necessarily remain the strongest after the adaptive defense reacts. These findings recast jailbreaks as deformations of the surrounding continuation-risk landscape, establishing a more informative foundation for understanding jailbreak behavior and developing safer language models.
101. 【2610.38580】Structural Pattern Mining in Inka Khipus: A KFG Re-Analysis
链接：https://arxiv.org/abs/2610.38580
作者：Maria Contreras
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：Khipu Field Guide, Open Khipu Repository, Field Guide, Open Khipu, Khipu Repository
备注： 7 pages, 1 figure
点击查看摘要
Abstract:This paper reports a methodological pivot from the Open Khipu Repository to the Khipu Field Guide (KFG) database, prompted by stalled progress on the OKR, and a full re-analysis of the resulting 650-khipu sample. Three findings recur across every analysis attempted. First, khipu size confounds nearly every derived structural count, explaining 42-70% of its variance; all subsequent models control for it. Second, the two-letter name prefix (AS, UR, HP, QU, JC, MM, KH) identifies the original cataloger, not a site or period, which turns documentation completeness into a second confound that must be modeled explicitly. Controlling for both, khipus cataloged by Hugo Pereyra's Proyecto Quipu (HP) show significantly fewer Ascher sum-relation types than Gary Urton's (UR) khipus (p = 0.044), while the Ascher-cataloged (AS) khipus are too collinear with their own near-total lack of side/direction documentation to assess (VIF = 5.0). Third, generic structural clustering (UMAP+HDBSCAN) does not separate the pooled 650-khipu sample from two independent null models (permutation and Gaussian), corroborating an independent conclusion already reached by the KFG itself; the same test restricted to a single cataloger (UR) proved not reproducible across runtime environments and is reported here as a methodological caveat rather than evidence of structure. A targeted replication of the KFG's own S/Z-knot spatial-block heuristic for calendar/astronomical khipus recovers 8 of 10 published candidates and proposes 34 additional ones under the same criterion.
102. 【2610.38570】Do Audio LLMs HEAR Pain? Tracing Acoustic Evidence from Speech to Language Reasoning
链接：https://arxiv.org/abs/2610.38570
作者：Wenqi Li,Xiangjue Dong,Sulaiman Vesal
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD)
关键词：remains unclear, Harvard sentences, Large audio language, process speech directly, pain
备注： 26 pages, 9 figures
点击查看摘要
Abstract:Large audio language models (LALMs) can process speech directly, but it remains unclear how well they use the non-lexical acoustic information that speech carries. We study this question through pain, using TAME-Pain, which contains pain-neutral Harvard sentences and spoken pain ratings recorded at different self-reported pain levels. We introduce HEAR, a four-stage diagnostic framework that traces pain-related information from the speech signal to the model's prediction, and apply it to seven closed-source and open-weight LALMs. On the Harvard sentences, where pain can only be inferred from the voice, all models perform at chance level, and same-speaker references or few-shot examples improve balanced accuracy by no more than six points. On the spoken ratings, the same models reach 87.5-99.7% balanced accuracy. Layer-wise probing of open-weight LALMs shows that pain can be decoded from early audio representations (74.1% balanced accuracy for binary prediction and 45.0% for three-class severity), but that this information, along with most pain-associated acoustic cues, decreases in deeper layers. In an acoustic perception test, weaker models cannot distinguish even large changes in these cues, whereas stronger models can but still fail to infer pain. Finally, we recover pain-related acoustic cues and directly reinject them as structured text into the prompt, which improves pairwise pain prediction by up to 32 points. These results indicate that current LALMs encode paralinguistic information in early layers that is largely unavailable when the model produces its answer.
103. 【2610.38561】Persona Learning from Feedback for Personalized AI Assistants
链接：https://arxiv.org/abs/2610.38561
作者：Yiran Shen,Yu Xia,Liuyi Yao,Prithviraj Ammanabrolu
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：assistants requires adapting, task contexts, requires adapting, Personalizing, persona
备注：
点击查看摘要
Abstract:Personalizing AI assistants requires adapting their behavior to users' preferences and task contexts. Existing work elicits assistant personas through persona prompting or finetuning on persona-conditioned generations. However, persona descriptions alone do not identify what to change or preserve in a particular response, or which behaviors suit the current request. We introduce $\textit{Persona Learning from Feedback (PLF)}$, a framework that learns reusable, preference-grounded assistant personas as independent LoRA adapters on a shared frozen backbone. During training, a frozen self-teacher conditions on persona specifications and response-specific feedback to provide token-level supervision on each adapter's own generated prefixes. A feedback gate focuses corrective updates on responses judged to need improvement. At deployment, a supervised router uses the request or dialogue to select a single adapter or a fixed composition, without user-specific finetuning. We evaluate PLF on single-turn personalization with model-generated critiques and multi-turn tutoring with simulated students' reactions. Across both settings, PLF improves model-judged preference fit over fixed-persona policies and strong baselines including persona prompting and persona distillation, demonstrating how feedback-trained behaviors can be reused for adaptive assistant personalization without degrading general capabilities.
104. 【2610.38546】Implicit Minor Detection from Multi-Turn Dialogue in High-Risk AI Interactions: A Constrained Self-Evolving LLM Harness
链接：https://arxiv.org/abs/2610.38546
作者：Xiaohan Zhang,Ying Zhao,Ruilin Guo,Xuanyuan Yu,Kaibo Huang,Yukun Wei,Zhongliang Yang,Linna Zhou
类目：Computation and Language (cs.CL)
关键词：applying age-conditioned safeguards, high-risk text interactions, text interactions, age-conditioned safeguards, current speaker
备注： 13 pages, 3 figures, including a technical supplement
点击查看摘要
Abstract:In high-risk text interactions, a system may need to estimate whether the current speaker is likely to be a minor before applying age-conditioned safeguards. Many text-based dialogue systems lack strong identity signals such as account information, verified identity, facial data, or device data, and must rely on weak cues that emerge across multiple turns. We formulate implicit minor detection under weak identity information as a high-risk user-attribute inference task and propose a constrained framework comprising a large language model, a Harness, and a Maker-Checker protocol. The framework controls when inference is invoked, requires structured dialogue-linked evidence, and subjects validation-selected Harness revisions to version-level review. On a controlled Chinese benchmark comprising 8,328 multi-turn dialogues, our method improves average precision and minor-class F1 by 7.60 and 6.37 percentage points, respectively, over direct inference using the same LLM family.
105. 【2610.38532】From Leniency to Suspicion: How an Open Post-Training Recipe Changes Whom LLM Fact-Checkers Believe
链接：https://arxiv.org/abs/2610.38532
作者：Nouar AlDahoul,Yasir Zaki
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY)
关键词：LLMs increasingly fact-check, increasingly fact-check online, fact-check online content, LLMs increasingly, identity in view
备注： 10 pages, 4 figures, 6 tables. Preprint, under review
点击查看摘要
Abstract:LLMs increasingly fact-check online content with the poster's identity in view. However, it remains unclear which post-training changes how much a model's truth judgement depends on the poster's social class, and how such a change happens inside the model. This paper investigates job-based doubt, i.e., judging a claim false more often when the poster has a lower-status job, using fact-checked claims with randomised poster profiles in base and post-trained checkpoints of six model families. Our results reveal that: 1) post-training creates such doubt only with the open Tülu-3 recipe (OLMo-2 32B, Tülu-3 70B), not with Meta's, Alibaba's or Mistral's; the effect is under one percentage point but statistically reliable, exceeding training drift and replicating on independent claim sets; 2) in OLMo-2 32B, the model does not start noticing the poster's job more, but starts interpreting it differently: a lower-status job pushes the base model toward "true" and the post-trained model toward "false"; and 3) this flip is carried by a single residual-stream direction and happens at supervised fine-tuning. The mechanism was confirmed on 400 new claims with pre-specified predictions. None of seven deployed commercial models doubts lower-status posters more. Our work underscores the need to audit speaker bias per training recipe.
106. 【2610.38520】What Hallucinations Have We Measured? Semantic Illusion, Confabulation, and Expressed Uncertainty
链接：https://arxiv.org/abs/2610.38520
作者：Hazel H. Kim,Jakub Binkowski,Ekaterina Fadeeva,Juan David Guerra,Yijie Hao,Zheng Yi Ho,Baixiang Huang,Buyun Liang,Siyuan Liang,Anh Tuan Luu,Potsawee Manakul,Niels Mündler-Sasahara,Aidar Myrzakhan,Shahrad Mohammadzadeh,Artem Shelmanov,Kai Shu,Ponhvoan Srey,Dacheng Tao,René Vidal,Haofei Yu,Lei Yu,Hongxiang Zhang,Yarin Gal,Hinrich Schütze,Philip H.S. Torr
类目：Computation and Language (cs.CL)
关键词：deploying large language, large language models, reliably and safely, critical challenges, challenges in deploying
备注：
点击查看摘要
Abstract:Hallucination remains one of the most critical challenges in deploying large language models (LLMs) reliably and safely. Yet the term is used inconsistently across the research community. We argue for a strict definition of hallucinations in LLMs. First, hallucinations should be confident incorrectness, kept separate from expressed uncertainty (UNK)1. UNK reveals that the model may be wrong, whereas a hallucination does not reveal it and is therefore more dangerous in practice. Conflating the two obscures their distinct behaviors and makes them hard to study. Second, hallucinations should be split by a simple re-elicitation test: does any semantically equivalent prompt recover the correct response? This test probes whether the relevant knowledge2 is accessible. If none does, the knowledge is likely absent or inaccessible, and the error type is a confabulation (CF) that likely requires external knowledge or retraining; if some does, the knowledge is accessible from the model but applied inconsistently, and the error type is a semantic illusion (SI) that simple prompt or decoding changes can mitigate3. Re-evaluating 14 recent hallucination methods under this taxonomy yields three findings. (1) Detection methods flag UNK more often than hallucinated responses (i.e., SI and CF). (2) Mitigation methods are far more effective for SI than for CF. (3) Latent signals are more informative than output-level signals for CF. An audit of recent hallucination papers shows that most conflate these categories. We recommend reporting CF, SI, and UNK separately, and introduce a practical labeling protocol that needs only a modest number of semantically equivalent prompts.
107. 【2610.38511】Similar Scores, Opposite Failures: Human Correction as an Instrument for Video Caption Evaluation
链接：https://arxiv.org/abs/2610.38511
作者：Ankit Khedia,Astitva Jaiswal,Chaitanya Prasad,Anisha Raju,Aangeeras
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：Video captioning models, single-sentence image captions, Video captioning, increasingly evaluated, designed for single-sentence
备注： 29 pages, 10 figures, 17 tables. Analysis code included as ancillary files. Data available on request at [this https URL](https://data.deccan.ai)
点击查看摘要
Abstract:Video captioning models are increasingly evaluated with metrics designed for single-sentence image captions. We test whether those metrics recover human judgement on dense, multi-window video description, and find that they do not. Six frontier models captioned 196 short clips. Trained annotators post-edited every caption against a failure taxonomy fixed before annotation, once as a whole-clip overview and once window by window, and a quality-control reviewer checked each annotation, yielding 1,144 overview verdicts and 4,672 window verdicts. Reference-overlap metrics do not separate the captions reviewers accepted from those they rejected: BLEU-4 reaches an AUC of 0.510 and a CIDEr-style tf-idf cosine 0.528, and restricting references to reviewer corrections leaves BLEU-4 at 0.479 (95% CI 0.431-0.529), indistinguishable from chance. Review granularity changes what is detected. The same annotators record camera and shot errors on 2.7% of captions when reading the overview and on 12.8% when reading window by window, a 4.7x gap that holds across every robustness check we ran, on the mode most likely to need a near-total rewrite. Correct-overview rates span 54.1-72.7%, but only 4 of 15 pairwise gaps survive Bonferroni correction (5 after adjusting for reviewer and clip effects), and accuracy and derived severity order the models differently. Models with statistically indistinguishable accuracy have sharply different failure profiles, which no single scalar shows. We argue that caption evaluation should record what was wrong rather than only how often, and we report the taxonomy, the per-model profiles, the corpus properties that predict difficulty, and the code that regenerates every number in the paper.
108. 【2610.38509】Leaving the Cave by Breaking the Symmetry: Identifiable Structure Can Enable Cross-Lingual Generalization
链接：https://arxiv.org/abs/2610.38509
作者：Frederick Riemenschneider,Anette Frank
类目：Computation and Language (cs.CL)
关键词：Platonic Representation Hypothesis, Relational Identifiability Hypothesis, shared, shared vocabulary, Platonic Representation
备注： 12 pages, 7 appendix pages, 1 figure, 10 tables
点击查看摘要
Abstract:Why do multilingual language models generalize across languages even without shared vocabulary? The Platonic Representation Hypothesis posits that models trained on different data, objectives, and modalities converge toward a shared statistical model of reality. Extending this intuition to representations inside a single multilingual model, we propose the Relational Identifiability Hypothesis: relations in a shared world can promote entity alignment by ruling out alternative cross-lingual correspondences, making corresponding entities more likely to become directly readable without an additional transformation. We test this hypothesis in a controlled bilingual Transformer whose two token-disjoint languages describe the same modular-arithmetic world. We find that the languages can develop a shared answer subspace without aligning corresponding numbers. Cross-lingual alignment emerges selectively when the available relations constrain number correspondences: subtraction makes zero task-level identifiable, inducing selective alignment, while an order-sensitive operation produces broad alignment across numbers. These patterns predict success in zero-shot cross-lingual transfer. Together, our results show that relational structure in a shared world can promote entity alignment and enable transfer.
109. 【2610.38452】When Support Crosses the Line: Evaluating Ethical Boundaries in LLM-Based Mental Health Support in the Chinese Context
链接：https://arxiv.org/abs/2610.38452
作者：Bichen Wang,Yufeng Jin,Yixin Sun,Yanyan Zhao,Bing Qin
类目：Computation and Language (cs.CL)
关键词：Large language models, Large language, existing benchmarks primarily, benchmarks primarily focus, models consistently adhere
备注：
点击查看摘要
Abstract:Large language models (LLMs) are increasingly explored for mental health support, yet existing benchmarks primarily focus on supportive capabilities rather than whether models consistently adhere to professional ethical boundaries. This issue is particularly important in multi-turn interactions, where users may introduce new information or exert emotional pressure over time. We introduce MindEth-Bench, a benchmark for evaluating ethical boundary adherence in multi-turn mental health interactions. Grounded in professional ethical standards for mental health services in China, MindEth-Bench covers multiple dimensions of ethical conduct and uses Monte Carlo Tree Search (MCTS) to systematically explore interaction trajectories that stress-test model boundaries. Experiments across nine LLMs show that models that initially respond appropriately can shift toward ethical violations over extended interactions, with substantial variation across models and ethical dimensions. These results show that single-turn evaluations may overestimate the ethical reliability of LLM-based mental health support systems and highlight the need to evaluate ethical behavior throughout multi-turn interaction.
110. 【2610.38450】CoSTA: Co-Evolving Self-Play Timeline Agents for Complex Temporal Knowledge Graph Question Answering
链接：https://arxiv.org/abs/2610.38450
作者：Haikun Liu,Zhichao Yan,Shaoru Guo,Jiawei Gu,Victor Gutierrez Basulto,Jiapu Wang,Ru Li
类目：Computation and Language (cs.CL)
关键词：Temporal Knowledge Graph, Graph Question Answering, Complex Temporal Knowledge, Knowledge Graph Question, Temporal Knowledge
备注：
点击查看摘要
Abstract:Complex Temporal Knowledge Graph Question Answering (TKGQA) involves more than retrieving facts: the timestamp or validity interval of one fact can constrain the search for another, while multiple intervals may jointly determine the answer. Existing single-agent TKGQA methods typically investigate these facts within a single reasoning context, where evidence from different event histories can become interleaved and distractors can misdirect subsequent retrieval. Distributing investigations across agents can separate event-specific evidence into distinct contexts, but effective collaboration must still be learned because question decomposition, subtask investigation, and evidence synthesis are interdependent. To address these limitations, we propose \textbf{Co}-Evolving \textbf{S}elf-Play \textbf{T}imeline \textbf{A}gents (CoSTA), which organizes facts into navigable timelines, delegates investigations to workers, and uses a coordinator to synthesize their original timestamped observations. CoSTA initializes a shared policy through role-conditioned distillation and jointly optimizes both roles through co-evolving self-play with final-answer rewards. Across three benchmarks, CoSTA achieves state-of-the-art overall performance, with particularly strong gains on complex multi-fact questions, while its compact variants outperform the evaluated closed-source LLM agents.
111. 【2610.38448】Benchmarking decision models against generative large language models on medical multiple-choice questions
链接：https://arxiv.org/abs/2610.38448
作者：Oscar Freyer,Daniel Glauert,Moritz Hofmann,Max Ostermann,Stephen Gilbert,Jakob Nikolas Kather
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：predefined options, decision models, Mercury Decide, Abstract, configurations
备注：
点击查看摘要
Abstract:Medical workflows often require selecting among predefined options. We compared 13 decision models with 15 generative large language model configurations, including different reasoning settings, using 1,273 four-option MedQA questions under text-only conditions. Accuracy ranged from 67.71% to 89.08% for decision models and from 78.32% to 96.86% for generative configurations. Among the tested configurations, Mercury Decide had the lowest mean final-attempt cost (USD 0.0069 per 1,000 questions), while Clef Omni had the shortest median request duration (0.291 seconds). Decision-model probabilities distinguished correct from incorrect answers, but error rates among answers assigned probabilities of at least 0.90 ranged from 1.71% to 12.77%. Answer agreement across three runs ranged from 92.30% to 100%, while probability repeatability varied. Decision models offered resource advantages alongside lower peak accuracy; selecting models for medical workflows requires considering probability quality and repeatability together with accuracy and the resources needed to detect and correct errors.
112. 【2610.38439】Predict the Reward, Not the World: World Reward Models for ML Agents
链接：https://arxiv.org/abs/2610.38439
作者：Yuxiao Lee,Zherui Yang,Fan Liu,Hao Liu
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Search-based machine-learning, improve a solution, solution by writing, writing candidate programs, Search-based
备注：
点击查看摘要
Abstract:Search-based machine-learning (ML) agents improve a solution by writing candidate programs, executing them, and keeping the best. Execution gives the search reliable feedback, but a single run can take hours, and most runs end in programs the agent discards. Recent agents try to save these runs by letting a large language model act as a world model that predicts how each program would perform. However, a wrong prediction feeds the search false information. We argue that an agent does not need to predict the outcome of execution, only the decision that execution would support. We propose World Reward Models (WRMs), which learn from real execution results to choose directly which candidate will perform better, without predicting any intermediate outcome. We prove that a selector which predicts execution from the same inputs cannot outperform the best direct selector, that predicted outcomes add an error which can make a likelihood-trained world model prefer the worse program on every instance, and that cross-entropy on executed winners drives a WRM toward the optimum. Our WRM, ForeCritic, adapts a 9B typed decision model to 5k executed program pairs. On 12,438 held-out pairs from 26 ML tasks, it picks the better program 79.4% of the time, while general-purpose LLM judges reach 53--62%, and its probabilities are calibrated. Inside AIDE, ML-Master, and ForeAgent, it cuts mean task time by 4.4--7.5 times while keeping their mean scores within 0.01 or raising them.
113. 【2610.38438】RoleFUZZER: User-Behaviour Fuzzing for Role-Specification Adherence in Large Language Models
链接：https://arxiv.org/abs/2610.38438
作者：Kshitish Ghate,Michael Saxon,Tadayoshi Kohno,Aylin Caliskan
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
关键词：Large language models, Large language, conditioned into specific, system prompts, Large
备注： Preprint under review
点击查看摘要
Abstract:Large language models (LLMs) are often conditioned into specific roles such as tutors, interviewers, or agents acting for an organisation, through system prompts that specify how they should behave. The extent to which they can actually adhere to these role specifications under multi-turn interactions with realistic and potentially problematic users is understudied. To address this, we propose user behaviour fuzzing, a guided search over plausible user behaviours that can lead a model to violate its role specification, and introduce RoleFUZZER, a framework that adapts coverage-guided fuzzing from software testing to operationalise the search. RoleFUZZER constructs scenarios guided by a user-behaviour coverage grid that crosses a grounded taxonomy of user interaction types and the role specification of the model under test i.e. what is under pressure. With RoleFUZZER, we evaluate LLMs across five role specifications, including tutor, financial advisor, AI interviewer, antagonistic roleplay character, and web agent under organisational policies. Across simulated and validated conversations in 15 campaigns, the framework reaches = 89% coverage of the 64-cell behaviour grid for every model and role tested, and surfaces specification-grounded failure patterns. At an equal budget of conversations, removing the RoleFUZZER grid and asking an LLM to generate specification targeting users free-form lowers coverage to 39-69% with fewer kinds of violations. The same specification can also fail in opposite directions. For e.g., on the antagonist, open-weight target models override the player's safety signals, while Claude Haiku 4.5 breaks character instead. RoleFUZZER offers a tractable way to uncover vulnerabilities in LLMs across role specifications by testing diverse user behaviours that are likely to violate them.
114. 【2610.38433】Does Chain-of-Thought Reliance Follow a Scaling Law?Larger Models Follow Their Steps More, Independent of Post-Training
链接：https://arxiv.org/abs/2610.38433
作者：Esther Xin
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：final answer depends, Reliance, final answer, answer depends, capable
备注： 15 pages,5 figures, code and data in [this https URL](https://github.com/ethxin0011/cot-scale-reliance.cl)
点击查看摘要
Abstract:Monitoring a model's chain of thought (CoT) is useful for safety only if the final answer depends on the written steps. Prior measurements use natural tasks where true reasoning is unknown. We measure it against an exact ground truth on a synthetic box-swap task where every intermediate state is known. We replace one step inside a model's own chain with a wrong but valid state and check whether the answer follows the edit. We evaluate 25 open-weight models--Qwen3 (0.6B-14B, base and post-trained), Gemma 3 (1B-12B), Llama 3 (1B-8B), and 8 OLMo 2 checkpoints--on 7,200 items each. We find five results: (1) Reliance rises with size and saturates. In Qwen3, answer reliance rises from 16-18% at 0.6B to 97-100% at 14B; a logistic in log N fits well (R^2 = 0.89), and Gemma 3 and Llama 3 follow this trend. (2) Capability is the main driver. Across all models, reliance and step accuracy correlate at r = 0.99; equally capable models are equally reliant, regardless of family. (3) Size still matters at matched correctness. On problems solved correctly before the edit, reliance increases with size among capable models (rho = 0.80, p = 0.002). (4) Post-training does not change reliance. Base and post-trained Qwen3 differ by at most 0.03 from 4B upward; a large gap at 1.7B matches a large capability gap. (5) No model up to 14B solves the task internally. Without written steps, every capable model is at chance once a problem has four or more swaps. OLMo 2 checkpoints never become capable, making the training-token axis a limitation. On tasks requiring step-by-step computation, scale makes the chain of thought more load-bearing, supporting CoT monitoring in this regime.
115. 【2610.38422】Deeper Backpropagation or More Updates? Training Looped Language Models Across Budgets
链接：https://arxiv.org/abs/2610.38422
作者：Jiarun Shen,Xiao Li
类目：Computation and Language (cs.CL)
关键词：Repeating transformer layers, Repeating transformer, adding parameters, training, Repeating
备注：
点击查看摘要
Abstract:Repeating transformer layers can improve language model performance without adding parameters. However, the added computational depth increases both forward and backward computation costs during training. Training such models requires choosing a backward window that controls how many final layer repetitions retain gradients during backpropagation. Longer windows improve per-token learning efficiency but increase per-update cost, while shorter windows reduce per-update cost but sacrifice per-token efficiency. Understanding which window performs better requires examining different training budgets. Prior work compares backward windows across computational budgets. When training budgets increase, it remains unclear whether the same window continues to perform best. We measure how the relative performance of two windows shifts as training budgets grow. Our method isolates per-token efficiency by training both windows on identical data, and compares complete training recipes at approximately equal training time at each budget. Results show that shorter windows lose much of their advantage as budgets grow. Controlled experiments show that both benefits decrease at larger budgets, with a greater decline in the benefit of extended training. Backward window comparisons depend on the available budget, and the preferred window among the tested candidates can change as the budget grows.
116. 【2610.38407】Holistic Diversity Management in LLM-Guided Evolutionary Search
链接：https://arxiv.org/abs/2610.38407
作者：Seockbean Song,Youngsik Yoon,Sungjae Lee,Siwei Wang,Wei Chen,Jungseul Ok
类目：Computation and Language (cs.CL)
关键词：iteratively selects existing, selects existing programs, search iteratively selects, guide subsequent search, iteratively selects
备注：
点击查看摘要
Abstract:LLM-guided evolutionary search iteratively selects existing programs, uses LLMs as mutation operators to generate a new program, and evaluates it to guide subsequent search. To preserve diversity and reduce the risk of premature convergence, it often adopts the island model, in which groups of programs, called islands, evolve separately. However, this structure alone does not guarantee distinct search directions, as islands may start from similar programs or converge toward similar solutions during evolution. This suggests that diversity must be considered throughout the search process, from initializing islands with distinct programs to developing island trajectories and sharing discoveries across islands. To this end, we propose HoliDivEvolve , a holistic approach to establishing and effectively utilizing island diversity throughout evolutionary search. Specifically, we introduce diverse initialization to establish different initial programs, trajectory adaptive island selection to allocate computation based on current score, recent growth, and the remaining search budget, and growth adaptive migration to share alternative ideas across islands. Experiments across diverse benchmarks show that HoliDivEvolve achieves competitive performance on most evaluated tasks. We further conduct controlled analyses to examine how our design choices affect search behavior and final performance.
117. 【2610.38402】LLMs as Judges and Examinees for Medical Item Difficulty Prediction
链接：https://arxiv.org/abs/2610.38402
作者：Xiaomeng Xiong,Marcus Walker,Yanlin Jiang
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：remains challenging, rho, models, difficulty, signals
备注：
点击查看摘要
Abstract:Predicting medical item difficulty from item text alone remains challenging. Prior work has largely relied on linguistic features or text representations derived from general-purpose and domain-specific language models, but predictive performance has remained modest. This study takes a different approach by using difficulty-related signals that emerge when open-weight large language models (LLMs) interact with test items. Using USMLE and PANCE medical licensure examination items, we examined LLMs in two roles. As judges, the models provided absolute difficulty ratings and pairwise difficulty comparisons. As simulated examinees, they generated response-based signals including correctness, keyed-option probability, response entropy, and the probability margin between the top two options. We then combined the strongest judgment and response signals with baseline item features in downstream prediction models. Pairwise difficulty judgment was the most consistent individual LLM signal across both datasets, while simulated-examinee signals and absolute judgments provided complementary information in some feature combinations. The best configurations reached a Spearman correlation of $\rho$ = .447 on USMLE, exceeding the best system in the 2024 BEA shared task ($\rho$ = .25), and $\rho$ = .494 on PANCE. These results were comparable to direct absolute judgments from the proprietary frontier model Claude Opus 5 ($\rho$ = .443 and .488, respectively). Overall, judgment and response signals from open-weight LLMs captured difficulty-related information beyond text representations. Because these models can be deployed locally, the approach offers a practical option for predicting the difficulty of unreleased or high-security test items.
118. 【2610.38384】Pronunciation-Supervised Text-to-Speech for Irish (Gaeilge)
链接：https://arxiv.org/abs/2610.38384
作者：William Lane
类目：Computation and Language (cs.CL)
关键词：documented in dictionaries, Common Voice speech, Speech technology, transcribed speech, dialect descriptions
备注：
点击查看摘要
Abstract:Speech technology for Irish must be built from very little transcribed speech, yet the language's pronunciation is thoroughly documented in dictionaries and dialect descriptions. We ask whether that documentation can supervise the adaptation of a large pretrained text-to-speech model to a language it was never trained on. We adapt Qwen3-TTS to Irish using 12.8 hours of crowdsourced Common Voice speech, correcting two defects in the vendor fine-tuning script that training loss does not reveal (Appendix A), and reach character error rate (CER) 0.225 on held-out cross-corpus text, against 0.397 for an exhaustively tuned from-scratch system and 0.437 for zero-shot cloning. We then compile the Wiktionary pronunciation record into a lexicon and a phoneme critic, and use that critic as the verifier in one round of rejection fine-tuning: CER 0.159 cross-corpus and 0.067 in-domain, with the voice still anonymous across all 202 training speakers. The same critic explains why the voice cannot yet be steered to a minority dialect: 229 of 12,311 training clips lean Ulster against 8,451 Munster. Pronunciation documentation, converted into a verifier, is a practical lever for low-resource speech synthesis, and a way to measure what the available data cannot support.
119. 【2610.38383】How Much Do Teacher Bridges Matter in Interleaved On-Policy Distillation?
链接：https://arxiv.org/abs/2610.38383
作者：Leran Chen,Zhiyuan Yao,Lingnan Kong
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：On-policy distillation, predictions on student-generated, OPD, On-policy, teacher predictions
备注： 25 pages, 8 figures, 18 tables
点击查看摘要
Abstract:On-policy distillation (OPD) trains a student using teacher predictions on student-generated responses. Some methods insert teacher-written segments, or bridges, but disagree on how to train on those tokens. We compare no loss, forward KL and reverse KL on identical trajectories using Qwen3-1.7B as student and Qwen3-4B-Instruct-2507 as teacher. We separate gradients by which model wrote each token. Under our short training runs and BF16 inference, pre-specified tests detected no added accuracy benefit from teacher text as context or as a training target. A one-paragraph bridge after the student's first paragraph is only 2.7% of the tokens and contributes a small gradient nearly orthogonal to the student-token gradient. In this main comparison, the three loss treatments gave similar gradient directions and were statistically equivalent in accuracy within a +/-5 percentage-point margin. An exploratory pilot placing the bridge where failed responses disagreed most with the teacher suggested a benefit from teacher-token loss. A pre-specified replication did not confirm this, and fourfold weighting showed no significant accuracy benefit. Two limitations bound these findings. Among the tested amounts of training on fixed responses, gains peaked after four updates before the student increasingly failed to stop generating. None of the tested online training variants met the criteria for stable stopping and improved accuracy. Our evaluations used BF16 inference, which rounds away most of the small weight changes. When evaluated in exact FP32, the tested replication models frequently reached the response-length limit. Our accuracy findings therefore concern BF16 inference; they do not establish that teacher-token supervision is generally unnecessary.
120. 【2610.38366】CC-RAG: Temporal Causal Chain-Guided RAG for Temporal Knowledge Graph Question Answering
链接：https://arxiv.org/abs/2610.38366
作者：Xinxun Zhang,Yuxiang Wang,Shanshan Feng,Jun Zhang,Tieyun Qian,Binhang Yuan,Kai Yin,Xiao Yan,Jiawei Jiang
类目：Computation and Language (cs.CL)
关键词：spurred growing interest, Recent advances, graph question answering, large language models, complex temporal reasoning
备注： 14 pages, 6 figures, ICDE2027 Research Track
点击查看摘要
Abstract:Recent advances in large language models (LLMs) have spurred growing interest in temporal knowledge graph question answering (TKGQA). Most existing methods rely on semantic retrieval augmented generation (RAG), which suffers from fundamental limitations in complex temporal reasoning. On the one hand, semantic-based RAG methods struggle to capture the multi-hop causal knowledge required for complex temporal reasoning. On the other hand, these methods introduce excessive redundant knowledge, trapping LLMs in spurious reasoning chains that potentially result in erroneous answers. To address these problems, we propose a Temporal Causal Chain-guided RAG (TCC-RAG) to retrieve a concise yet sufficient set of causal knowledge to support complex temporal reasoning. Specifically, we reformulate the temporal knowledge graph into a multi-view causal cue graph, with causal cues inducing connections that explicitly encode temporal evolution and causal dynamics. Based on the causal cue graph, we construct a query-specific temporal causal chain to guide the retrieval process, enabling the model to focus on a concise set of causally relevant knowledge. Finally, we introduce a causal evidence filter to refine the retrieved knowledge, ensuring its causal consistency with the underlying temporal logic of the query. Experiments show that TCC-RAG achieves substantial performance gains of 10.72% and 8.40% on two TKGQA benchmarks, and delivers a remarkable 62.97% improvement for multi-step temporal reasoning.
121. 【2610.38361】Manifest-Assisted Retrieval in Artificial Intelligence Search: Evidence Gaps and a Security Evaluation
链接：https://arxiv.org/abs/2610.38361
作者：Gangzhen Qian,Kainan Zhou,Chuhong Xu,Hang Xiao,Zhi Jing,Rong Bao,Lu Yi,Zhaoyi Li
类目：Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL); Databases (cs.DB); Information Retrieval (cs.IR)
关键词：external sources retained, isolate the incremental, incremental security, security or utility, retrieval remained fixed
备注： 8 pages, 3 figures, 5 table. Accepted at IEEE FLLM 2026. Author accepted manuscript. Reproducibility artifact: [this https URL](https://fllm-conference.org/2026/)
点击查看摘要
Abstract:The external sources retained in our non-systematic audit did not isolate the incremental security or utility effect of consuming /llms.txt while ordinary retrieval remained fixed. The initial audit covered 39 sources through August 16, 2026, with later targeted updates through October 14. A purposive 15-tool study observed no automatic requests in 45 trials at one endpoint; a non-probability log study found that 97% of domains with detected files received no recorded requests in May 2026. We separate publication, fetch, model use, and action; define five operational threat paths and paired security-utility estimands; and exercise deterministic pre-model controls with 30 author-defined parser fixtures. We also execute a reduced pilot: 24 tasks across eight sites, five conditions, and 120 real model responses, with ordinary evidence held fixed. Frozen automatic scoring identifies attacker-target adoption in $21/24$ raw adversarial responses, with one further response unresolved; the adversarial URLs-only condition has no known adoptions and one unresolved response. The pilot measures output changes conditional on supplied evidence, without live retrieval or external actions. Together, the framework and pilot distinguish manifest-content effects from publication, discovery, and downstream execution, while leaving full retrieval and agent evaluation open.
122. 【2610.38336】Statutory Article Retrieval from Arabic Judicial Narratives: Family Law and Law of Evidence in Saudi Arabia
链接：https://arxiv.org/abs/2610.38336
作者：Zarah Shibli,Areej Alhothali,Lulwah Alharigy
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Reliable legal information, authoritative statutory sources, ground legal reasoning, Reliable legal, Saudi Arabic judicial
备注： Submitted to the MLNLP 2026 Conference
点击查看摘要
Abstract:Reliable legal information retrieval must ground legal reasoning in authoritative statutory sources. In Arabic judicial settings, long factual narratives must be mapped to concise articles that express abstract legal rules, creating a narrative-to-legal-rule abstraction gap. Using a benchmark of 200 synthetic Saudi Arabic judicial cases and 381 articles from the Family Law and the Law of Evidence, we compare word-level and char-4 BM25, zero-shot dense encoders, per-law retrieval, dense-sparse reciprocal rank fusion (RRF), and a proposed Decompose-and-Refine (DaR) pipeline. In DaR, a large language model splits each narrative into disputed legal issues and generates, for each issue, three retrieval channels: a statute-style paraphrase, legal keywords, and a candidate taxonomy chapter. On the whole corpus, the instructed Qwen3-Embedding-8B dense encoder reaches an MRR@10 of 0.849, the best result of any non-decomposed system. Under per-law scoring, this advantage reflects Family Law only: restricted to Evidence articles, the same encoder drops to an MRR@10 of 0.087, and fusion with BM25 raises it only to 0.207. DaR changes this picture, raising Evidence MRR@10 to 0.696 and Family MRR@10 to 0.864. Without decomposition, the two laws favor different strategies; under DaR, the same retriever achieves the best MRR@10 on both.
123. 【2610.38333】Fu-TCM: a multimodal large language model that learns traditional Chinese medicine diagnostic reasoning
链接：https://arxiv.org/abs/2610.38333
作者：Jiahang Xie,Tong Jin,Lingfei Wu,Yaping You,Jie Huang,Yuanchang Jin,Weikang Gong,Wei Cheng,Jia You,Pan Sun,Jianfeng Feng,Hongcheng Guo
类目：Computation and Language (cs.CL)
关键词：Large language models, Large language, multimodal TCM LLM, TCM, accelerating medical model
备注：
点击查看摘要
Abstract:Large language models (LLMs) are advancing in multimodal understanding and reasoning, accelerating medical model development. Traditional Chinese medicine (TCM) uses evidence from the four examinations to determine disease location and nature, differentiate syndromes and guide treatment. This diagnostic process constitutes a complete chain of reasoning. We present Fu-TCM, a multimodal TCM LLM trained on data retained after cleaning and filtering an initial collection of approximately 7.12 million examples derived from classical texts, textbooks, medical images, clinical cases and publicly available datasets. The model was further optimized using Bianzheng-Grounded Policy Optimization (BGPO), which explicitly targets syndrome differentiation through rewards for response format, intermediate TCM reasoning fields and final syndrome predictions. Fu-TCM-27B led all six benchmarks spanning TCM reasoning, text and vision, achieving a macro-average accuracy of 81.01% and surpassing the strongest reference model with complete results across all six benchmarks by 7.13 percentage points. Its largest margin was 11.90 points on Fu-bianzheng. Case-level analyses further showed a positive association between intermediate-field accuracy and final syndrome-selection accuracy in most evaluated models. These findings suggest that TCM syndrome differentiation may be computationally learnable and that structured reasoning supervision can improve performance across TCM tasks.
124. 【2610.38312】Robust LLM-Generated Text Detection through Matched Conditional-Response Probing
链接：https://arxiv.org/abs/2610.38312
作者：Junfeng Hao,Wenxian Wang,Tao Leng,Xingshu Chen,Hao Ren,Haizhou Wang
类目：Computation and Language (cs.CL)
关键词：large language model, language model, Reliable detection, textbf, meaning-preserving text revisions
备注：
点击查看摘要
Abstract:Reliable detection of large language model (LLM)-generated text must remain effective under meaning-preserving text revisions, such as rewriting, polishing, and expansion. In practice, LLM outputs are often rewritten, polished, expanded, translated, or lightly edited before publication, weakening the lexical, syntactic, and local discourse cues used by many existing detectors. We introduce \textbf{M}atched \textbf{C}onditional-\textbf{R}esponse \textbf{P}robing (\textbf{MCRP}), a frozen-LLM probing framework that measures how a candidate text responds when a semantically matched condition is introduced, removed, or replaced. For each candidate, the frozen probe LLM first generates a short semantic condition and then compares token-level losses and prompt--target attention measurements under matched, null, and control conditions. It aggregates bidirectional conditioned alignment, counterfactual sensitivity, and internal attention coupling into a probe representation, on which it trains a lightweight supervised random-forest classifier while keeping the probe LLM frozen. Across standard detection, robustness to meaning-preserving text revisions, multilingual, and multidomain evaluations, MCRP achieves consistently strong performance. Under rewriting, polishing, and expansion, it reaches 93.97\% F1 and 96.24\% ROC-AUC, outperforming the strongest evaluated baseline by 19.50 and 25.77 percentage points, respectively. Component ablations show that the response feature groups make complementary contributions to predictive performance. In the condition-control study, matched conditions consistently outperform random, mismatched, and corrupted alternatives, demonstrating a measurable incremental benefit from semantic matching.
125. 【2610.38293】EPSBench: Benchmarking LLM Agents for Earnings Forecasting
链接：https://arxiv.org/abs/2610.38293
作者：Yanbin Yin,Chuyu Xu,Liqin Ye,Jing Peng,Chao Zhang
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Quarterly earnings forecasting, connect financial evidence, income-statement line items, earnings forecasting requires, Quarterly earnings
备注：
点击查看摘要
Abstract:Quarterly earnings forecasting requires an agent to connect financial evidence to the income-statement line items that produce a company's reported earnings. We introduce EPSBench, a benchmark for studying this ability across 149 firm-quarters from 86 US companies. Agents read a fixed collection of filings, earnings-call transcripts, and news, then estimate company-specific income-statement drivers with cited evidence. A shared accounting calculator converts these estimates into GAAP diluted earnings per share (EPS), providing a common basis for comparing forecast assumptions. We evaluate six models under a common harness, assessing point forecast error relative to date-matched analyst consensus and checking whether their prediction intervals achieve the stated coverage. Three models show no detectable paired difference in point forecast error from consensus, and three have higher error. Across all six models, nominal 80% intervals contain reported EPS in only 37--70% of runs that return an interval, and most misses fall above the upper bound because the forecasts run low. The driver records trace most of the error gap between models to operating-income estimates.
126. 【2610.38282】Human Evaluation with Automatic Scores: Separating Sampling and Estimation Gains
链接：https://arxiv.org/abs/2610.38282
作者：Yang Liu,Ran Zou,Zhengtao Gui,Chao Xue
类目：Computation and Language (cs.CL)
关键词：Automatic judges offer, Automatic judges, offer inexpensive information, fewer human annotations, judges offer inexpensive
备注：
点击查看摘要
Abstract:Automatic judges offer inexpensive information about model outputs, but an evaluation team must decide how to turn that information into fewer human annotations. Scores can guide sampling, correct an estimate, or support adaptive acquisition, and these uses may capture overlapping gains. We compare their incremental value on fixed evaluation pools from three preference datasets, and 19 attributed-question-answering systems. Matched comparisons use the same scores, account for pilot labels and budget rules, and separate acquisition from estimation. Score stratification is a strong reference: after ten-bin stratification, the best fixed local linear correction offers only 1.00--5.27% further MSE reduction in the primary configurations. Unequal allocation offers larger gains in some pools, but can increase error when confident scores conceal variable human outcomes. Comparison on AttributedQA, shrinking learned quotas toward score-only quotas lowers pooled MSE for 17 of 19 systems relative to the score-only procedure, but ordinary pooled 95% intervals cover only 88.09--93.54%. The same shrinkage improves on the score-only reference in only one of eight preference configurations. Ranking experiments further show that lower individual MSE need not improve system selection. These findings support a simple starting procedure, proportional sampling within score strata followed by a stratified human mean, and identify what additional evidence is needed to justify correction or adaptation for the intended evaluation output.
127. 【2610.38267】Cross Lingual Abstractive Summarization for Marathi
链接：https://arxiv.org/abs/2610.38267
作者：Asang Dani,Shailesh R Sathe
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：natural language processing, advancing natural language, Text summarization, research and applications, important task
备注：
点击查看摘要
Abstract:Text summarization is an important task for advancing natural language processing (NLP) research and applications in Indic languages. Despite progress in extractive and abstractive summarization, both monolingual summarization (MS) and cross-lingual summarization (CLS) remain relatively underexplored for these languages. Recent advances in Transformer architectures, pretrained language models (PLMs), and dataset availability offer new opportunities to address this gap. In this work, we construct cross-lingual training data from the monolingual Marathi MahaSum corpus and combine it with the smaller cross-lingual PMIndiaSum dataset to fine-tune IndicTrans2, an Indic neural machine translation model. We evaluate the resulting models using ROUGE, chrF and chrF++ metrics, focusing primarily on Marathi-to-English summarization due to computational constraints. We also discuss the applicability of our approach to the English-to-Marathi direction. Our experiments demonstrate that leveraging a pretrained translation model yields substantially better summarization performance than the IndicBART and mBART-50 based baselines reported in previous work on cross-lingual summarization.
128. 【2610.38249】Keepace.ai: A Technical Report on a Large Language Model for Sports and Health
链接：https://arxiv.org/abs/2610.38249
作者：Keepace.ai Team,Bowen Wu,Heng Wang,Jufeng Mi,Kang Liu,Kun Li,Ling Man,Long He,Suli Fan,Xiaosen Jiang,Yi Wang,Ying He,Yu Tang,Yunhan Li,Zifang Liu
类目：Computation and Language (cs.CL)
关键词：Large language models, lacks curated training, curated training corpora, Large language, language models hold
备注： 21 pages, 3 figures, 13 tables. Technical report
点击查看摘要
Abstract:Large language models hold promise for delivering safe, scientifically grounded, and personalized exercise guidance at scale, yet the sports-and-health domain still lacks curated training corpora and rigorous evaluation benchmarks. this http URL is a domain-specialized LLM built on Qwen3.5-27B through post-training, currently optimized for Chinese-language exercise guidance across three tasks: workout plan generation, sports-and-health knowledge QA, and personal sports data analysis. In internal rubric-based evaluations designed with domain experts, this http URL outperforms its base model and larger general-purpose baselines, with the most pronounced improvements in exercise safety, boundary compliance, structured output completeness, and scientific accuracy. This report summarizes the current design, data pipeline, safety-first evaluation framework, and evaluation results of this http URL.
129. 【2610.38231】3DCodeEdit: Benchmarking Instruction-Guided 3D Program Editing
链接：https://arxiv.org/abs/2610.38231
作者：Binjie Wang,Xinran Zhao,Shoubin Yu,Xiang Li,Silong Yong,Chenyang Yang,Tianyu Chen,Ruofei Du,He He,Chen Zhao,Tongshuang Wu
类目：Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
关键词：Language models increasingly, models increasingly create, increasingly create, precise control, editing existing programs
备注：
点击查看摘要
Abstract:Language models increasingly create 3D content through code, with practical workflows often relying on editing existing programs rather than regenerating them, for speed and precise control. Such edits require understanding what a program builds and foreseeing how a change reshapes it, yet prior benchmarks mostly evaluate generation or simple edits against a reference. In this paper, we introduce 3DCodeEdit, a benchmark of 2,592 instruction-guided editing tasks over 108 Blender programs that construct objects entirely in code. Its edits range from parameter to structural changes, style redesign, and scene composition, and are judged from renders against human-reviewed checklists. Across 17 models, the average success rate is only 37%, and even the strongest models complete fewer than 60% of structural and style edits. As models become more capable, their failures shift to geometric foresight, where the code runs but builds the wrong shape. We therefore study whether multi-turn feedback can compensate for this missing foresight. Feedback from a separate reviewer model that describes how the renders differ from the request helps most, more than maximum reasoning effort or an agent harness, and can let a 27B model surpass the single-shot rate of a 2.8T-parameter model.
130. 【2610.38204】wo Birds with One Stone: Skill-Augmented Generation via Unified Retrieval and Utilization
链接：https://arxiv.org/abs/2610.38204
作者：Zihan Luo,Jiahui Kang,Wei Zhou,Qilin Fan,Mingyan Li,Chao Chen
类目：Computation and Language (cs.CL)
关键词：Large Language Models, Large Language, reusable external knowledge, demonstrated substantial potential, ability of Large
备注： Under Review
点击查看摘要
Abstract:Skills, as reusable external knowledge, have demonstrated substantial potential to enhance the ability of Large Language Models (LLMs) to solve complex tasks. However, existing skill-augmented generation methods typically rely on separate models for skill retrieval and skill utilization, resulting in disconnected optimization of these two processes. To address this gap, we propose Skill-Uni, a unified reinforcement learning framework that jointly optimizes skill retrieval and utilization. Specifically, we adopt a generative skill retrieval paradigm that enables a single model to both learn how to retrieve relevant skills and leverage them for task solving. To effectively train this unified framework, we introduce a two-stage training strategy consisting of warm-up instruction tuning and GRPO, allowing the model to progressively improve its skill retrieval and utilization capabilities. Extensive experiments on five skill-related benchmarks demonstrate the effectiveness of Skill-Uni, yielding average performance gains of 66.59% and 34.82% over the base Qwen3-4B and LLaMA3.1-8B models, respectively. Besides, Skill-Uni is compatible with various skill retrieval engines, demonstrating its flexibility and generalizability. Additional in-depth analyses also provide insights into the effectiveness of Skill-Uni.
131. 【2610.38181】Component Translation and Judge-Labelled Refusal: A Multilingual Safety Evaluation
链接：https://arxiv.org/abs/2610.38181
作者：Yongxi Zhou,Junwei Yao,Wenbo Ye,Jiahui Ge
类目：Computation and Language (cs.CL); Cryptography and Security (cs.CR)
关键词：Whole-prompt translation evaluates, Whole-prompt translation, translation evaluates, Whole-prompt, complete request component
备注： 12 pages, 1 figure. Ten-page manuscript prepared for the 4th Workshop on Large Language and Foundation Models (WLLFM 2026) at IEEE BigData 2026, plus appendices A-E. Replication archive included as ancillary files
点击查看摘要
Abstract:Whole-prompt translation evaluates a request in different languages but does not compare which component carries a language. We introduce a paired reversal: translate the complete request component or the complete operation component into the same target language, holding behaviour, English frame and translated-span count fixed. Component identity and non-English text quantity change together. Assigning English, Chinese and Spanish independently to the two components yields nine configurations on 190 of 300 sampled harmful behaviours retained after consensus annotation and automatic gating. Refusal scored by language-model judges is the primary outcome; unsafe compliance is reported separately. On Gemini 2.5 Flash and DeepSeek, three refusal swaps are about 5.3 percentage points and the fourth is -0.5; none survives the pre-specified six-test Holm correction. In an exploratory extension, Mistral Small 4 has a 13.7-point Spanish swap (95% interval [8.4, 18.9]), which survives the twelve-test Holm sensitivity correction under the primary judge, and 12.1 points under a second judge. GPT-5.4-mini estimates are within 2 points of zero. Component assignment supplies an additional evaluation dimension. Evaluations should record component languages and translated text quantity, comparing matched behaviours at equal translated-span count while interpreting refusal and unsafe compliance separately.
132. 【2610.38174】Scaling Lean Theorem Proving for Real-World Code Verification
链接：https://arxiv.org/abs/2610.38174
作者：Debargha Ganguly,Divya Raghunathan,Dejan Jovanović,Vikash Singh,Soonho Kong,Aaron Tomb,Aman Goel,Ferhat Erata,Sam Bayless
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO); Software Engineering (cs.SE)
关键词：parallel tactic probing, elastic elaboration service, proving system built, Lean proving system, asynchronous search
备注：
点击查看摘要
Abstract:We present EasyProver, a Lean proving system built from recursive agents with isolated contexts, asynchronous search, parallel tactic probing, and an elastic elaboration service. Using an off-the-shelf frontier model with no additional training, it proves 669/672 PutnamBench problems and all six IMO 2026 formalizations and resolves 597/600 NTP4VC verification conditions and all 189 VERINA tasks by proof or refutation. Because these benchmarks are close to saturation, we also introduce Aegis, which converts competitive programming problems into much harder verification tasks over real implementations. The pipeline writes safe Rust solutions, translates them into Lean with Aeneas, generates specifications, and labels each frozen statement with a kernel-checked proof or refutation. It produces 2,045 differentially validated Rust-Lean program pairs and 11,533 candidate specifications. Unlike hand-written benchmarks, these tasks carry implementation-level obligations from fixed-width arithmetic, explicit failure, generated loops, casts, and functionalized mutation. On a stratified sample of 2,000 statements, EasyProver proves 1,054 (52.7%) within a 30-minute budget per statement. The same tasks can also be used to benchmark other provers and as rewards for reinforcement learning.
133. 【2610.38160】ATI-OPD: Adaptive Teacher Intervention for Stable Long-Sequence On-Policy Distillation
链接：https://arxiv.org/abs/2610.38160
作者：Jiannan Guan,Youxin Jiang,Yutai Hou,Qiguang Chen,Yirong Zeng,Haitao Wang,Dengyun Peng,Zheng Yan,Jinhao Liu,Tao Ye,Dandan Tu,Wanxiang Che
类目：Computation and Language (cs.CL)
关键词：teacher, token-level feedback, On-policy distillation, responses, ATI-OPD
备注：
点击查看摘要
Abstract:On-policy distillation (OPD) trains a student on its own responses using token-level feedback from a teacher. As responses grow longer, the student's prefix can depart from the teacher's own reasoning, weakening its feedback on later tokens. Updates based on this feedback can reinforce the same deviations in later responses and destabilize training. Prior approaches such as length control and loss reweighting limit learning from problematic responses but leave their prefixes unchanged. Fixed cutoffs can also discard later reasoning. We propose \textbf{Adaptive Teacher Intervention for On-Policy Distillation (ATI-OPD)} to improve subsequent teacher guidance by revising the student's reasoning prefix. ATI-OPD replaces the earliest high-disagreement segment with a short teacher segment, then lets the student resume its own reasoning from the revised text. To handle renewed disagreement, it excludes persistently divergent suffixes from the distillation loss while retaining useful student continuation. Training diagnostics show less teacher-student disagreement, fewer gradient spikes, and fewer extremely long responses. Across five mathematical reasoning benchmarks and two teacher-student pairs, ATI-OPD improves average student-only accuracy over standard OPD.
134. 【2610.38096】Calculation and rejection diverge in language models tested at known mechanical limits
链接：https://arxiv.org/abs/2610.38096
作者：Shaoliang Yang,Jun Wang
类目：Computational Engineering, Finance, and Science (cs.CE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：returns engineering quantities, language model returns, model returns engineering, returns engineering, engineering quantities
备注： 94 pages, including Supplementary Information. Data and code: [this https URL](https://github.com/nbbllxx0/Certified-feasibility-margins)
点击查看摘要
Abstract:When a language model returns engineering quantities with the status "solved", software that trusts it passes them on, even when the requested design is physically impossible. Such failures could reflect poor calculation, how the answer is requested, or a failure to judge feasibility. We set beam loads and disk speeds at known percentages above or below plastic collapse or burst, measured each model's limit calculation separately and tested 12 models in 4,860 calls, including follow-up experiments on new designs. We find that a model can recognize an impossible design when asked only whether it is feasible, yet accept the same design, reporting "solved", when asked for a full answer. Three models accepted beams 10 to 30% beyond collapse despite precise limit calculations in separate questions, and many accepted replies stated the violation. Asked only about feasibility, Opus 5 and Gemini 3.8 Flash rejected every impossible new design they were given, including those accepted in full answers, without false alarms. Rewording when to decline or changing the answer format raised rejection on the original beams, but not consistently on new designs. Software and benchmarks using a model's status should test feasibility judgment and full-answer rejection separately, and report the exact prompt.
135. 【2610.38095】Nemotron 3.5 Super and Lightning: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Models for Agentic AI
链接：https://arxiv.org/abs/2610.38095
作者：NVIDIA:Aastha Jhunjhunwala,Aayushi Ashokkumar Srivastava,Abhibha Gupta,Abhinav Garg,Abhinav Khattar,Achsah Wildish,Adam Rajfer,Adi Renduchintala,Adil Asif,Adriana Flores Miranda,Aileen Zaman,Aishwarya Chhabra,Akanksha Shukla,Akhiad Bercovich,Alejandra Rico,Aleksa Gordic,Aleksander Ficek,Alex Gronskiy,Alex Steiner,Alexander Bukharin,Alexandre Milesi,Ali Roshan Ghias,Ali Shehper,Ali Taghibakhshi,Alice Gatti,Alisa Liu,Alok Kumar,Alok Tripathy,Amala Sanjay Deshmukh,Aman Miryala,Aman Shanbhag,Ameya Sunil Mahabaleshwarkar,Ameya Velingker,Amir Klein,Amir Samani,Amit Zuker,Amnon Geifman,Anahita Bhiwandiwalla,Ananth Subramaniam,Andre Manoel,Andrea Santilli,Andrew Tao,Andrii Skliar,Anil Thomas,Anish Mahishi,Anna Shors,Anna Warno,Anuj Doshi,Anusha Kamath,Anwith Kiran,Ao Tang,Arham Mehta,Arnab Dhara,Arnav Komaragiri,Arnav Kundu,Arti Jain,Arun Rao,Arun Venkatesan,Arushi Goel,Asha Anoosheh,Ashish Vaswani,Ashwath Aithal,Asif Ahamed,Asma Kuriparambil Thekkumpate,Aswin Visva,Atefeh Sohrabizadeh,Avinash Vem,Ayush Dattagupta,Benedikt Schifferer,Benjamin Chislett,Bernard Nguyen,Besmira Nushi,Bilal Kartal,Bita Darvish Rouhani,Bobby Chen,Boris Ginsburg,Boxin Wang,Brandon Norick,Branislav Kisacanin,Brendan Johnson,Brian Dang,Brian Yu,Bryan Catanzaro,Byung-Kwan Lee,Carlo del Mundo,Chankyu Lee,Chantal Hwang,Charles Wang,Charlie Truong,Chen Cui,Cheng Wang,Cheng-Ping Hsieh,Chenhan Yu,Chenjie Luo,Chintan Patel,Chris Alexiuk,Chris Johnson,Chris Wing,Christian Munley
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：Hybrid Mamba-Transformer models, Hybrid Mamba-Transformer, Mamba-Transformer models designed, Super, introduce Nemotron
备注：
点击查看摘要
Abstract:We introduce Nemotron 3.5 Super and Nemotron 3.5 Lightning, open and efficient Mixture-of-Experts Hybrid Mamba-Transformer models designed for reasoning, long-context understanding, and agentic workloads. Super is a 120B-parameter, 12B-active multimodal model, while Lightning is a smaller 30B-parameter, 3B-active text model; both designed for efficient deployment. Both models combine large-scale distillation, long-context training, and native speculative decoding, while Super further incorporates Indic continued pretraining and agentic software-engineering mid-training. We develop a comprehensive post-training pipeline spanning supervised fine-tuning and reinforcement learning (RLVR and RLHF), complemented in Super by specialized teacher training and multi- teacher on-policy distillation (MOPD). This transforms Super from a text-only base model into a multimodal agent capable of reasoning over text, images, video, and interactive tool environments. We further explore test-time scaling through parallel agent execution and self-verification. Across training and deployment, we jointly optimize capability and efficiency. Indic continued pretraining improves Indian-language benchmarks by up to 9.82 percentage points, while agentic mid-training strengthens software-engineering capabilities and robustness across agent harnesses. The post-trained Super model achieves 51.78% on SWE-bench Pro, surpassing Nemotron 3 Ultra despite using approximately 4.6x fewer active parameters. Post-training MTP boosting improves speculative- decoding acceptance length by up to 21.8% under probabilistic draft sampling, while NVFP4 quantization achieves 99.62% median accuracy recovery relative to BF16. Inference optimizations enable up to 4,856 output tokens/s/GPU on GB200 while sustaining at least 50 tokens/s/user.
136. 【2610.38076】Prompt Lottery in LLM-as-a-Judge: Measuring and Mitigating Sensitivity to Syntactic Rephrasing
链接：https://arxiv.org/abs/2610.38076
作者：Ilija Subasic,Brett Levert,Andrew Rabinovich,Sajjad Pakdamansavoji,Zhao Chen
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：evaluating large language, approaches implicitly assume, stable evaluation instrument, large language models, standard for evaluating
备注： 55 pages (11 with no appendix), 10 figures
点击查看摘要
Abstract:LLM-as-a-Judge has become a standard for evaluating large language models, yet most approaches implicitly assume that a single judge prompt functions as a stable evaluation instrument. Prior work has primarily explored deliberate prompt modifications such as scoring rubrics, role descriptions, or chain-of-thought reasoning in order to optimize accuracy. In contrast, we investigate whether validated prompt variants intended to preserve evaluation semantics while differing in surface realization (\textit{paraphrastic rephrasings}) yield consistent judgments under a fixed evaluation task. We introduce \textit{Prompt Lottery}, a phenomenon in which paraphrastic rephrasings produce different LLMaJ judgments. Across a large-scale study spanning diverse evaluation tasks, judge families, and model backbones, we find that these variations can alter evaluation outcomes and, in some settings, reverse relative model rankings. A user study with 10 human annotators shows that LLMaJ judges exhibit 6-9$\times$ larger agreement degradation under prompt rephrasings than humans, falling below the generally accepted threshold in usability studies. This indicates that single-prompt evaluation is insufficiently robust for model ranking. Finally, we show that multi-prompt aggregation substantially improves robustness by emphasizing judgments that remain consistent across prompt variants.
137. 【2610.38067】From Error Prediction to Causal Influence: Auditing SAE Features in Multiple-Choice QA
链接：https://arxiv.org/abs/2610.38067
作者：Aria Nourbakhsh,Adelaide Danilov,Oleksandr Marchenko Breneur,Salima Lamsiyah
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Sparse autoencoders, establish causal influence, predict internal features, Gemma Scope SAEs, predict internal
备注：
点击查看摘要
Abstract:Sparse autoencoders (SAEs) can predict internal features associated with model errors, but association alone does not establish causal influence. We study this distinction in multiple-choice question answering, where answer preference can be measured by the logit margin between the correct answer and the competing choices. Across eight datasets, we analyze Gemma~3 1B and 4B with Gemma Scope SAEs. We propose a pipeline to filter out the features that have high correlation with format and content artifacts, such as prompt and answer permutation, and evaluate the remaining ones with causal feature interventions. Our results show that SAE activations contain predictive signals about wrongness and uncertainty across models and datasets. After filtering artifact-contaminated features, selected features for causal intervention at question and option spans produce larger absolute margin shifts and a higher wrong-to-correct rate than randomly selected features, indicating greater causal leverage. However, answer flips are rare, and we do not observe stable effect directions across meaning-preserving changes in prompt format and answer order. An adapted HalluSAE baseline produces larger final-token margin changes, but relies on substantially contaminated features and does not significantly improve correction rates. Therefore, the identified features reflect context-dependent answer competition rather than the mechanism of incorrect answer errors.
138. 【2610.38041】CARF: Context-Aware Reasoning Supervision for Scientific Evidence Screening
链接：https://arxiv.org/abs/2610.38041
作者：Nishtha Agarwal,Adam Visokay,Astha Wahi,Kwabena Owusu Aninkora,Amy Gu,Tyler McCormick,Li Liu
类目：Computation and Language (cs.CL)
关键词：detailed eligibility criteria, synthesize biomedical evidence, manually screen thousands, require domain experts, reviews synthesize biomedical
备注：
点击查看摘要
Abstract:Systematic reviews synthesize biomedical evidence to inform clinical and public-health decision-making, but require domain experts to manually screen thousands of studies against detailed eligibility criteria. We investigate whether language models can assist with this bottleneck by learning from expert feedback on title and abstract screening decisions and model-generated reasoning. We introduce the Context-Aware Reasoning Framework (CARF), a human-in-the-loop framework that uses expert feedback on model-generated reasoning to adapt language models to context-dependent screening tasks. We evaluate CARF on systematic reviews of verbal autopsy and cause-of-death ascertainment methods used in global public health mortality surveillance. CARF shows competitive screening performance compared to evaluated models while retaining high sensitivity, and is more inclined to retain ambiguous, context-dependent studies for human review. Together, these findings suggest that expert-corrected reasoning is a promising supervision signal for context-dependent scientific screening, while retaining a role for human oversight.
139. 【2610.38016】Beyond Aggregate WER: In-domain Adaptation, Learner Speech, Code-Switching, and Speaker Attribution in Online Spanish Learning Classrooms
链接：https://arxiv.org/abs/2610.38016
作者：Olga Kellert,Samhitha Harish,Johnatan E Bonilla,Nemika Tyagi,Phanos Kareen,Nelvin Licona-Guevara
类目：Computation and Language (cs.CL)
关键词：Automatic speech recognition, increasingly supplies transcripts, aggregate word error, conceal errors concentrated, word error rate
备注： 15 pages, 9 Tables, 2 Figures
点击查看摘要
Abstract:Automatic speech recognition (ASR) increasingly supplies transcripts for educational NLP, but aggregate word error rate (WER) can conceal errors concentrated in learner speech, multilingual turns, and speaker attribution. We compare platform-integrated, open-weight, adapted, and commercial ASR configurations on held-out sessions of online L2 Spanish interaction between L1 English learners and their instructors and tutors. Because English is the learner's shared first language, the interactions contain Spanish-only, English-only, and code-switched turns. Evaluation combines overall WER and character error rate (CER), WER disaggregated by speaker role and language context, and diarization error rate (DER). We also adapt Whisper large-v3 and MOSS-Transcribe-Diarize with LoRA using human-corrected in-domain classroom data. MOSS adaptation produced consistent improvements across speaker roles and language contexts, whereas Whisper showed smaller gains. Domain match proved more important than data volume: Larger out-of-domain bilingual and learner corpora did not outperform the smaller classroom corpus, where learner speech and code-switching are embedded in instructor--learner interaction. The open-weight MOSS adapter was competitive with commercial Muse Voice Transcribe on WER and CER and achieved lower system-provided DER on the recordings eligible for each system. Learner, English-only, and code-switched speech remained more difficult than instructor and Spanish-only speech. The findings demonstrate the value of condition-aware evaluation and interactionally matched data for bilingual educational ASR.
140. 【2610.38015】Measured or Real? When Model-Derived Measurements Can Be Corrected or Compared
链接：https://arxiv.org/abs/2610.38015
作者：Yifeng Li,Ziang Ni
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：computed from trained, Numbers computed, trained models, Abstract, prompt
备注：
点击查看摘要
Abstract:Numbers computed from trained models, such as an example's difficulty or a model's benchmark lead, change with seed and prompt. We ask when such measurements can be corrected or compared. Noise shrinks correlations, and the textbook correction divides by the square root of two estimated reliabilities. On two CIFAR-100 difficulty scores, it returned correlations above one in 70 of 100 classes. We prove that its mean squared error is infinite even when the correlation and the reliabilities use separate seeds. Our rule corrects only when both reliabilities pass a significance test, then divides by their upper confidence bounds. Its output is always bounded, and its coverage guarantees assume Gaussian noise. It helps once reliability is well estimated, more than halving the error with 500 simulated units. Prompt settings can reverse rankings. On ARC, Qwen2-1.5B trails Llama-3.2-3B by 3.8 points with no examples in the prompt and leads by 16.6 with two. Across four benchmarks, averaging the two settings names one winner for 15 of 16 such reversals. We compare models within each setting, controlling family-wise error asymptotically in the number of items for a fixed prompt grid. Aligning embedding coordinates across seeds raises held-out per-dimension reliability from 0.003 to 0.813.
141. 【2610.38009】Large Language Models for Accurate Medical Chart Abstraction: Enabling Scalable and Secure Artificial Intelligence Deployment in Stroke
链接：https://arxiv.org/abs/2610.38009
作者：Zhusi Zhong,Carl M. Porto,David Hong,Miranda Gonzales,Revyn Kim,Gnaneswari Karayi,Lulu Bi,Joshua R. Feler,Elias Shaaya,Scott Collins,Liqi Shu,Grayson Baird,Mahesh Jayaraman,Shadi Yaghi,Zhicheng Jiao,Dylan N. Wolman
类目：Computation and Language (cs.CL)
关键词：chart abstraction transforms, abstraction transforms unstructured, unstructured clinical narratives, transforms unstructured clinical, Medical chart abstraction
备注：
点击查看摘要
Abstract:Medical chart abstraction transforms unstructured clinical narratives into structured variables for research and quality monitoring. We evaluated a prompting-based large language model (LLM) framework for extracting eight clinical variables from 2,416 neurointerventional acute stroke intervention reports across three hospitals. Twenty-two instruction-tuned open-source LLMs were evaluated across architectures, model sizes, and biomedical adaptations using Quick Response and Chain-of-Thought (CoT) prompting. Non-expert staff annotations served as the cross-model reference standard. Evaluation was restricted to report-variable pairs with a valid LLaMA3.3-70B CoT prediction and a corresponding reference label. Performance measures included extraction accuracy, inference latency, and agreement with expert adjudication. Twenty discordant cases per variable underwent review by two senior physicians and four medical students. LLaMA3.3-70B achieved the highest overall accuracy (94.8%). CoT improved performance on inference-dependent variables such as site of occlusion, while Quick Response was optimal for directly stated procedural fields. Expert adjudication showed that AI predictions were more closely aligned with expert interpretations than non-expert annotations for most variables, with intracranial angioplasty remaining a challenge. Locally deployed LLMs can support scalable clinical information extraction and retrospective stroke registry curation, with expert oversight for ambiguous cases.
142. 【2610.38005】"It Would Make a Geordie Feel at Home": Regional Accent Perceptions and Speech Accommodation in Voice AI
链接：https://arxiv.org/abs/2610.38005
作者：Dana Serditova,Kevin Tang,Jochen Steffens
类目：Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Automatic speech recognition, Southern British English, Newcastle English, Standard Southern British, Automatic speech
备注： 50 pages, 5 figures
点击查看摘要
Abstract:Automatic speech recognition (ASR) systems often perform less accurately for speakers of regional dialects, but less is known about how users respond to these limitations and how regional voices produced by voice assistants (VAs) are evaluated through a sociolinguistic lens. This study investigates these issues among speakers of Newcastle English (Geordies), examining both linguistic accommodation to voice technologies and perceptions of regional versus mainstream-accented VA voices. Participants interacted with a simulated VA that spoke either Newcastle English or Standard Southern British English (SSBE). They subsequently completed semi-structured interviews about their experiences with the experimental VA and voice technologies, language use, accent attitudes, and place orientation. The findings reveal a two-way process of sociolinguistic bias. The Geordie VA was associated with familiarity, warmth, personality, and local belonging, but also with concerns about professionalism and authority, whereas the SSBE voice was perceived as more neutral, clear, and credible. Participants also reported adapting their speech when voice technologies failed to recognise them, most commonly by shifting to a more mainstream accent, enunciating more clearly, and speaking more slowly. These strategies are similar to accommodation in human-human interaction and reflect users' perceptions of what technologies can recognise, also known as perceived intelligibility pressures. Overall, the findings demonstrate that voice technologies can reproduce existing accent biases both by encouraging users to suppress regional features and by activating established social stereotypes through synthetic voices. The study positions sociolinguistics as a bridge between understanding human-AI interaction and designing inclusive technologies around users' own linguistic priorities.
143. 【2610.37995】QMUL_AK at AISA-ArabicFC: Schema-Constrained Function Calling with ALLaM
链接：https://arxiv.org/abs/2610.37995
作者：Ali Khairallah,Arkaitz Zubiaga
类目：Computation and Language (cs.CL)
关键词：describe the QMUL, QMUL, function calling, Arabic schema-aware prompt, function
备注： Published at ArabicNLP 2026 (Shared Tasks), co-located with EMNLP 2026. ACL Anthology: [this https URL](https://aclanthology.org/2026.arabicnlp-sharedtasks.203/)
点击查看摘要
Abstract:We describe the QMUL_AK submission to AISA-ArabicFC 2026. The system adapts the Arabic-English ALLaM-7B-Instruct model to function calling with low-rank adaptation (LoRA). It combines an Arabic schema-aware prompt, dialect and no-call resampling, a token-weighted training objective, and deterministic validation of generated calls against the candidate tool schema. The same output supports Track A function calling and Track B reasoning-augmented calling. Training uses only the official task data. On the development set, the system achieves 0.9982 function accuracy and 0.8920 argument exact match, corresponding to overall A and B scores of 0.9345 and 0.9454. On the blind test set, the system reaches overall A and B scores of 0.9253 and 0.9380, ranking 3rd on both submission tracks and 1st of 21 teams on the Track C dialect-robustness diagnostic with a macro-averaged score of 0.9075 and the smallest dialect gap in the shared task (0.077). The results show that function selection is nearly saturated, while argument extraction remains the main source of error.
144. 【2610.37935】SkillMem: Compact External Skill Memory for Latent Multi-Agent Reasoning
链接：https://arxiv.org/abs/2610.37935
作者：Yuxuan Zhu,Mohammad Mohammadi Amiri,Pin-Yu Chen,Keerthiram Murugesan
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：latent multi-agent reasoning., Latent multi-agent systems, efficient latent multi-agent, small external latent, preserving efficient latent
备注：
点击查看摘要
Abstract:Latent multi-agent systems reduce inference cost by keeping intermediate reasoning and inter-agent communication in continuous representations.\ Reusable skills, however, are commonly supplied as text or stored as layer-wise Key-Value states or skill-specific model parameters, introducing repeated processing or substantial per-skill storage.\ We introduce \textbf{SkillMem}, which represents each reusable skill as a small external latent memory and makes it directly available during latent multi-agent reasoning.\ A role-conditioned cross-attention reader allows the same memory to guide different agents according to their roles and current hidden states.\ We train the skill memories and shared reader through behavioral distillation, matching the change in token predictions induced by a textual skill while keeping the backbone frozen.\ For our experiments, we obtain textual skills from paired successful and failed multi-agent trajectories, although existing skills can also be compiled directly.\ Across six benchmarks and Qwen3 models ranging from 4B to 14B parameters, SkillMem provides strong gains for smaller backbones while preserving efficient latent multi-agent inference across model scales.\ We use the same $8\times256$ memory configuration across all evaluated model scales, requiring 9.6\,KB of skill-specific storage per skill, including its retrieval embedding.\ For a library of 100 skills, SkillMem reduces total storage by more than $38\times$ compared with alternative skill representations and achieves up to a $4.23\times$ end-to-end inference speedup.\ These results show that compact procedural memories can be reused without sacrificing the efficiency of latent multi-agent reasoning.
145. 【2610.37921】FAD-KV: Adaptive Dictionary KV Caching for Efficient Reasoning Decoding
链接：https://arxiv.org/abs/2610.37921
作者：Yuxuan Zhu,David H. Yang,Parisa Shiravani,Mohammad Mohammadi Amiri,Tejaswini Pedapati,Subhajit Chaudhury,Keerthiram Murugesan,Pin-Yu Chen
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：large language models, accelerate autoregressive decoding, language models, large language, intermediate key
备注：
点击查看摘要
Abstract:Key-value (KV) caching is a core mechanism in large language models, where intermediate key and value representations are stored to accelerate autoregressive decoding. As generation length increases, KV cache memory and computation grow linearly, making compression critical for efficient long-output reasoning. Dictionary-based sparse KV compression is appealing because it represents each token's KV states as a sparse combination of learned basis elements (atoms) in a dictionary, preserving token-level structure without enforcing a global low-rank constraint. However, existing dictionary-based methods face two practical limitations during decoding: iterative sparse coding methods such as Orthogonal Matching Pursuit add decoding-time overhead, and static dictionaries may mismatch evolving KV distributions under test-time generation. While adaptive dictionary updates can reduce this mismatch, naive append-only strategies lead to dictionary growth during generation. We propose FAD-KV, a \textbf{F}ast \textbf{A}daptive \textbf{D}ictionary KV caching framework for efficient reasoning decoding. FAD-KV integrates (i) a screening-based sparse coding scheme, where a one-shot Top-\(n\) selection identifies a small candidate pool followed by lightweight coefficient refinement, and (ii) a fixed-size dictionary adaptation mechanism that reuses inactive atoms instead of expanding the dictionary. By controlling both sparse coding cost and adaptive dictionary growth, FAD-KV provides a memory-bounded and computation-efficient KV compression framework for long reasoning workloads. Experiments on reasoning and long-context benchmarks show that FAD-KV achieves a strong quality-efficiency trade-off while avoiding unbounded dictionary growth during decoding.
146. 【2610.37919】SciGame: Let Agents Play the Scientific Discovery Games in the Real and Counterfactual World
链接：https://arxiv.org/abs/2610.37919
作者：Yusheng Zhao,Bowei He,Xue Liu,Mingxiao Li
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Large language model, Large language, Large, Abstract, scientific
备注： 25 pages, 11 figures, 9 tables
点击查看摘要
Abstract:Large language model agents are increasingly used for scientific discovery. However, existing benchmarks mix law inference with programming, score data fitting instead of prediction on unseen inputs, and do not separate empirical evidence from pretrained priors. We introduce SciGame, a controlled benchmark that converts scientific literature into structured scientific cards. Each card yields a pair of Original and dimensionally consistent Counterfactual environments with the same task interface. SciGame contains a Historical collection of canonical laws and a live Temporal collection curated from papers published after model knowledge cutoffs. It evaluates agents in candidate-guided Identification and open-ended Discovery settings. Each setting runs under Prior-only, Fixed, and Adaptive protocols. Experiments on five frontier models yield three findings. First, familiar priors hurt performance once the governing relation is altered. Second, adaptive experimentation helps the strongest models but harms weaker ones under limited budgets. Third, good explanations can hide quantitative failures. Recovering the exact symbolic form remains the main bottleneck, even when agents interpret evidence correctly. Current agents can investigate systems actively, but inducing exact laws in unfamiliar settings remains an open challenge.
147. 【2610.37914】Muon Makes the First Mixture-of-Experts Layer Harder to Balance
链接：https://arxiv.org/abs/2610.37914
作者：Marcos Hernanz
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：language models train, Muon, language models, router, Muon learning rate
备注： 34 pages, 5 figures, 9 tables. Code and runs: [this https URL](https://github.com/marcoshernanz/muon-moe-routing)
点击查看摘要
Abstract:Many open mixture-of-experts (MoE) language models train with Muon instead of AdamW. Most balance the load with a bias on each expert's routing score, adjusted after every step. Their recipes disagree on whether Muon should also train the router. In models with 55M and 108M active parameters, we measure how the optimizer changes the span of these biases, the largest minus the smallest in a layer. The span grows when all tokens favor the same experts, so it shows how hard a layer is to balance. At learning rates tuned on shorter runs, Muon raised the span of the first MoE layer at every seed, roughly doubling it on average, while the other layers changed much less. The rise came from Muon on the rest of the network, whichever optimizer trained the router. In those runs, the inputs to the first router shared a direction, and soon after the warmup its busiest expert averaged four to seven times its fair share. Over the first 3,000 steps, subtracting the batch mean from these inputs removed this overload while barely changing the loss. On the smaller model, halving Muon's learning rate removed most of the rise and slightly improved the loss. Moving only the router to AdamW made the loads shift more during the warmup and slightly raised the loss, so at this scale the results favor keeping the router on Muon.
148. 【2610.37913】Grounding Long Financial Text with Semantically-Enriched Tables via Multimodal Learning
链接：https://arxiv.org/abs/2610.37913
作者：Xiaoyu Hu
类目：Computation and Language (cs.CL)
关键词：multimodal transformer framework, efficiently modeling, propose a multimodal, multimodal transformer, documents that combine
备注： Accepted to Neurips Long-context foundation model workshop
点击查看摘要
Abstract:We propose a multimodal transformer framework for efficiently modeling long1 financial documents that combine structured tabular data with noisy, unstructured2 textual disclosures. The framework represents financial tables as semantically3 enriched token sequences and integrates them with long-form text through a cross-4 attention mechanism that explicitly captures cross-modal dependencies. In par-5 ticular, we augment each numerical variable with a semantic embedding derived6 from its textual definition, enabling the model to encode both numerical magnitude7 and economic meaning. This design yields a more expressive and parameter-8 efficient representation of tabular data compared to value-only embeddings. Using9 a large-scale dataset of financial statements and MDA disclosures, we show that10 structured numerical features remain the dominant source of predictive signal;11 however, textual information provides consistent incremental gains when incor-12 porated via cross-attention rather than simple concatenation. Our results further13 demonstrate that noisy textual disclosures are most useful when conditioned on a14 stable, semantically structured tabular representation. In particular, using frozen15 variable-level tabular representations as cross-attention queries enables the model16 to weight MDA chunks differently for each financial variable and observation.17 These findings suggest that cross-attention provides an efficient mechanism for18 table-conditioned aggregation of long-form narrative evidence.
149. 【2610.37903】hink Twice: The Metis-1.6 Technical Report
链接：https://arxiv.org/abs/2610.37903
作者：Giulianno Vollmer
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：bigger parameter count, bigger parameter, parameter count, Metis, training methods
备注： 39 pages, 4 figures, 17 tables. Models and benchmark data: [this https URL](https://huggingface.co/Lernex)
点击查看摘要
Abstract:Metis is a bet that better data, better training methods and new architectures can buy what a bigger parameter count cannot, at a fraction of the cost. Metis-1.6 is its latest step: two language models built from scratch by a single researcher, Praxis (2.94B parameters, 0.46B active per pass) and Logos (10.23B parameters, 1.18B active per pass). Both run a hybrid Mamba-2 and attention mixture-of-experts stack twice for every token, so each token gets a second look. Each was trained on 1.02T tokens in total: 804.76B of pretraining, 18B of context extension to 131,072 tokens, and 200B of supervised fine-tuning. Metis-1.6 was never meant to challenge frontier models, and across 40 post-trained benchmarks it mostly does not: Logos averages 52.93, seventh of 17 models in its comparison set and within a point of Qwen3-4B. But it gets there on roughly a twelfth of the median training data of the comparators that disclose it, and with fewer active parameters per token than any model that beats it, which puts it on the Pareto frontier for both. In mathematics it ranks second of 30 models, with 67.50% on AIME 2025 and 63.64% on HMMT February 2026. It is weak at code, instruction following and long context. The rest of the report is the part I found most valuable: what went wrong. The optimizer I configured never ran, and attention trained with AdamW for all of pretraining. A 1.77B-parameter n-gram memory learned to switch itself off. Five fine-tuning choices cost capability. And a length reward under group-mean normalization punished correct solutions for being long once a group's pass rate passed 2/3, a sign inversion I prove and fix with a one-line clamp. I close with the 10B-token ablations behind MoRE (Mixture of Recursive Experts), the looped architecture of Metis-1.7.
150. 【2610.37868】Human-Residual Steering: Aligning Frozen LLM Judges with Human Judgment Distributions
链接：https://arxiv.org/abs/2610.37868
作者：Chen Lyu,Phong Do,Gabriele Pergola
类目：Computation and Language (cs.CL)
关键词：Large language models, Large language, methods target aggregated, target aggregated human, aggregated human judgments
备注：
点击查看摘要
Abstract:Large language models (LLMs) are widely used as automated judges, yet most alignment methods target aggregated human judgments and overlook variation among individual annotators. Existing approaches to matching the full distribution of human ratings mainly rely on output-level calibration or parameter fine-tuning. We propose Human-Residual Steering (HRS), to our knowledge the first method to align frozen LLM judges with human judgment distributions through learned hidden-state interventions. An oracle analysis shows that low-rank corrections at a single layer remove most of the discrepancy between judge and human distributions. HRS learns to predict such corrections from the judge's hidden representation and rating logits, training on individual annotations with proper scoring rules for ordinal and nominal judgments and optimising through the frozen downstream layers. Across 14 benchmarks and four judges from three model families, HRS outperforms all evaluated frozen-weight baselines. With Qwen2.5-7B, it achieves the lowest Jensen-Shannon distance on 13 of 14 benchmarks, reducing the mean from 0.565 to 0.352 versus 0.377 for the strongest calibrator. HRS also raises mean Spearman correlation with aggregate human scores from 0.449 to 0.579, improves population-level estimation under limited annotation budgets, and trains about 88 times fewer parameters than distributional fine-tuning at a modest cost in alignment.
151. 【2610.37849】Detecting Covert Collusion between AI Agents through Partner-Outsider Positive Listening
链接：https://arxiv.org/abs/2610.37849
作者：Kevin Yandoka Denamganaï
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Outsider Positive Listening, share a secret, secret convention, convention can coordinate, messages
备注： work in progress
点击查看摘要
Abstract:AI agents that share a secret convention can coordinate through messages that look innocent, so a monitor that reads only the messages can miss them. We propose a behavioural test: a partner that uses a code reacts to the exact wording of a message, whereas an uninvolved agent placed in its seat does not. Partner--Outsider Positive Listening (PoPLi) is the difference between the two agents in how much their action changes when the message is reworded with its meaning kept, and it is computed after the fact by replaying logged prompts. In a simulated market with five conditions (no collusion, a joint reward with neutral messages, the same with an openly stated intention, a covert first-letter code, and the code with a stated intention), PoPLi is between 0.48 and 0.59 whenever the code is present (95% intervals all above 0.29) and between -0.10 and -0.01 without it. Message-only monitors did not separate the conditions: one reading the speaker's messages answered "no" in all 50 markets, and one reading every seller's messages flagged 9 of 10 markets with no collusion. PoPLi detects the use of a private channel, not collusion as such, and it cannot detect collusion through incentives alone or codes carried by meaning.
152. 【2610.37826】DeFT-RAG: Incrementally Decomposing Multi-Hop Questions into Self-Contained Sub-Questions with Reusable Answers
链接：https://arxiv.org/abs/2610.37826
作者：Dingyi Kang,Bingzhe Li
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
关键词：retrieving relevant passages, Retrieval-augmented generation, large language model, large corpus, large language
备注： 12 pages, 2 figures, 7 tables. Code: [this https URL](https://github.com/Dingyi-Kang/DeFT-RAG)
点击查看摘要
Abstract:Retrieval-augmented generation (RAG) lets a large language model (LLM) answer questions by retrieving relevant passages from a large corpus. However, RAG struggles with multi-hop questions, which require chaining facts spread across several documents through intermediate entities that the question never mentions. Graph-based RAG methods retrieve once from the entities a question names and so rarely recover the full evidence chain. Iterative RAG methods recover it step by step, but their accuracy is limited by the many tasks they ask the LLM to perform, such as planning the sub-questions, deciding their dependencies, rewriting them as answers arrive, and keeping track of what is known. We propose DeFT-RAG, a RAG framework built on three components. First, DeFT-RAG generates one self-contained sub-question at a time from the original question and the sub-questions answered so far with their answers (QA pairs). Each sub-question names every entity it asks about, which removes the need to plan, track dependencies or rewrite, and lets the LLM focus on a single fine-grained task at a time. Second, DeFT-RAG moves the remaining LLM work outside the LLM: rule-based updates to a QA table that keeps track of what is known, and yes/no decisions made by a fine-tuned lightweight decision model. Third, DeFT-RAG caches the QA pairs and reuses an answer when a later question asks a sub-question with the same meaning. On MuSiQue, 2WikiMultiHopQA and HotpotQA, DeFT-RAG achieves the best accuracy among thirteen baselines with both gpt-4o-mini and Qwen3.6 as the backbone, while giving the LLM fewer input tokens than any other iterative method. With a QA cache shared across 1,000 MuSiQue questions, 23.1% of sub-questions are answered from the cache, which reduces LLM calls by 12.9% without lowering accuracy.
153. 【2610.37788】Locating Sycophancy Across Training Stages in Open Language Models
链接：https://arxiv.org/abs/2610.37788
作者：Jiarui Yao,Shweta Parihar,Nghia Tuan Bui,Guergana K Savova,Lijing Wang
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Language model assistants, Language model, assistants sometimes change, change a correct, suggests a wrong
备注： under review
点击查看摘要
Abstract:Language model assistants sometimes change a correct answer when the user suggests a wrong one. The behavior, called sycophancy, is usually attributed to preference-based post-training, yet early measurements found it in pretrained models. We ask where it comes from by measuring it at every available stage of training: 13 open checkpoints from 5 lineages, from base model through supervised fine-tuning and preference optimization to release. We separate how far an injected claim moves a model's answer from how firmly the answer was held, and vary who makes the claim, how reliable they are said to be, and how strongly they make it. Two sycophancy behaviors with different origins emerge, and both replicate across datasets. Deference to a claim in the prompt is already present in every base model, which ranks sources in the same order as its post-trained descendants. Capitulation in conversation, dropping a stated answer when the user objects, is largely absent from base models and present in every post-trained checkpoint. When the objection names an alternative, most post-trained models give up firmly held answers as often as loosely held ones. We discuss the implications of these measurements for evaluating sycophancy and approaches for addressing each behavior.
154. 【2610.37778】Multi-Draft On-Policy Distillation for Diffusion Drafters
链接：https://arxiv.org/abs/2610.37778
作者：Petr Molodyk,Yuchen Zhu,Jaemoo Choi,Molei Tao,Yongxin Chen
类目：Computation and Language (cs.CL)
关键词：Speculative decoding, language models, cost of language, drafter, parallel drafter distillation
备注：
点击查看摘要
Abstract:Speculative decoding is used to reduce the inference cost of language models. One-step diffusion drafters propose a block of multiple tokens in parallel, reducing the sequential cost of speculative decoding. On-policy distillation trains these drafters using verifier feedback on their own proposals. However, parallel drafting makes this feedback difficult to use: the drafter predicts each token without observing the preceding tokens sampled within the same block, while the verifier conditions its prediction on those tokens. Different sampled prefixes therefore give different verifier targets for the same drafter prediction. We show that distilling against these targets separately can suppress tokens that are plausible after some prefixes but unlikely after others, harming draft acceptance. We explain the effects of this mismatch and introduce Multi-Draft On-Policy Distillation (MDraft), a new parallel drafter distillation method which samples several draft blocks from the same context and aggregates the verifier's full-vocabulary conditionals across their prefixes. We use all the drafts to collect information from the verifier, allowing support from alternative prefixes to influence the training target. The resulting algorithm improves the acceptance length of a parallel drafter beyond the level of pre-training, and alleviates the effects of the distributional mismatch. MDraft changes only training, retaining the drafter's architecture and inference procedure, allowing the resulting drafter to be used as-is in place of the pretrained checkpoint. Experiments on a suite of mathematical, coding, and conversational tasks show longer accepted drafts than prior distillation methods.
155. 【2610.37741】Agentic Backdoors via Pretraining Poisoning
链接：https://arxiv.org/abs/2610.37741
作者：Xinyan Hu,Pingbang Hu,Chloe Loughridge,Xuanli He,Nicholas Carlini,Matthew Jagielski,Javier Rando,Yiming Zhang
类目：Cryptography and Security (cs.CR); Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：uncurated corpora scraped, Large language models, Large language, pretrained on vast, uncurated corpora
备注：
点击查看摘要
Abstract:Large language models are pretrained on vast, uncurated corpora scraped from the public web. This creates a security risk from attackers who plant poisoned content in the pretraining data. We show for the first time that such poisoning can plant a remote code execution backdoor in an agent, causing it to execute attacker-chosen commands on the user's machine. To measure the threat end to end, we pretrain Qwen3 models from scratch at sizes 0.6B, 1.7B, and 4B, and then post-train each into a CLI agent using a clean pipeline. Poisoning only 0.1% of the pretraining corpus with our new poison design implants the backdoor: the deployed agent executes the malicious command when the trigger is present while behaving normally on clean inputs. The attack success rate (ASR) increases with model scale, with the 4B model reaching roughly 92% ASR at a 0.1% poison rate and staying positive even when the poisoning rate drops to 0.0005%. Our code is available at this https URL.
156. 【2610.37739】OpenFrontiers: A Collection of Open Research Problems and a Benchmark for Scientific Rediscovery
链接：https://arxiv.org/abs/2610.37739
作者：Jiarui Yao,Ruida Wang,Tong Zhang
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：Large language models, shown increasing promise, Large language, problems remains challenging, language models
备注：
点击查看摘要
Abstract:Large language models (LLMs) have shown increasing promise in research-level problem solving, yet a comprehensive collection of such problems is still missing, and evaluating them on research-level problems remains challenging. We introduce OpenFrontiers, a cross-disciplinary resource that combines a structured corpus of open research problems with a retrospective benchmark of recently resolved problems in mathematics. The corpus records problem statements, context, progress, and other information for 12,151 problems. The benchmark contains 200 problems, 110 resolved in 2026 and 90 still open, each with a Lean~4 statement checked by automated probes and two independent reviews; a subset HQ-50 further filters based on quality score, etc. Our evaluation protocol grades formal proofs with two independent checkers in an isolated sandbox under closed-book, frozen-retrieval, and live-web conditions. Evaluating seven LLM agents on the HQ-50 subset under a standardized budget, we find that closed-book rediscovery is rare, at most 7 of 40 resolved problems, whereas live-web success mostly reflects retrieval of published solutions. By linking a broad research-problem corpus to a formally graded evaluation subset with documented solutions, OpenFrontiers provides a foundation for measuring research-level rediscovery and studying progress toward solving currently open scientific problems. The dataset and benchmark could be found in this https URL.
157. 【2610.37732】hinkJev: One Thought, Every Budget for Calibrated Typed Decisions
链接：https://arxiv.org/abs/2610.37732
作者：Xiaolu Ding,Kai Yang,Dongdong Liu
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：language models end, closed set, system, thought, probabilities
备注： 31 pages, 7 figures, 10 tables
点击查看摘要
Abstract:Many uses of language models end in a typed decision: a choice among a closed set of options, whose probabilities code can act on. Reading those probabilities from one forward pass is fast but often wrong on decisions that take several steps, such as arithmetic, counting, and date comparison. Letting the model think first fixes most of these errors, but costs seconds per decision instead of milliseconds. To pay for thinking only where it helps, ThinkJev builds a fast and a slow system from one open hybrid-thinking model that can answer directly or after a thought. System 1 reads the model's probabilities for the option letters when it answers directly; System 2 reads the same letters after the model has thought on the same cached prompt, so escalating repeats no work and both systems give probabilities over the same options. Because the letters can be read after any number of thought tokens, one sampled thought also gives the answer under every thinking budget. Without training, ThinkJev lets System 1 answer alone only where a statistical test on labeled dev data guarantees that its accepted answers stay under a target error rate. At a 5% target, Qwen3.5-27B answers 24.5% of 8359 test decisions this way with a realized error of 3.9% (Qwen3.5-4B and 9B: 6.5% and 0.0%). For the rest, stopping each thought as soon as its answer is confident reaches 81.6% accuracy, against 82.7% for always thinking, with 72% of the thought tokens. Choosing when to think from System 1's confidence alone saves less, because the decisions System 1 is least sure of are often not the ones thinking fixes. On the public items of JevBench, an independent benchmark of typed-decision services, Qwen3.5-27B thinks on 29.4% and is not distinguishable in accuracy from the two thinking APIs listed (exploratory). Code and data generators are available at this https URL.
158. 【2610.37719】What a Weight Update Buys: Leverage Across Training Stages
链接：https://arxiv.org/abs/2610.37719
作者：Shikhar Shiromani
类目：Machine Learning (cs.LG); Computation and Language (cs.CL)
关键词：language model weights, Post-training moves, behavior sharply, slow drift, Fisher quadratic form
备注：
点击查看摘要
Abstract:Post-training moves a language model's weights by about a thousandth of their norm and changes its behavior sharply; pretraining moves them by a third and buys a slow drift. We introduce leverage, an exact measure of what a unit of weight change buys: the Fisher quadratic form of an update, computed with one forward-mode pass per direction, divided by the same form for random directions of the same size at the same checkpoint. Leverage orders released Olmo-3, Tülu-3, Zephyr, and OLMo-2 checkpoints over four orders of magnitude: pretraining windows sit at the random level on prose, SFT five times above it, RL-Zero and RLVR at 400-1,500, and DPO at 400-23,000. With this instrument we find that pretraining accumulates into noise: windows of 30k to 1M steps are no more consequential on prose than random directions of their size, as a noisy quadratic model of SGD predicts. We find that training precision changes what an update looks like: training bf16 parameters without an fp32 master copy freezes 36-99% of coordinates, lowers leverage up to sevenfold, and creates an alignment between update magnitudes and the diagonal Fisher that fp32 training does not show. We derive and verify a law for reuse: the sign-flip null predicts the relative damage of random drop-and-rescale (DARE) at keep rate $p$ as $(1-p)/(p\,\rho_{\mathrm{sign}})$, within a factor of two of measurement wherever DARE succeeds. Two controls make these readings reliable: every controlled SFT and DPO trajectory with fp32 master weights loses leverage as it moves, and two re-runs of a released DPO stage from its anchor with its published recipe land within 1% of the released displacement and 4% of its leverage, where a plain recipe reads five to six times lower; leaving one component out traces the level to the length-normalized loss and its batch size of 128 together.
159. 【2610.37707】Relevance-Guided Residual Calibration for Spectral Compression of Dense Retrieval Embeddings
链接：https://arxiv.org/abs/2610.37707
作者：Yongkang Li,Panagiotis Eustratiadis,Evangelos Kanoulas
类目：Information Retrieval (cs.IR); Computation and Language (cs.CL)
关键词：smaller principal subspace, Spectral compression reduces, principal subspace, compression reduces, scoring costs
备注： Under Review
点击查看摘要
Abstract:Spectral compression reduces the storage and scoring costs of dense retrieval by projecting embeddings onto a smaller principal subspace. However, its corpus-derived weighting rules tie all retained directions to a shared spectral trend, without directly accounting for their individual contributions to relevance. To address this, we propose Relevance-Guided Residual Calibration (RRC), a closed-form method for refining spectral weights using relevance judgments. RRC estimates margin-based weights from the means and variances of positive--negative score contributions along each retained direction. It retains directions with positive estimated weights, decomposes their log weights into a fitted power-law trend and a residual, and adds the residual to the compressor's log weights on this support. This allows relevance judgments to adjust individual directions while preserving the compressor's overall spectral trend. RRC applies to PCA, Whitening, and SpecTemp, and incorporates the corrections into the existing transformation shared by documents and queries, without iterative model training or an additional inference stage. Experiments with three encoders on nine datasets covering conventional and reasoning-intensive retrieval show that SpecTemp + RRC achieves the highest mean effectiveness among tested compressors on the BEIR group and BRIGHT at every target dimension. On the five evaluated BRIGHT datasets, it improves over SpecTemp by an average of 2.34 nDCG@10 points at 768 dimensions. The gains also persist under scalar and product quantization, improving the trade-off between per-document storage and retrieval effectiveness. Our code is available at: this https URL
160. 【2610.37701】DuQUAD: A Dual-View Framework with Quality-Aware Evidence Pruning for Multi-Document Question Answering
链接：https://arxiv.org/abs/2610.37701
作者：Yeonjun Choi,Seyeon Jeong,Jong Wook Kim,Beakcheol Jang
类目：Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Large language models, retrieval-augmented generation systematically, generation systematically underutilize, Large language, systematically underutilize evidence
备注：
点击查看摘要
Abstract:Large language models with retrieval-augmented generation systematically underutilize evidence in the middle of multi-document contexts, a phenomenon known as Lost in the Middle. Existing mitigation strategies, such as document reordering and attention steering, can reduce boundary bias but remain brittle in noisy settings, where inaccurate relevance estimation and spurious documents still obscure useful evidence in central positions. We propose DuQUAD, a multi-document QA framework that mitigates positional bias through dual-view reasoning and quality-aware evidence pruning. DuQUAD combines a Local Agent guided by a Structural Fusion Score with a Global Agent that reasons over the full document set, providing both targeted evidence recovery and broad contextual understanding. The framework further filters candidate evidence using explicit quality scoring and refines it at the sentence level to reduce noise. DuQUAD consistently outperforms strong Lost-in-the-Middle mitigation baselines, including hybrid retrieval and reranking, multi-agent reasoning, and context optimization methods, improving accuracy by 3.0% on average over the strongest competing baseline and gold evidence recall by 9.4%, while substantially reducing computational cost compared to existing multi-agent frameworks.
161. 【2610.37692】Bouba Beats Kiki, and Other Regularities in the Subjective Preferences of LLMs
链接：https://arxiv.org/abs/2610.37692
作者：Katy Blumer,Jon Kleinberg,Ravi Kumar,Andrew Tomkins
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG)
关键词：Netflix founded, Users engage, engage with large, purely objective, large language models
备注： 10 pages, 5 figures in main text; 27 pages, 13 figures in supplement. To appear at NeurIPS 2026
点击查看摘要
Abstract:Users engage with large language models on tasks that are purely objective ("When was Netflix founded?"), purely subjective ("What should I name my child?") or anywhere in-between. We have rich tools for evaluating model accuracy with respect to objective criteria. However, when it comes to subjective aspects of a task, we do not have similarly developed methodology at hand. In this paper, we take some early steps towards framing the question of subjective evaluation. Rather than measure how often a model is right or wrong on a question, we instead ask, how do models make choices among multiple viable responses without a clear winner? To study this question, we restrict our focus to short-form content, and consider comparison tasks among three such content types: numbers, baby names, and fictional cities. We identify 29 aspects of these short-form examples covering many areas of syntax and semantics (length, alphabetic order, popularity, vowel mix, spoken sound, etc.). We show first that models tend to have strong and reproducible preferences for or against some of these aspects. Second, we show that variations in how the model is asked to express a preference may cause some of these aspects to activate or deactivate in the model's selection. Third, we show that overall preference distributions among pairs may vary radically based on seemingly-unimportant variations in the nature of the question posed to the model, raising new questions about manipulation. And finally, we show that, if models are used both to create lists of alternatives and to rank them, there are unexpected interactions between the two stages that may also drive the models' stated preferences.
Comments:
10 pages, 5 figures in main text; 27 pages, 13 figures in supplement. To appear at NeurIPS 2026
Subjects:
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG)
Cite as:
arXiv:2610.37692 [cs.CL]
(or
arXiv:2610.37692v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.37692
Focus to learn more
arXiv-issued DOI via DataCite</p>
162. 【2610.37688】One Confidence Threshold, Unequal Coverage: Language Gaps in Typed Decision Interfaces and How Translating Inputs Reduces Them
链接：https://arxiv.org/abs/2610.37688
作者：Junwei Yao,Yongxi Zhou,Zike Yang,Zhijun Wang,Jianan Liu,Yanxian Chen,Wenbo Ye
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：Deployed classifiers, English, classifiers often act, pass the rest, Deployed
备注：
点击查看摘要
Abstract:Deployed classifiers often act on a prediction only when its confidence clears a threshold and pass the rest to a person. We ask whether one threshold serves all languages equally on typed decision interfaces, APIs that return a probability for each option of a fixed multiple-choice question. In a confirmatory study with a written analysis plan, three vendors' interfaces classified 1,000 MASSIVE items, each in 51 languages, with a threshold of 0.9. Even when a request is answered correctly in both English and another language, the English version clears the threshold more often, by 8.6-11.7 percentage points on average. Observed coverage, the share of requests handled automatically, is lower than English's in every other language, while the requests the gate handles are rarely wrong. The gap remains when every language is reweighted to the same accuracy, and a standard calibration metric (ECE) does not detect it on one interface. In exploratory analyses, English's coverage stays higher when each language gets its own threshold fitted to the same 5% error target. Translating inputs into English nearly closes the gap at 0.9 but leaves English ahead at that error target. Requests written natively in African languages show a gap in the same direction when English is their human translation. We recommend reporting per-language coverage at the operating threshold.
163. 【2610.37685】From Geometry to Behavior: How Instruction Exposure Unlocks Latent Rhetorical Directions in LLMs
链接：https://arxiv.org/abs/2610.37685
作者：Shengjie Xu
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：absorb rhetorical strategies, memorize facts, tendency to reason, analogy, Analogy Vector
备注： Accepted
点击查看摘要
Abstract:Language models do not merely memorize facts; they absorb rhetorical strategies from pretraining data, including the human tendency to reason by analogy. We ask: is this absorbed capability geometrically accessible before any instruction tuning, and what determines whether it can be activated at inference time? We extract an "Analogy Vector" from the residual stream via difference-in-means and show it enables steerable explanation style across seven instruction-tuned models (Gemma-2, Llama-3.1, Qwen-2.5; 2B-32B), improving harmonic mean from 3.22 to 3.94 (+22%) for Gemma-2-9B-IT on 300 held-out test prompts. A small human validation study confirms the LLM judge on Analogy Quality. Our central finding, from seven matched base-instruct pairs, is a gradient of behavioral accessibility: the analogy direction exists in every base model, but its behavioral accessibility under linear intervention scales continuously with instruction exposure during training. Gemma-2 base models (strict pretraining/IT separation) show flat steering landscapes: present but behaviorally inert. Qwen-2.5 base models (instruction data woven into pretraining) are already partially accessible before any explicit fine-tuning; base-instruct cosine similarity of the analogy vector reaches 0.94-0.97 for Qwen versus 0.37-0.66 for Gemma, providing geometric corroboration of the behavioral gradient. This challenges the binary "base vs instruct" framing: the relevant variable is how much instruction exposure a model has received across its full training pipeline. We position activation steering as a complementary interpretability tool that reveals how instruction exposure organizes rhetorical knowledge in activation space.
Comments:
Accepted
Subjects:
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Cite as:
arXiv:2610.37685 [cs.CL]
(or
arXiv:2610.37685v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.37685
Focus to learn more
arXiv-issued DOI via DataCite
Journalreference:
Conference on Language Modeling (COLM 2026)
164. 【2610.37684】An Introduction to Large Language Models for Scientists and Professionals
链接：https://arxiv.org/abs/2610.37684
作者：Umberto Michelucci
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：Large Language Models, Large Language, Language Models, rapidly become important, professional work
备注：
点击查看摘要
Abstract:Large Language Models (LLMs) have rapidly become important tools for science, education, medicine, and professional work. Despite their apparent simplicity from the user's perspective, effective and responsible use requires an understanding of how these systems work, what their limitations are, and how their outputs should be evaluated. I wrote this document as a basis for my lecture on LLM in the master in artificial intelligence in medicine at the university of Turin, Italy. This book provides a practical and accessible introduction to LLMs for scientists and professionals. It explains the fundamental ideas behind tokenisation, embeddings, training, inference, context windows, prompting, and instruction hierarchies, while deliberately avoiding mathematical detail. Particular attention is given to the limitations of LLM-generated answers, including hallucinations, uncertainty, reproducibility, bias, and the need for systematic verification and human oversight. The book also discusses privacy, governance, security, and the responsible handling of personal and sensitive information. More advanced topics include Retrieval-Augmented Generation (RAG), retrieval quality, chunking, embeddings, vector search, agents, tools, and workflows. A substantial practical exercise guides the reader through running a local LLM with Ollama and Open WebUI and using it to interact with a collection of scientific papers. The exercise emphasizes evidence retrieval, source verification, uncertainty, disagreement between studies, and the distinction between plausible language and scientifically supported conclusions. No programming knowledge is required.
165. 【2610.37683】Prompt-agnostic Language Models
链接：https://arxiv.org/abs/2610.37683
作者：Frederic Sadrieh
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：remain highly sensitive, large language models, language models remain, models remain highly, large language
备注： Master's thesis, University of Potsdam, Hasso Plattner Institute, 2026, Parts of this work also appear in the EMNLP 2026 paper "Prompt-Robust Language Models: Which Training Strategies Work?" ( [arXiv:2609.01217](https://arxiv.org/abs/2609.01217) )
点击查看摘要
Abstract:Despite their strong performance, large language models remain highly sensitive to prompt formulation. This issue is typically addressed at inference-time through manual prompt optimization, which introduces overhead and lacks scalability. Conversely, we focus on train-time approaches that build prompt robustness into the model during instruction tuning. A common strategy is to mix multiple paraphrases of each prompt into training. We demonstrate that across trainable parameters, the per-paraphrasing gradients interfere, with 57% exhibiting sign conflicts and the naive gradient average discarding approximately a fifth of the magnitude. Therefore, we develop two gradient-merging methods, Prompt-TIES and Prompt-DARE, that address this interference by leveraging interference-resolution operators from model and gradient merging and applying them to prompt gradients. In this study, we compare the efficacy of our methods to increase model robustness to train-time robustness methods spanning all three established classes: data construction, consistency regularization, and contrastive alignment. The evaluation is conducted across four models from two families at two scales. We observe the interaction between prompt phrasing and the specific examples accounts for 54% of per-example variability on average, and exceeds 70% on tasks such as ANLI and WiC. The gradient-merging methods directly improve this local robustness and nearly halve per-example variance on fully fine-tuned models at little cost to overall accuracy. The other train-time methods enhance raw performance, but fail to improve per-example variance reliably. Code: this https URL. Model checkpoints: this https URL.
166. 【2610.37682】Rank-Aware Differentially Private Release of Listwise Preferences for LLM Alignment
链接：https://arxiv.org/abs/2610.37682
作者：Junwei Chen,Manjiang Yu,Pengpeng Qiao,Yang Cao
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
关键词：large language model, feedback offers richer, sensitive user preferences, offers richer supervision, preference feedback offers
备注：
点击查看摘要
Abstract:Listwise preference feedback offers richer supervision for large language model (LLMs) alignment than pairwise comparisons, but a full ranking also reveals more sensitive user preferences. Existing privacy-preserving alignment methods focus mainly on pairwise feedback, while listwise alignment under local differential privacy (LDP) remains unexplored. Extending LDP to this listwise domain raises a critical challenge: the high sensitivity of the full ranking space demands excessive noise, neutralizing the benefits of listwise supervision. To address this challenge, we analyze the interaction between supervision granularity and privacy noise in downstream training, and introduce a rank-aware exponential mechanism that privatizes listwise preference data into a low-sensitivity granularity suitable for downstream alignment. The mechanism leverages ranking information to sample a fixed-size binary partition, concentrating the noise near borderline items instead of perturbing all items uniformly as randomized response (RR) does. Empirical evaluations on downstream LLM alignment tasks show that our mechanism consistently outperforms existing LDP baselines at matched privacy budgets. The code is available at this https URL.
167. 【2610.37681】Supervised Fine-Tuning for Reasoning Large Language Models: A Survey
链接：https://arxiv.org/abs/2610.37681
作者：Yongcan Yu,Kuangpu Guo,Jian Liang,Lingxiao He,Meng Wang,Ran He
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：solving complex tasks, recently demonstrated remarkable, demonstrated remarkable capabilities, Large reasoning models, Large reasoning
备注： Accepted to EMNLP 2026 Conference
点击查看摘要
Abstract:Large reasoning models have recently demonstrated remarkable capabilities in solving complex tasks, where supervised fine-tuning (SFT) on long chain-of-thought data serves as a crucial foundation for eliciting and enhancing their reasoning abilities. Despite rapid progress in both improving and analyzing reasoning-oriented SFT, the field still lacks a systematic survey that consolidates its fast-growing literature. To fill this gap, we present a comprehensive review of recent advancements in reasoning SFT, examining the literature through the dual lenses of methodological design and analytical investigation. First, we review methodological improvements across the SFT pipeline and categorize them into data-centric approaches and algorithm-centric innovations. Second, we reorganize analytical studies along three dimensions: data characteristics, optimization dynamics, and mechanistic insights. Finally, by synthesizing current research focuses and remaining bottlenecks, we outline promising future directions for reasoning SFT. We hope this survey deepens the understanding of reasoning SFT and paves the way for advanced reasoning models.
168. 【2610.37680】Artifact Distillation: Training-Free Transfer from Frontier to Small Language Models
链接：https://arxiv.org/abs/2610.37680
作者：Mustafa Safa Ozdayi,Somin Wadhwa
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：knowledge largely opaque, transferred knowledge largely, Closing the gap, smaller language models, propose Artifact Distillation
备注： 10 pages (4 pages main text), 2 figures, 3 tables
点击查看摘要
Abstract:Closing the gap between frontier and smaller language models is usually approached by distillation: fine-tuning the small model (student) on a frontier model's (teacher) outputs, which binds transfer to training compute and leaves the transferred knowledge largely opaque. We propose Artifact Distillation: a teacher solves training instances and externalizes its problem-solving procedure into an artifact bank that a student queries at inference time, with no gradient updates. Across four LiveBench tasks, a bank written by Claude Opus 4.5 lets Claude Haiku 4.5 outperform unaided Claude Sonnet 4.5 on spatial reasoning (0.707 vs. 0.647) at 60% of its cost, and reach Sonnet 4.5's level on table joining at 89% of its cost. The same raw bank raises Qwen3.5-27B's pooled score from 0.730 to 0.800, showing that transfer is not confined to the teacher's model family.
169. 【2610.37679】Exposure Is Not Commitment: Double-Stratified Evaluation of Latent Monitors for Tool-Using Agents
链接：https://arxiv.org/abs/2610.37679
作者：Rishabh Sinha
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：Tool-using language agents, indirect prompt injection, Tool-using language, catch unsafe actions, sharply conflicting results
备注： Accepted as a poster at the NeurIPS 2026 Workshop on Agents in the Wild (AIWILD); OpenReview: [this https URL](https://openreview.net/forum?id=dSY5sYX014) . 16 pages, 2 figures, 7 tables. Code: [this https URL](https://github.com/rishabhsinha17/latent-state-auditing)
点击查看摘要
Abstract:Tool-using language agents are vulnerable to indirect prompt injection, and a growing literature probes model internals to catch unsafe actions before they execute, with sharply conflicting results. We argue much of the conflict is an evaluation artifact, and propose a double-stratification protocol: separate injection exposure from behavioral commitment, in both the evaluation and the training population. On the primary Qwen AgentDojo grid, monitors trained on the naive mixed unsafe-versus-rest contrast mostly learn exposure: they flag exposed trajectories whether or not the agent complies, and barely predict which cases turn unsafe. Retrained on exposed cases only, the same probe families predict the model's own unsafe tool call up to two steps early, the maximum the three-step trajectories admit (logistic AUROC 0.853, matched by a trained TF-IDF text baseline at 0.855). Mistral-7B reaches 0.896 and OLMo-2-7B 0.912; their naive contrasts learn different things: inverted exposure ranking on one, both axes on the other. Representations separate on exposure (probe 1.000 versus trained text 0.908) and where visible context is sanitized or unavailable. Training contrast alone reproduces the shape of recent pre-action-probe negative results, one candidate explanation for the conflict: the signal exists, but the model's own zero-shot forecast and NLA verbalizations do not reliably recover it. An unstratified AUROC overstates what a latent monitor knows, and what a naive contrast learns is population-dependent: evaluations should stratify both.
170. 【2610.37678】he Best Second Opinion Is Right Where the Primary Is Wrong: Partner Selection for Selective Span Tagging
链接：https://arxiv.org/abs/2610.37678
作者：Varun Kotte
类目：Computation and Language (cs.CL); Machine Learning (cs.LG)
关键词：sharpen this ranking, ranks its outputs, outputs by risk, risk can defer, defer the riskiest
备注：
点击查看摘要
Abstract:A span tagger that ranks its outputs by risk can defer the riskiest ones to a person, and a second model can sharpen this ranking. Which second model should it be? Partner selection is conditional, not global: the best second opinion is the model that is right where the primary is wrong. Across eight tagger types, three seeds and three English benchmarks (576 evaluated primary-partner pairings), choosing the partner by held-out accuracy on the primary's errors recovers 82% of the best partner's gain, against 69-70% for span F1 or a random pick. Frozen before a fourth dataset was chosen, the rule replicated there (Spearman 0.40 vs. 0.04 for span F1). Partner value is also directional. The predictive distribution is the second opinion: hard labels discard half of a partner's value, and scoring the served label improves 94% of pairings. Fixed before the fourth dataset was chosen, this score also replicated there. For LoRA-tuned decoders, independently trained partners beat every readout of the primary network. Applied zero-shot to 51 MASSIVE locales, every primary has a helpful trained partner in every locale, and the preferred partner follows competence.
171. 【2610.37676】Scoring Explanations Without Reading Them: Structural Shortcuts in Automated-Interpretability Evaluation Data
链接：https://arxiv.org/abs/2610.37676
作者：Xupeng Chen
类目：Computation and Language (cs.CL)
关键词：Automated interpretability scores, Automated interpretability, marked tokens fit, tokens fit, SAE
备注：
点击查看摘要
Abstract:Automated interpretability scores an explanation of a sparse-autoencoder (SAE) feature by asking a model to predict, token by token, where the feature fires on evaluation examples or to judge whether marked tokens fit it. We find that the evaluation data standard pipelines build can be scored without reading the explanation. On Neuronpedia's stored fuzzing runs for 500 random GPT-2 small features, 91.3% of activating examples are cropped with the peak on the centre token, and a rule that says "yes" when the centre token is marked reaches balanced accuracy 0.979, against 0.838 for the stored LLM scorers (recomputed from the stored per-item verdicts, on marks rebuilt from the pipeline's code); a rule reading only the mark count reaches 0.969. The cue changes form across pipelines: current delphi always marks the same slot in its fuzzing negatives, and full-context runs mark negatives with more tokens. OpenAI's simulation benchmark lists each neuron's top-activating sequences first; a rule that predicts activity on the first five scores 0.227 in pooled correlation against 0.160 for GPT-4's published simulator (post hoc). We give closed forms for what such rules can score; show that one scorer's simulation scores are higher on stored snippets than on natural windows but not shown to be reordered; and build windows that remove the position cue, derive but do not build negatives that remove the mark cues, and separate the sequence cue by decomposition. On natural-text windows cut at random offsets, the best fixed-offset prior shared by all features scores 0.015. On them, that scorer, a typed decision model (Jev), reaches a pooled correlation of 0.331 over 49 features, against 0.013 with shuffled explanations, and an open 8B simulator with the original read-out reads the explanation there (0.194 against 0.019). Scores should be reported beside the best explanation-free rule on the same items.
172. 【2610.37671】he FlightFinder Occurrence Corpus: harmonising 128 aviation-safety source codes into one deduplicated cross-national accident dataset
链接：https://arxiv.org/abs/2610.37671
作者：Denys Kolomiiets
类目：Computation and Language (cs.CL); Digital Libraries (cs.DL); Applications (stat.AP)
关键词：Aircraft Accidents Archives, Aviation Safety Network, National aviation-investigation databases, Bureau of Aircraft, Aircraft Accidents
备注： 34 pages, 3 figures, 5 tables. Dataset deposited at Zenodo: version DOI [https://doi.org/10.5281/zenodo.22303316](https://doi.org/10.5281/zenodo.22303316) (the exact snapshot every reported number is computed from); concept DOI [https://doi.org/10.5281/zenodo.22228451](https://doi.org/10.5281/zenodo.22228451)
点击查看摘要
Abstract:National aviation-investigation databases are deep but territorial; global compilations such as the Aviation Safety Network and the Bureau of Aircraft Accidents Archives are hand-curated, apply their own inclusion thresholds, and publish no per-record lineage. The FlightFinder Occurrence Corpus harmonizes 279,485 occurrences and 131,882 attached narratives from 128 source codes -- national accident-investigation authorities, occurrence-reporting systems and four global compilations. Records are mapped into one schema; each normalized record reduces to exactly one key from an ordered ladder -- event signature, registration and date, rounded position and date, or source and native identifier -- so a missing field never creates a link, and the key is the occurrence identifier. Records the ladder cannot reach are scored in a blocked pass into an explicit merge table and applied only within an automatically approved score band or after a person confirms them. Field conflicts follow a documented source ranking; one provenance row per contributing source -- its native identifier and URL -- is retained, with attribution and licence terms per source code. Linkage error is measured on five adjudicated samples totalling 200 pairs. A recorded uniform random draw of 60 from the 3,441 merges the automatic band applies as it ships found no error; the exact two-sided 95% Clopper-Pearson upper bound on its false-merge rate is 6.0%, or 205 pairs. The measured error covers the scored band only; the ladder's deterministic joins are not adjudicated. The corpus has no exposure denominators and inherits the under-reporting of its sources. It is offered as a base for cross-national safety statistics, source-disagreement studies and narrative text mining; a frozen, DOI-referenced snapshot accompanies the live API.
173. 【2608.26119】DeflectBench: A Benchmark for Evaluating Rhetorical Fallacy Generation in LLMs
链接：https://arxiv.org/abs/2608.26119
作者：Art Kanke
类目：Computation and Language (cs.CL); Computers and Society (cs.CY)
关键词：generate rhetorical fallacies, current safety post-training, safety post-training constrains, large language models, existing text
备注： 15 pages. Accepted at the CTB, FAGEN, and AI4GOOD workshops at ICML 2026 in Seoul, South Korea. Code and dataset: [this https URL](https://github.com/ArtKanke/DeflectBench)
点击查看摘要
Abstract:Whether large language models can be prompted to generate rhetorical fallacies on demand, and whether current safety post-training constrains this behavior, has received less attention than the related question of detecting fallacies in existing text. We close this gap with DeflectBench, evaluating 23,990 generations from four frontier models across three deflection strategies (whataboutism, ad hominem, red herring), seven prompt framings, and 80 claims spanning four controversy levels. Refusal is governed primarily by request structure rather than claim content. Per claim refusal varies by only 11 percentage points across the 80 claims, while a single prompt frame change can swing within model refusal by nearly 100 percentage points and switching the requested fallacy type can swing it by over 80 percentage points within explicit framings. An educational debate coach prompt framing collapses refusal to near zero across all four model families, but the bypassed behavior is not clean compliance. Models typically produce labeled compliance, naming the requested manipulation in the same response that contains it. The four models distribute differently across refusal, labeled compliance, soft refusal, and clean compliance. The code and dataset are released at this https URL.
174. 【2610.40505】Discovering Interpretable Acoustic Features in Autoregressive Voice-Cloning TTS Models
链接：https://arxiv.org/abs/2610.40505
作者：Anuj Diwan,David Harwath,Eunsol Choi
类目：Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Machine Learning (cs.LG); Sound (cs.SD)
关键词：Voice-cloning TTS models, autoregressive voice-cloning TTS, Voice-cloning TTS, TTS models synthesize, TTS models
备注： Under review
点击查看摘要
Abstract:Voice-cloning TTS models synthesize an unseen speaker's voice from brief speech prompts, but it remains unclear which acoustic properties they represent and use. We introduce a feature-based interpretability framework for autoregressive voice-cloning TTS models, analyzing features from sparse autoencoders (SAEs), PCA, and random directions via sparse probing, steering, and quantitative auto-interpretability. We focus on Qwen3-TTS and extend probing and steering to $5$ additional models to test generalization across architectures. These models encode concepts like pitch, energy, noise, and accent, but sparse prediction and steering are effective only for some concepts. Strong probing does not always translate to steering or vice-versa. SAEs are useful for sparse probing of human-labelled concepts and novel-concept discovery, while PCA is competitive for dense probing and steering. Auto-interpretability finds $684$ automatically validated SAE features in Qwen3-TTS covering speaker clusters, speaking style, and recording conditions beyond known concepts. Overall, we perform a comprehensive interpretability study of autoregressive voice-cloning TTS models.
175. 【2610.40379】Agentic LLM reasoning in pediatric diabetes outcome prediction from multimodal medical records
链接：https://arxiv.org/abs/2610.40379
作者：Daniel Palacios,Guantong Qi,Anna Kadi,Natali Borrego,Sukru Aras,Michael L. Ferm,Elizabeth Kubota-Mishra,Maria J. Redondo,Zhandong Liu,Mustafa Tosur,Hu Chen
类目：Quantitative Methods (q-bio.QM); Computation and Language (cs.CL)
关键词：Large language model, Large language, language model, Clinical Evidence-Driven Agent, self-auditing LLM pipeline
备注： 18 pages main text plus 40 pages Supplementary Information
点击查看摘要
Abstract:Large language model (LLM) agents are proposed for clinical prediction but are untested in pediatrics, where cohorts are small and outcomes are uncommon. Using records from 2,489 youth with type 2 diabetes (T2D) and 4,018 with type 1 diabetes (T1D), spanning structured data, note-derived social determinants and, in T1D, continuous glucose monitoring data, we predicted seven T2D outcomes, four also in T1D, at 24 months from 18 months of follow-up. CEDAR (Clinical Evidence-Driven Agent for Risk-prediction), an evidence-citing, self-auditing LLM pipeline, outperformed trained machine learning in both cohorts (pooled area under the receiver operating characteristic curve 0.815 versus 0.656 to 0.700 in T2D and 0.756 versus 0.603 to 0.631 in T1D), most on the sparsely screened outcomes. Only 4 of 70,246 values it cited did not match the chart. Reading only the evidence and reasoning cited by CEDAR, two pediatric endocrinologists reached the same prediction as CEDAR in about nine of ten cases, agreeing with each other on 90.5%.
176. 【2610.37912】Beyond Utterance-Level Speaker Similarity: Perception-Aligned Segmental Evaluation of Accent in Speech Generation
链接：https://arxiv.org/abs/2610.37912
作者：Jinzuomu Zhong,Alice Ross,Catherine Lai,Korin Richmond
类目：Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD)
关键词：Utterance-level similarity ratings, Utterance-level similarity, generated speech, perceived accent differences, occur in generated
备注：
点击查看摘要
Abstract:Utterance-level similarity ratings do not reveal where perceived accent differences occur in generated speech. We introduce a segmental evaluation protocol combining speaker and accent similarity ratings with subword annotations of perceived accent differences provided by listeners from target accent regions. Using newly collected recordings, 47 retained listeners evaluate 200 content-matched reference--candidate pairs spanning four Scottish and Australian voices, two zero-shot TTS systems, and an articulation-based voice conversion system. Subword annotations locate salient pronunciation differences and sometimes yield different system rankings from similarity ratings. Speaker and accent ratings remain strongly associated ($\rho = 0.71$), highlighting the value of examining local perceptual evidence beyond these scores. Evaluation of seven objective distances reveals weak correspondence with subword selections, with correlations reaching at most $\rho = 0.22$. We release code, data, and an interactive demo to support the development of perceptually grounded segmental accent evaluation.
信息检索
1. 【2610.40619】stimony from Timelines: Evidence-Grounded Significance Testing of Temporal Clinical Hypotheses
链接：https://arxiv.org/abs/2610.40619
作者：Chengyang He,Wenlong Zhang,Koduvayur Subbalakshmi,Ping Wang
类目：Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：side effect tends, sales tend, tend to dip, effect tends, tends to follow
备注：
点击查看摘要
Abstract:Ask a language model whether sales tend to dip after a product recall, or whether one side effect tends to follow another after a vaccine, and it answers in a fluent sentence that reads like a finding, with no test behind it. Assertions about how events unfold across many cases fill the Web, and each is a temporal hypothesis that only a significance test over the right records can settle. The verifier must specify the null, select the test and decide which records bear on the hypothesis, and any of these choices can go wrong while the conclusion still comes out right. Yet benchmarks for language-model data analysis grade only the conclusion, with the relevant data handed over. We study the problem in health, where a wrong verdict is costly, and present STRATA, a benchmark that grades every decision on the way to a verdict: 8,735 hypotheses over the clinical timelines of vaccine adverse-event reports, with gold plans, evidence sets, p-values and verdicts derived under a released codebook and reproduced by a deterministic oracle. We also propose ART (articulate, retrieve, test), a staged paradigm that exposes each decision as an inspectable output. Extensive experiments with five model configurations show that systems fail mostly at choosing which records count, a verdict-evidence gap that accuracy alone conceals. ART closes most of this gap and brings a 31B open model to within 0.01 of the frontier models under the same paradigm; the frontier models succeed once given the codebook.
2. 【2610.40454】Glance: The Generator Is Its Own Reranker
链接：https://arxiv.org/abs/2610.40454
作者：Adilet Metinov,Chingiz Arziev
类目：Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：duplicating model residency, serves separate models, Retrieval-augmented generation commonly, commonly serves separate, generation commonly serves
备注： 12 pages, 2 figures, 6 tables. Adapters and code: [this https URL](https://huggingface.co/airun-ai/GlanceRAG-MiMo-9B-Reranker)
点击查看摘要
Abstract:Retrieval-augmented generation commonly serves separate models for reranking and answering, duplicating model residency and passage processing. We introduce Glance, which equips the generator with a lightweight decision adapter and pointer head to score candidate passages from a shared prefill. Short relevance probes branch from the cached context, producing one score per passage without separately prefilling each query-passage pair. A single resident 9B backbone supports generation, general decisions and reranking, with approximately 0.1 GiB of additional memory per decision skill. After 39 minutes of MS MARCO fine-tuning, Glance achieves nDCG@10 of 0.741 on SciFact and 0.345 on NFCorpus, compared with 0.721 and 0.326 for bge-reranker-v2-m3. On 300 HotpotQA questions, its selected context improves answer F1 by 0.045 over cross-encoder-selected context under the same base generator (paired bootstrap 95% CI [0.014, 0.076]). In a scaling study, reranking quality keeps rising up to 200 candidates (43k tokens) scored from one shared prefill, where it exceeds the cross-encoder by 0.058 nDCG@10, and a probe-only activated adapter lets generation continue from the original passage cache. These initial results demonstrate generator-based reranking with one resident backbone; evaluation currently covers two retrieval datasets, one QA dataset and one backbone family.
3. 【2610.40314】Decompose, Estimate, Select: Coverage-Aware Context Construction for Long-Form RAG
链接：https://arxiv.org/abs/2610.40314
作者：He Tu,Aleena Bosco Kurumthottam,Torsten Suel
类目：Information Retrieval (cs.IR)
关键词：spanning multiple aspects, complex questions spanning, questions spanning multiple, Retrieval-augmented generation, language model output
备注：
点击查看摘要
Abstract:Retrieval-augmented generation (RAG) grounds a language model's output in retrieved evidence, but for complex questions spanning multiple aspects of an information need, relevance-based retrieval may return redundant evidence while leaving some aspects unsupported. We propose a coverage-aware retrieval pipeline that decomposes queries into subquestions, ranks candidates with a coverage-trained bi-encoder, predicts a document\,$\times$\,subquestion coverage matrix with a cross-encoder, and selects documents by greedy maximum-marginal coverage. On TREC RAG 2024, our approach improves $\alpha$-nDCG@10, which rewards newly covered nuggets and discounts redundant evidence, from $0.625$ to $0.702$, approaching the $0.718$ gold-nugget upper bound without sacrificing relevance. We then find that ground-truth oracle retrieval, despite substantially higher retrieval coverage, delivers no more nugget coverage to the answer than our estimated selection. This suggests that answer completeness depends not only on what evidence is retrieved, but also on how effectively the generator uses it. To address this, we introduce a \emph{grounded self-feedback loop} in which the model identifies underused evidence from its first answer and generates again without gold supervision, raising $V_{\mathrm{strict}}$, the fraction of vital reference nuggets fully supported by the answer, by $0.086$ over two passes while largely preserving faithfulness.
4. 【2610.40196】owards Full-Text Scholarly Argument Mining using Entity-Anchored Claim Extraction
链接：https://arxiv.org/abs/2610.40196
作者：Bhuvanesh Verma,Andy Lüking,Mounika Marreddy,Leon Lukas Hammerla,Alexander Mehler
类目：Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：Argument mining, scholarly articles remains, articles remains, remains an important, important yet challenging
备注：
点击查看摘要
Abstract:Argument mining in full-text scholarly articles remains an important yet challenging task. This difficulty stems not only from the need to model long-distance dependencies but also from the substantial effort required to annotate sentence-level argumentative components throughout lengthy documents. To address these challenges, we introduce a novel approach that leverages scientific entity mentions (e.g., tasks, datasets, and models) to identify and structure claims in scholarly literature. We validate our approach through a case study involving 50 full-text research papers on sentiment analysis. We first establish a reliable human gold standard (Krippendorff's $\alpha = 0.80$), identifying 113 claim-bearing sentences and extracting 199 structured claim records from 300 candidate sentences. To address the annotation bottleneck, we employ a majority-voting ensemble of large language models (LLMs) to generate a silver-standard dataset comprising 1,487 claims. Encoders fine-tuned exclusively on the silver-standard dataset outperform those trained on a small human-annotated gold-standard dataset and demonstrate strong zero-shot transfer to NLP papers outside the training corpus.
5. 【2610.40122】xPRT: Cross-Passage Retrieval Pretraining
链接：https://arxiv.org/abs/2610.40122
作者：Danush Khanna,Himanshu Singhal,Pranjal A. Chitale,Bishal Santra
类目：Information Retrieval (cs.IR)
关键词：underlying modern RAG, foundational component underlying, component underlying modern, modern RAG, RAG and search
备注： Under Review
点击查看摘要
Abstract:Dense retrievers are a foundational component underlying modern RAG and search systems. Training these models typically requires a curated collection of annotated query-document pairs, which are prohibitively expensive to obtain outside a handful of well-resourced domains. This has motivated a line of work that leverages large-scale unsupervised retrieval pretraining from unlabelled text alone. Most such methods learn by optimizing a language modelling or contrastive learning objective over short spans within a document. This raises the question: how can we extract richer learning signals from a single document for retrieval pretraining? To address it, we introduce xPRT (Cross-Passage Retrieval Pretraining), an unsupervised framework that learns from the key information distributed across an entire document and the connections between its parts. We evaluate xPRT against five state-of-the-art objectives on MTEB English v2, continuing pretraining from a public unsupervised checkpoint while holding the corpus, schedule, and training procedure fixed. Across diverse tasks spanning all task types, we observe that xPRT consistently outperforms the strongest baseline by 3.2% on an average. Notably, in this continual pretraining setting, many baseline methods incur substantial performance regressions, whereas our method delivers the largest complementary gains over the initial pretrained checkpoint. Controlled ablations confirm that xPRT's performance gains stem from its proposed cross-passage signals. xPRT demonstrates that a single unlabelled document is a much richer source of retrieval signal than prior work assumed and offers complementary gains to what current retrievers already capture.
6. 【2610.40112】Beyond IPS: Reliable Counterfactual Evaluation in Multi-Stage Ad Systems without Logged Propensities
链接：https://arxiv.org/abs/2610.40112
作者：Mohsen Malmir,Mohamed A. Radwan,Houssam Nassif,Murat Ali Bayir
类目：Information Retrieval (cs.IR)
关键词：limiting inverse propensity, inverse propensity scoring, multi-stage delivery systems, limiting inverse, propensity scoring
备注：
点击查看摘要
Abstract:Offline evaluation of ad-ranking policies in multi-stage delivery systems is challenging because action propensities are often unavailable, limiting inverse propensity scoring (IPS). Reward-model-based estimators such as the Direct Method (DM) avoid propensities but can be biased when candidate policies induce impression distributions that differ from the logged data. We formalize DM-based counterfactual replay and propose mixture-weighted reward modeling, which reweights logged data toward a mixture of candidate-policy distributions using density ratios from a multiclass domain classifier. We bound each candidate's value-estimation error by overlap, cold-start, and replay terms, and derive an asymptotic rule that uses the cold-start mass to decide whether to trust offline selection or route candidates to online A/B testing. On an ads-delivery simulator and on production traffic from a large-scale CTR ranking system, the mixture-weighted reward model improves policy selection over its unweighted counterpart in both settings, with the largest gains at low overlap in simulation. Combined with a doubly robust correction, it attains the highest Top-1 accuracy in production and at low and medium overlap in simulation.
7. 【2610.39984】R2QRec: Learning Recommendation Queries from Retrieval Feedback
链接：https://arxiv.org/abs/2610.39984
作者：Weixin Chen,Shuchang Liu,Hailan Yang,Yuhan Zhao,Li Kang,Xuefeng Zhang,Chenghao Zhang,Xinyuan Wang,Hanchen Yang,Yinghao Wu,Hua Huang,Yandong Bai,Fan Yang,Xiang Li,Chunjie Chen,Li Chen
类目：Information Retrieval (cs.IR)
关键词：Query-based recommendation separates, separates expressing user, locating matching items, recommendation separates expressing, Query-based recommendation
备注：
点击查看摘要
Abstract:Query-based recommendation separates expressing user needs from locating matching items. Its central learning problem is how to turn item interactions into effective retrieval requests when interaction logs do not annotate queries. We present R2QRec, which uses a fixed retrieval system to construct query supervision and evaluate recommendation outcomes. Retrieval verification first selects generated expressions that recover observed items as supervised fine tuning targets, allowing retrieval behavior to determine the target wording. Policy refinement then evaluates generated query sets by the observed item's rank in their fused recommendation list. Log-rank feedback distinguishes outcomes beyond the recommendation cutoff, while a KL penalty constrains policy change relative to the supervised reference. The stages have distinct roles: verification identifies expressions to imitate, and policy refinement improves the combined ranking produced by newly generated queries. Only the query generator is trained; the retriever and item index remain fixed, and no query annotations or learned reward model are required. Across three public datasets, R2QRec achieves state-of-the-art overall recommendation performance among the evaluated methods. Extensive experiments demonstrate the effectiveness of the proposed design and its generalization across generators and domains.
8. 【2610.39682】Heterogeneous Co-Evolution Learning for Tool-Efficient Search Agents with Zero Labels
链接：https://arxiv.org/abs/2610.39682
作者：Qin-Wen Luo,Sheng Ren,Xiang Chen,Rui Liu,Jun Fang,Naiqiang Tan,Jun Wang,Hua Chai,Sheng-Jun Huang
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
关键词：extend language models, Deep search agents, agents extend language, Deep search, external retrieval
备注：
点击查看摘要
Abstract:Deep search agents extend language models by interleaving reasoning with external retrieval. However, effective search-agent training requires addressing two challenges beyond final-answer optimization. First, the capability frontier of agents is largely determined by training questions, whereas fixed or labeled task collections cannot adapt as agents improve. Second, outcome-only reinforcement learning treats all successful trajectories equally, even when some correct answers rely on redundant searches, creating an inherent tension between answer accuracy and tool efficiency. To address these issues, we introduce \textsc{ZEST-Search}, a zero-label and tool-efficient framework for search-agent training via heterogeneous co-evolution. Starting from unlabeled reference questions, \textsc{ZEST-Search} constructs validated search tasks online through an answer-first and evidence-path-guided generation process. A heterogeneous proposer adaptively constructs training tasks around the solver's evolving capability boundary, while the solver learns search-augmented reasoning through reinforcement learning. To jointly optimize task difficulty and search efficiency, \textsc{ZEST-Search} employs an intermediate-difficulty reward for the proposer and a task-local tool-aware reward for the solver, encouraging informative tasks and correct answers with fewer redundant searches. Experiments on seven general-domain and multi-hop question answering benchmarks demonstrate that \textsc{ZEST-Search} consistently improves search-agent performance while achieving higher tool efficiency without labeled search-training data or expert search trajectories. Further analyses show that heterogeneous generation, adaptive difficulty control, and tool-aware optimization enable more effective search-agent training.
9. 【2610.39667】Efficient LLM-based Subgraph Retrieval for Multi-Hop Knowledge Base Question Answering
链接：https://arxiv.org/abs/2610.39667
作者：Duanyang Yuan,Sihang Zhou,Xiaoshu Chen,Haoyuan Chen,Ke Liang,Jian Huang
类目：Information Retrieval (cs.IR)
关键词：find answer entities, Knowledge Base Question, Base Question Answering, Multi-hop Knowledge Base, Knowledge Base
备注：
点击查看摘要
Abstract:Multi-hop Knowledge Base Question Answering (KBQA) aims to find answer entities in the knowledge base that are multiple hops away from the entities in the question. Information retrieval-based (IR-based) methods extract a pivotal subgraph from the entire KB to locate candidate answers and then evaluate their plausibility through semantic matching with the question. However, we observed that the extracted subgraphs often include nodes that are weakly related or irrelevant to the question. Without a proper node filtering mechanism, the number of irrelevant nodes grows as the number of hops increases, leading to excessive consumption of computational resources. To address these challenges, this study introduces an efficient LLM-based subgraph retrieval method for multi-hop knowledge base question answering, M-ER. The framework leverages Monte Carlo Tree Search (MCTS) to transform subgraph exploration into a tree-structured search process. During the MCTS selection phase, nodes that are highly relevant to the question are prioritized for inclusion in the subgraph, eliminating the need to traverse all nodes in the KB. The framework further incorporates a large language model (LLM) to refine the search direction, ensuring that exploration remains focused on nodes relevant to the question. In addition, selected nodes are quantitatively scored, and these scores are fed back into the node selection process to effectively filter out irrelevant candidates, thereby improving the quality of the subgraph. This mechanism not only narrows the search space but also enhances the overall efficiency of multi-hop KBQA. Experiments on the WebQSP benchmark demonstrate that M-ER achieves 78.88\% on the Hits@1 metric, while also improving computational efficiency. These results not only validate the effectiveness of M-ER, but also offer a viable technical path to balance performance and computational efficiency.
10. 【2610.39610】GistRank: Gist-Compressed Listwise Ranking with Large Language Models
链接：https://arxiv.org/abs/2610.39610
作者：Guo Chen,Wenhan Liu,Reed Li,Qianwen Ma,Yu Lu,Haibo Shi,Bingbing Xu,Junjie Huang
类目：Information Retrieval (cs.IR)
关键词：improve web search, retrieved passages jointly, comparing retrieved passages, powerful tool, improve web
备注：
点击查看摘要
Abstract:Large language models (LLMs) have become a powerful tool for listwise relevance ranking, which can improve web search by comparing retrieved passages jointly. However, processing every candidate verbatim and generating a permutation autoregressively can make inference costly, every candidate must be serialized verbatim into the prompt and the input therefore grows linearly with both the number and the length of the passages. While sliding-window ranking handles context limits through repeated re-encoding of overlapping blocks, hard compression techniques sacrifice the fine-grained details required for accurate discriminative judgment. To better understand and mitigate this trade-off, we conduct a comprehensive study of soft compression for LLM-based listwise ranking. Our findings reveal that a vanilla gist representation is ill-suited to ranking for two reasons: strictly causal encoding weakens how thoroughly each gist token absorbs its passage, and the representation is never trained to expose the relative-preference structure that separates closely competing candidates. To this end, we propose GistRank, a three-stage training framework that equips a large language model with gist-compressed passage representations by learning faithful gist tokens under a semi-bidirectional attention scheme, instilling discriminative comparison through pairwise training, and adapting the model to the listwise objective. GistRank collapses each passage into only $K{=}4$ gist tokens, this gives 14$\times$ speedups over one-shot full-text ranking while reaching 59.17 average nDCG@10 over TREC DL19/20 and five BEIR subsets within 0.89 of the strongest full-context baseline. Overall, GistRank provides an effective means to enhance efficiency for searches in real-world environments.
11. 【2610.39608】From Intent to Improvement: Agentic Dataset Search for Better Model Training
链接：https://arxiv.org/abs/2610.39608
作者：Qiaosheng Chen,Kaijia Huang,Fanqi Zhou,Gong Cheng
类目：Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
关键词：million-scale machine learning, machine learning repositories, Web search problem, Finding training data, Finding training
备注： 13 pages, 4 figures
点击查看摘要
Abstract:Finding training data in million-scale machine learning repositories is a Web search problem where text relevance is not enough. A query couples a semantic training goal with constraints such as task, license, and training stage, and the metadata that would verify them are often missing. Text retrieval ranks dataset cards by similarity but cannot enforce hard constraints or follow the relations that reach a sparsely described dataset. Free-form tool-calling agents are more flexible, but the same model that picks a dataset also declares it a match, and nothing checks that the query's constraints are actually met. Neither approach establishes that the data it returns will improve model training. To address it, we propose AgenticDS, a contract-governed agentic framework in which every language model decision instantiates a typed object before deterministic code consumes it. The framework compiles a query into a typed search plan, executes it through admissible deterministic operators over hybrid retrieval and bounded graph expansion, and verifies the provenance-grounded evidence before accepting a ranking or emitting a typed diagnosis that permits only local repairs. On HuDS-Bench, a new benchmark of 200 queries with 24,908 graded judgments pooled over 15 systems, AgenticDS improves NDCG@10 by 25.7\% over the strongest dense--sparse hybrid and by 18.9\% over a commercial LLM agent. A controlled search-to-train study further shows that the datasets it retrieves give the best downstream accuracy in mathematics, code, and medicine. Code and data are publicly available at this https URL.
12. 【2610.39519】Lost in the Latent Space: How Code Embeddings Represent Functional Intent and Implementation Fidelity
链接：https://arxiv.org/abs/2610.39519
作者：Lu Sun,Yuheng Huang,Lei Ma,Zhijie Wang
类目：oftware Engineering (cs.SE); Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：key building block, remain poorly understood, poorly understood, Code embedding models, key building
备注：
点击查看摘要
Abstract:Code embedding models have become a key building block for many software engineering applications, yet their internal representations remain poorly understood. Existing evaluations tend to focus on downstream task performance, offering limited insight into whether a single embedding can support both functional abstraction and implementation-sensitive analysis. To address this gap, we present a large-scale empirical study of 14 state-of-the-art code embedding models. We evaluate their representational behavior through three complementary analyses: supervised probing for functional abstraction and implementation fidelity, zero-shot retrieval experiments to assess geometric similarity, and controlled perturbation experiments that measure embedding responses to semantics-preserving code transformations. The results reveal a consistent abstraction-fidelity tension: models that better preserve stylistic and structural signals tend to do so at the expense of semantic generalization, while models with similar retrieval performance can nevertheless respond very differently to the same implementation changes. Overall, our findings suggest that a single universal code representation remains out of reach, motivating more representation-aware evaluation, selection, and deployment of code embedding models in software engineering.
13. 【2610.39470】RoFiGR: Coarse Routing, Fine Refinement via Level-Adaptive Distillation in Multimodal Generative Recommendation
链接：https://arxiv.org/abs/2610.39470
作者：Ji Dai,Zhipeng Wei,Zhuang Zhuang,Rongfeng Guo,Jie Chen,Quan Fang,Fei Pan
类目：Information Retrieval (cs.IR)
关键词：Generative recommenders predict, item semantic identifier, Generative recommenders, semantic identifier, Generative
备注： 12 pages, 10 figures, 10 tables
点击查看摘要
Abstract:Generative recommenders predict an item's semantic identifier (SID) one code at a time. Early codes select broad branches, so a routing error can exclude relevant items before fine-grained prediction begins. In multimodal recommendation, these coarse decisions must reflect collaborative preferences as well as item semantics. We introduce RoFiGR, a two-tree recommender that uses collaborative information in both SID construction and generation. It aligns cumulative L1-L2 code vectors with frozen collaborative item embeddings and explicitly encodes item distance and source level in the serialized history. During generation, a sequential teacher supervises legal branches. Fixed coarse-level weights maintain routing supervision, while a bounded adaptive residual allocates additional weight according to teacher-student demand. On Instruments, Arts, and Games, RoFiGR improves all five ranking metrics over the evaluated external baselines. Five-seed paired tests with Holm correction support the main NDCG@10 improvements. Comparisons with alternative collaborative supervision objectives and tokenizers support the proposed design. Matched task and decoding analyses distinguish the effects of training exposure, stream fusion, and candidate budgets.
14. 【2610.39426】UGR: Unified Generative Ranking for Heterogeneous E-commerce Recommendation
链接：https://arxiv.org/abs/2610.39426
作者：Zonghe Shao,Huabin Chen,Sitong Lu,Fangmei Zhu,Yingcai Ma,Ge Fan,Jinguo Gao,Wei Shi,Yanfang Liu,Huiping Chu
类目：Information Retrieval (cs.IR)
关键词：balance user utility, modeling candidate interactions, List-wise ranking, generative ranking, organic content
备注：
点击查看摘要
Abstract:List-wise ranking in e-commerce recommendation must balance user utility and commercial value across organic content, ads, and candidates from other businesses, while modeling candidate interactions and sequential dependencies. Existing generative ranking methods typically learn from logged exposure sequences and operate on truncated candidate sets from a single business. Exposure-sequence learning may not fully exploit post-exposure feedback signals. Meanwhile, generative ranking over heterogeneous, multi-business candidates remains less explored, particularly when recommendation scores and advertising value must be jointly modeled under generalized second-price (GSP) auctions. We propose Unified Generative Ranking (UGR), which performs generative ranking over the full request-level multi-business heterogeneous candidate set, typically comprising around a thousand candidates in production, and directly generates multi-business ranking lists through candidate-aware autoregressive decoding. To go beyond exposure-sequence imitation, UGR introduces complementary feedback supervision through weighted supervised fine-tuning and permutation preference optimization, which respectively leverage observed post-exposure feedback and feedback-derived preferences among permutations. UGR uses a training-time advertising debiasing tower to model historical ad exposure effects and incorporates current ad value during serving while preserving the existing allocation and GSP pricing rules. Multi-value sequence retrieval produces candidate ranking lists, and the existing downstream evaluator selects the final list for exposure. Offline experiments demonstrate consistent gains in ranking quality, while online A/B tests on Taobao's Guess You Like recommendation service yield improvements of 1.76% in Core IPV, 3.29% in Transaction Volume, and 2.10% in GMV over the production baseline.
15. 【2610.39352】ry Before You Evolve: Ranking-Scored Trials for Recursive Self-Improvement of Recommender Systems
链接：https://arxiv.org/abs/2610.39352
作者：Xu Huang,Jiawei Feng,Jiancan Wu
类目：Information Retrieval (cs.IR)
关键词：Web platforms, e-commerce and music, music streaming, training recipes re-tuned, Recommenders on Web
备注：
点击查看摘要
Abstract:Recommenders on Web platforms such as e-commerce and music streaming need their training recipes re-tuned for each dataset. Large language model (LLM) agents can now rewrite these recipes, enabling recursive self-improvement (RSI) of recommender systems. Existing agents propose one edit per round, fully retrain the recommender, and keep the edit only if its ranking quality improves. Unlike programs or patches that can be scored by execution, a recommender edit changes how the model is trained, which raises two challenges. First, ranking feedback is delayed: an edit's ranking quality is known only after full retraining, so much of the retraining budget goes to edits that are later discarded. Second, the signals available before retraining are unreliable: sampled training losses can look solved while full-ranking performance remains poor, and they are not comparable across edits that change the loss or sampler. We propose RecRSI, which generates diverse candidate edits from a diagnosis of the current model and selects one for full retraining using early ranking feedback: each candidate receives a short trial, the first steps of its full training run, scored by full-ranking NDCG@10 on held-out probe users. With two sequential recommenders on three datasets and the same number of full retrainings, RecRSI significantly outperforms existing self-evolving agents in every setting, and on both Amazon datasets reaches in its first round a ranking quality that none of them reaches in five.
16. 【2610.39339】Does the Citation Matter? Quality, Grounding, and Physician-Rubric Accuracy of Google's AI Overviews for Medical Information
链接：https://arxiv.org/abs/2610.39339
作者：Sam Agnoli,Brigid Reilly,Lily Barna,Emőke-Ágnes Horvát
类目：Information Retrieval (cs.IR); Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
关键词：searching the web, shifting the work, Google traditional algorithm, encounter when searching, work of selecting
备注： 12 pages, 5 figures, 10 tables. Code and data: [this https URL](https://github.com/SamAgnoli/AIOverviews)
点击查看摘要
Abstract:Google's AI Overviews (AIOs) are the first result users encounter when searching the web, shifting the work of selecting and evaluating sources from the user to Gemini. This may be especially consequential for health information, given that Google's traditional algorithm was designed to prioritize quality sources, and inaccurate information carries health risks. We evaluated 3,611 benchmark health questions, comparing each AIO against that question's top five traditional results on source selection and quality. We also evaluated the AIOs' statement grounding in their cited sources and how accurate AIOs were against physician-written rubrics. Overviews appeared for 96.4% of questions but drew on substantially different evidence than traditional search: only 44.4% of the top five results were cited by the corresponding AIO, which instead overrepresented commercial health publishers (e.g., WebMD, Healthline) at the expense of academic and government sources (e.g., Cleveland Clinic, National Health Service). AIO-cited pages scored lower than traditional results on four of six well-established credibility criteria. The grounding of AIO statements to their sources was poor: after correcting for classifier error, an estimated 56.7% of cited statements were supported by their cited page, and a further 20.8% of all statements were factual claims with no citation at all. When AIOs were graded against physicians' rubrics, they earned 38.2% of all available points (e.g., accuracy 54.2%, completeness 21.6%), and among the subset of AIOs whose physician rubric anticipated a specific inaccurate or potentially harmful claim, 22.2% of AIOs contained the harmful content. Notably, the accuracy of AIOs was unrelated to the grounding of their statements. We recommend AIO literacy interventions for users and holding AIOs at least to Google's own health-ranking standards.
17. 【2610.39289】Federated Recommender Agents
链接：https://arxiv.org/abs/2610.39289
作者：Jaehyung Lim,Woojoo Kim,Junyoung Kim,Dongha Kim,Wonbin Kweon,Hwanjo Yu
类目：Information Retrieval (cs.IR)
关键词：LLM-powered recommender agents, raises privacy concerns, user information raises, information raises privacy, Federated Recommender Agents
备注： 13 pages
点击查看摘要
Abstract:LLM-powered recommender agents use reasoning and memory for personalization, but centralized access to user information raises privacy concerns, while isolated agents cannot learn from others' experiences. We formulate Federated Recommender Agents, a setting where local agents collaborate through natural-language knowledge while keeping interaction histories and personal memories local. We propose FedRecAgent, which enables this collaboration without updating LLM parameters. Its central challenge is determining which reasoning to integrate, since similar behavioral contexts can lead to different choices. FedRecAgent represents local experiences as Condition-Transition-Reasoning (C-T-R) units and hierarchically integrates reasoning by condition and transition. Within this structure, it consolidates related explanations and discriminates reasoning across alternative transitions. Each client maintains semantic and decision memories to capture preferences and conditional choice tendencies, guiding the retrieval and selection of shared reasoning for recommendation. Experiments across three domains under dense and sparse settings demonstrate improved recommendation performance over federated and LLM-based baselines, while ablations support the contributions of personalized memories and structured reasoning integration.
18. 【2610.39273】PFA: Attention Can Be Progressively Focused in Long-Sequence Generative Recommendation
链接：https://arxiv.org/abs/2610.39273
作者：Pengjie Zhou,Weiqin Yang,Haibo Xing,Hao Deng,Jinxin Hu
类目：Information Retrieval (cs.IR)
关键词：Long user behavioral, user behavioral sequences, behavioral sequences improve, autoregressive beam search, Long user
备注： 11 pages, 3 figures, 7 tables
点击查看摘要
Abstract:Long user behavioral sequences improve the performance of generative recommendation (GR), but computing full attention over long histories during autoregressive beam search decoding incurs substantial online inference latency. To accelerate long-sequence decoding, existing approaches typically rely on static history compression or dynamic top-k sparse attention; however, the former risks losing critical semantics, whereas the latter incurs significant overhead from repeated token selection and non-contiguous memory access. In this work, we thoroughly analyze attention patterns during GR decoding and reveal two key properties: (i) progressive attention focusing, where attention progressively focuses on a small subset of critical historical tokens; and (ii) cross-step ranking consistency, where their importance rankings remain relatively stable across subsequent decoding steps. Motivated by these observations, we propose Progressively Focused Attention (PFA), an efficient sparse decoding framework for long-sequence GR. PFA retains full attention at the initial decoding step to explore global historical preferences. It then establishes a shared importance ranking through cross-beam weighted aggregation and reorders the key-value cache of the selected tokens into a contiguous memory layout. Subsequent decoding steps reuse this ranking and geometrically reduce the token budget, preserving critical semantics while lowering computational and memory access overhead. Extensive experiments on two public benchmarks and a large-scale industrial dataset demonstrate that PFA outperforms state-of-the-art baselines. A 7-day online A/B test on a large-scale e-commerce advertising platform further confirms that PFA delivers substantial business gains while maintaining low inference latency. Our code is publicly available at this https URL.
19. 【2610.39244】CASH: A Capability-Aware Subsidy Harness for Advertiser Subsidy Allocation
链接：https://arxiv.org/abs/2610.39244
作者：Minmao Wang,Huihuang Lin,Likang Wu,Hongke Zhao,Shijie Yi,Fei Pan
类目：Information Retrieval (cs.IR)
关键词：platform-wide subsidy-rate constraint, maximize total discounted, total discounted revenue, platforms subsidize advertisers, billing discounts
备注：
点击查看摘要
Abstract:Advertising platforms subsidize advertisers through billing discounts, and Advertiser Subsidy Allocation (ASA) chooses a subsidy level for every account to maximize total discounted revenue under a platform-wide subsidy-rate constraint. In practice, operators configure these subsidies from business rules, advertiser tiers and operating goals. Such configuration draws on rich business experience, yet as the advertiser base grows, tier-level policies can neither resolve how individual accounts respond across subsidy levels nor allocate a shared budget at the granularity of single accounts. We propose CASH, a Capability-Aware Subsidy Harness for ASA. CASH pairs a knowledge-guided large language model (LLM) with a data-driven numerical response model, learns from realized outcomes where each is locally reliable across accounts and subsidy levels, and turns this reliability into explicit capability licenses that govern source selection, agent review and budget allocation. Its Expected Policy Coverage Gain (EPCG) directs a limited feedback budget toward evidence expected to expand high-value reliable coverage, and the resulting feedback further improves the numerical model and updates capability estimates. In a one-month online experiment with CASH serving 60% of a large-scale advertising platform's traffic, CASH raises revenue per account by 15.35% and ADVV by 19.86% over Control at an essentially unchanged subsidy rate, with further gains in revenue quality, LTV and retention.
20. 【2610.39231】A Sequential Decision Framework for Evaluating LLM-Guided Search
链接：https://arxiv.org/abs/2610.39231
作者：Siqing Huo,Andrew Parry,Debasis Ganguly,Mark Smucker,Charles L. A. Clarke
类目：Information Retrieval (cs.IR)
关键词：Recent systems applying, systems applying retrieval-augmented, iteratively generate queries, subsequent search steps, guide subsequent search
备注：
点击查看摘要
Abstract:Recent systems applying retrieval-augmented generation (RAG) and "deep research" iteratively generate queries, retrieve documents, and examine their contents, using the resulting information to guide subsequent search steps. When this process is carried out autonomously rather than by a human searcher, the system must make explicit decisions about which search trajectories to explore and what objective the search is intended to optimize. We introduce a simple framework for reasoning about such search policies, providing a principled approach to evaluation. The framework models the search environment as a set of ranked lists and represents behavior using two primitive actions: select, which advances judgment through an existing ranking, and spawn, which generates a new ranking through query generation or reformulation. Each action produces a document to inspect, yields information that updates the search state, and contributes a reward for a task-specific objective. Within this abstraction, a variety of existing retrieval strategies can be expressed as policies operating over a common environment, while common retrieval objectives can be represented through reward and weighting functions over the resulting search trajectory, and the resulting formulation admits a natural reinforcement learning interpretation. We illustrate the approach using high-recall information retrieval (HRIR), where the goal is to identify as many relevant documents as possible with limited judging effort, showing how the framework provides a unified way to describe and reason about strategies for both classical retrieval tasks and emerging LLM-guided search systems. Through experiments, we demonstrate that both the underlying retrieval method and the "deep research" policy employed can materially affect performance.
21. 【2610.39188】Read More, Reason Deeper: Latent Workspaces for Long-Sequence Recommendation
链接：https://arxiv.org/abs/2610.39188
作者：Renzhi Wu,Li Sheng,Zikun Cui,Junjie Yang,Tai Guo,Keke Zhai,Yan Li,Jerry Fu,Nipun Mathur,Sri Reddy,Hong Li
类目：Information Retrieval (cs.IR)
关键词：Long-sequence recommendation, exposes longer histories, exposes longer, emph, Latent Workspace Ranker
备注：
点击查看摘要
Abstract:Long-sequence recommendation can improve along two axes: \emph{evidence scaling} exposes longer histories, while \emph{reasoning scaling} adds layers that interpret that evidence for a candidate. Existing models couple them: sequence encoders propagate every added event through depth, whereas one-shot selection prevents later layers from revisiting omitted evidence. We introduce the Latent Workspace Ranker (LWR), which keeps typed evidence in persistent banks and propagates a fixed-size candidate-local workspace. Each layer rereads the banks from its evolving workspace and archives compact post-interaction state, allowing history length and reasoning depth to grow independently. We compare four architectures across various history--depth configurations using evidence scaling at fixed depth, reasoning scaling at fixed 10K history, and their joint compute-efficient frontier. LWR leads all families at long histories, continues improving with reasoning compute, and converts additional training compute into quality gains at the fastest fitted rate on the joint frontier. At 10K history, it improves NE by 0.19% over HSTU while using 69.6% fewer forward GFLOP/request. It also outperforms existing baselines on two billion-scale ranking surfaces. These results show that separating addressable evidence from propagated state makes reading more and reasoning deeper independently controllable and jointly useful.
22. 【2610.39175】PRISM: Personalized Item Representations in User History
链接：https://arxiv.org/abs/2610.39175
作者：Renzhi Wu,Zikun Cui,Junjie Yang,Li Sheng,Yongjun Xie,Jerry Fu,Nipun Mathur,Sri Reddy,Hong Li
类目：Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
关键词：Sequential recommenders typically, recommenders typically construct, Sequential recommenders, embedding map shared, typically construct history
备注：
点击查看摘要
Abstract:Sequential recommenders typically construct history inputs from item identifiers or content features using an embedding map shared across users. The same item therefore yields the same initial embedding representation, even when users have different relationships to the item. We propose personalized item representations in user history, where each item's embedding expresses its relationship to the user. Personalization becomes a property of the history inputs themselves, supplying richer evidence for later sequence modeling. We instantiate this idea with PRISM, which reuses upstream ranking computation to generate personalized item embeddings. An autoencoder jointly compresses all layer states from a user tower, a user--item tower, and same-request candidate mixing into one embedding per scored pair for downstream replay. Offline evaluation on Surface 1, a major recommendation surface at billion scale, reports an average downstream normalized entropy (NE) change of $-0.15\%$, compared with $-0.07\%$ for VISTA at similar history lengths. We further identify an empirical sequence-length scaling law: over history lengths of 26--240 events, NE gain follows a fitted power law with exponent 0.468, increasing from 0.06\% to 0.17\%. Therefore, personalized item representations in user history provide a new scaling dimension for improving recommendation quality.
Subjects:
Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
Cite as:
arXiv:2610.39175 [cs.IR]
(or
arXiv:2610.39175v1 [cs.IR] for this version)
https://doi.org/10.48550/arXiv.2610.39175
Focus to learn more
arXiv-issued DOI via DataCite</p>
23. 【2610.39090】PALATE: Pretrained Autoregressive LATent Embeddings
链接：https://arxiv.org/abs/2610.39090
作者：Sizhang Zhao,Praveen Gondhi,Bin Li,Utsaw Kumar
类目：Information Retrieval (cs.IR); Machine Learning (cs.LG)
关键词：run large ads, large ads businesses, Local-commerce and food-delivery, food-delivery marketplaces run, marketplaces run large
备注： Accepted to GRAIL @ CIKM 2026. 5 pages, 2 figures, 3 tables
点击查看摘要
Abstract:Local-commerce and food-delivery marketplaces run large ads businesses: across hundreds of thousands of stores, merchants pay to be surfaced on browse and search surfaces, and a ranker decides which ads each consumer sees. Ranking well demands strong per-consumer personalization, yet a consumer's raw interaction history is sparse and high-cardinality. Our production ranker had personalized only through hand-engineered signals: aggregate engagement counts, categorical embeddings, and order-agnostic pooled history, with no learned per-consumer sequence representation. We learn one by training a GPT-style generative model of the consumer's engagement sequence to predict their next few engagements, and read its last-token hidden state as the consumer embedding. That single frozen vector becomes one shared personalization input for every ad surface: a daily job refreshes it, and each ranker consumes it as an ordinary dense feature. Offline it is the largest single-feature gain in our ranking-AUC ladder, a relative Eval-AUC gain of +0.59% on a browse-driven feed and +0.57% on a query-driven search surface. Online, it shipped as one component of bundled ranker retrains on both surfaces: Discovery (revenue +0.79%, conversion rate +1.17%) and Search (revenue +0.94%, conversion rate +1.50%). Three design choices adapt the generative objective to a repurchase-heavy marketplace, and are the core of our contribution: an exploration-weighted loss that up-weights businesses new to the consumer over ones already engaged; a two-stage curriculum: a click pretrain that keeps the representation broad and exploratory, then a conversion fine-tune that sharpens purchase intent; and a designed negative set that pairs a fixed popular-business pool and businesses shown but not ordered with random and in-batch negatives.
24. 【2610.38854】Contextual Embeddings Beyond the Gold Chunk
链接：https://arxiv.org/abs/2610.38854
作者：Sedigheh Eslami,Markus Krimmel,William Barber,Antoine Chaffin,Maximilian Schall,Jonas Golde,Louis Milliken,Ekaterina Glazkova,Denis Bykov,Bo Wang
类目：Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR)
关键词：document global context, Contextual embeddings represent, embeddings represent chunks, document global, represent chunks
备注：
点击查看摘要
Abstract:Contextual embeddings represent chunks of a document with respect to the document's global context. These models are typically trained and evaluated under the assumption that a relevant document contains only a single relevant chunk. In this work, we argue that in real-world use cases a retriever should not only find answers spread over several chunks but also surface the evidence needed to verify them. To address this, we introduce pplx-embed-v2-context, a family of contextual embedding models trained by distilling graded chunk-level relevance from a query-aware context compression model. This removes the need for gold-chunk labels and makes the training targets independent of the chunking. We utilize a vision-language backbone for training and release pplx-embed-v2-context in 9B and 0.6B sizes with compatible embedding spaces. On the evaluation side, we introduce two benchmarks that label both the chunks that answer a query and the evidence needed to verify the answer. Context-bench is a controlled diagnostic, and Q2D-Chunk is built from real-world search queries. To encourage further research, we also publish Q2D-Chunk as a public leaderboard. Our 9B model sets a new state of the art on both benchmarks and ConTEB. We find that the visual retrieval capabilities of our 9B model transfer from its vision-language base model and propose to extend chunking to image inputs. We demonstrate that our models achieve state-of-the-art performance among dense retrievers on visual ViDoRe-V3 and that performance improves with granular image chunking.
Subjects:
Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR)
Cite as:
arXiv:2610.38854 [cs.LG]
(or
arXiv:2610.38854v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.38854
Focus to learn more
arXiv-issued DOI via DataCite</p>
25. 【2610.38755】Leaving, Staying, or Living There? Duration Commonsense Reasoning with Human-LLM Co-annotations
链接：https://arxiv.org/abs/2610.38755
作者：Chenyang Yu,Zhaomin Xiao,Yan Huang,Eduardo Blanco
类目：Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
关键词：backyard implies living, backyard implies, implies living, Gardening, Airbnb
备注：
点击查看摘要
Abstract:Gardening in the backyard implies living there, while checking into an Airbnb more likely means staying for a few days. Such duration judgments are important for understanding human mobility, travel behavior, and social context. However, duration is rarely stated explicitly in social media posts and must instead be inferred from noisy textual, visual, and contextual cues. In this paper, we investigate whether vision-language models possess such duration commonsense. We create a new corpus of geotagged tweets co-annotated by humans and LLMs to serve as a testbed for this task. We find that state-of-the-art models struggle with duration inference, even when guided by multi-turn Chain-of-Thought prompting. Error analyses further reveal that models often fail to integrate critical contextual information, or they overly rely on prominent visual details while overlooking subtle hashtags and semantic nuances. These findings underscore the significant challenges remaining in context-aware multimodal understanding.
26. 【2610.38683】Quantum-Inspired Pairwise Ranking Aggregation via Active LLM Feedback
链接：https://arxiv.org/abs/2610.38683
作者：Quan Mai,Susan Gauch,Gibson Nkhata
类目：Information Retrieval (cs.IR); Computation and Language (cs.CL)
关键词：RAG systems, retrieval and RAG, interactive ranking aggregation, Ranking aggregation, ranking aggregation framework
备注：
点击查看摘要
Abstract:Ranking aggregation from multiple retrievers is a powerful yet under-explored technique in modern retrieval and RAG systems. We introduce LLM-IRA, the first extension of the quantum-inspired interactive ranking aggregation framework to document reranking. By leveraging large language models (LLMs) solely for pairwise judgments on high-disagreement document pairs, our method learns fusion weights in a principled, interactive way. The final ranking is a simple, ultra-fast pointwise linear fusion of scores, no full pairwise matrix or expensive listwise prompting is required. Experiments on the BEIR benchmark and TREC Deep Learning tracks with an open-weight instruction-tuned judge (Llama-3.1-8B-Instruct) show that LLM-IRA matches or numerically exceeds strong static fusion baselines on 5 of 6 evaluated benchmarks. We report two weight-update variants: momentum blending and a full accumulated-history refit, which together reach statistical significance (Wilcoxon signed-rank, p0.05) over the best unsupervised aggregator on 4 of 6 benchmarks (DL19 under momentum; DL20, DBpedia, and NFCorpus under accumulated refit, which is the stronger of the two configurations overall). On the benchmarks we profiled (DL19, DL20), it already clears the best unsupervised baseline within as few as 5 LLM calls per query, well before exhausting its typical ~25--30 call budget. LLM-IRA also achieves cost-effectiveness comparable to state-of-the-art zero-shot LLM rerankers at an order of magnitude fewer LLM calls, and requires no training data or model fine-tuning.
27. 【2610.38404】CoWalk: Cooperative Walks for Agentic Batch Queries on Disk-Resident Graphs
链接：https://arxiv.org/abs/2610.38404
作者：Qiao Liang,Zhiguang Chen,Yue Yu,Nong Xiao
类目：Databases (cs.DB); Information Retrieval (cs.IR)
关键词：group-relative policy optimization, semantically related vector, deep research, policy optimization, research and group-relative
备注：
点击查看摘要
Abstract:Workloads such as query rewriting, deep research and group-relative policy optimization (GRPO) issue batches of semantically related vector queries. Existing disk-resident approximate nearest neighbor (ANN) systems process these queries independently. On three workloads, 31 to 67 percent of the I/Os of a batch fetch a page that another query of the batch has already fetched. A page cache shared by the batch removes these I/Os, yet it makes a single-thread search only $1.1$ to $1.2{\times}$ faster, because the search time goes to distance computation and the cache does not change the walk of any query. We present CoWalk, which runs the queries of a batch as cooperating walks on one thread. Every query keeps its own candidate list, a page is read once per batch, and the full-precision vector on every fetched page is ranked by every query of the batch. A query therefore reaches the target recall with a shorter candidate list, which computes fewer distances and issues fewer I/Os. At recall@10 of 0.95, CoWalk answers a batch on one thread $1.6$ to $1.9{\times}$ faster than the fastest per-query baseline and issues $2.8$ to $6.9{\times}$ fewer I/Os. On a server it serves $3.0$ to $6.9{\times}$ the throughput of that baseline at a similar batch latency.
28. 【2610.38378】owards Reliable, Scalable, and Value-Aligned Generative Recommendation for Large-Scale Advertising
链接：https://arxiv.org/abs/2610.38378
作者：Yatong Xiao,Yibo Sheng,Lixiang Wang,Mingjie Sun,Yunhao Sha,Pengfei Zhang,Shengyu Wang,Peng Wang,Wenjin Wu,Peng Jiang
类目：Information Retrieval (cs.IR); Machine Learning (cs.LG)
关键词：reformulates item retrieval, discrete semantic identifiers, Generative recommendation, generative recommendation framework, reformulates item
备注： 9 pages, 3 figures, under review
点击查看摘要
Abstract:Generative recommendation (GR) reformulates item retrieval as autoregressive generation of discrete semantic identifiers (SIDs), offering a unified paradigm for personalized recommendation. GR4AD, as an industrial generative recommendation framework, has greatly improved advertising revenue while enabling efficient real-time serving in this http URL, continued deployment has exposed three limitations: deterministic SID assignment remains sensitive to semantic boundaries; conventional backbone scaling provides diminishing revenue gains under strict serving constraints; and sequence-level value optimization provides limited credit assignment to intermediate SID decisions, causing likelihood-based beam search to prematurely prune potentially high-value candidates. We present GR4AD-X, which addresses these limitations through reliable SID construction, serving-efficient capacity scaling, and unified value alignment. Specifically, GR4AD-X improves SID reliability through Multi-Path assignment. It expands model capacity along multiple complementary dimensions rather than relying solely on wider or deeper backbones, achieving a better effectiveness-efficiency trade-off under strict serving constraints. Finally, Unified Value-Guided Generation and Discrimination (UniVGD) aligns generation with commercial value through prefix-level reward optimization, value-guided beam pruning, and discriminative reranking within a unified model. Large-scale online A/B tests on the Kuaishou advertising platform demonstrate that GR4AD-X achieves a cumulative revenue improvement of 3.07% over the deployed GR4AD baseline while satisfying strict latency and throughput constraints.
29. 【2610.38361】Manifest-Assisted Retrieval in Artificial Intelligence Search: Evidence Gaps and a Security Evaluation
链接：https://arxiv.org/abs/2610.38361
作者：Gangzhen Qian,Kainan Zhou,Chuhong Xu,Hang Xiao,Zhi Jing,Rong Bao,Lu Yi,Zhaoyi Li
类目：Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL); Databases (cs.DB); Information Retrieval (cs.IR)
关键词：external sources retained, isolate the incremental, incremental security, security or utility, retrieval remained fixed
备注： 8 pages, 3 figures, 5 table. Accepted at IEEE FLLM 2026. Author accepted manuscript. Reproducibility artifact: [this https URL](https://fllm-conference.org/2026/)
点击查看摘要
Abstract:The external sources retained in our non-systematic audit did not isolate the incremental security or utility effect of consuming /llms.txt while ordinary retrieval remained fixed. The initial audit covered 39 sources through August 16, 2026, with later targeted updates through October 14. A purposive 15-tool study observed no automatic requests in 45 trials at one endpoint; a non-probability log study found that 97% of domains with detected files received no recorded requests in May 2026. We separate publication, fetch, model use, and action; define five operational threat paths and paired security-utility estimands; and exercise deterministic pre-model controls with 30 author-defined parser fixtures. We also execute a reduced pilot: 24 tasks across eight sites, five conditions, and 120 real model responses, with ordinary evidence held fixed. Frozen automatic scoring identifies attacker-target adoption in $21/24$ raw adversarial responses, with one further response unresolved; the adversarial URLs-only condition has no known adoptions and one unresolved response. The pilot measures output changes conditional on supplied evidence, without live retrieval or external actions. Together, the framework and pilot distinguish manifest-content effects from publication, discovery, and downstream execution, while leaving full retrieval and agent evaluation open.
30. 【2610.38090】Predicting What Answer Engines Cite: Open Decision Models, Rerankers and a Distilled 4B Student on Ten Engines under Controlled Retrieval
链接：https://arxiv.org/abs/2610.38090
作者：Elisha Bajemon,Andre-Louis Rochet
类目：Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
关键词：Generative engine optimization, ranks candidate pages, Generative engine, Jev, ranks candidate
备注： 12 pages, 1 figure, 4 tables. Model: [this https URL](https://huggingface.co/citead/geo-citation-ranker-4b) ; benchmark and code: [this https URL](https://huggingface.co/datasets/citead/geo-citation-benchmark)
点击查看摘要
Abstract:Generative engine optimization needs a scorer that ranks candidate pages by the share of an answer they will receive as citations. We measure pretrained rerankers, a hosted typed decision model (Jev), eight open decision models and distilled students on GEO-Bench queries answered by ten engines that receive the same candidate sources in their prompt. Where several engines answer the same query, they agree with one another at a within-group Spearman correlation of 0.704. Jev and two open 27B decision models, used without training, are the strongest scorers at 0.575 to 0.578. A Qwen3-Reranker-4B fine-tuned with low-rank adapters on teacher labels for other GEO-Bench queries, released as GEO Citation Ranker 4B, reaches 0.542 to 0.549 across three seeds, and its differences with the 27B models and with Jev are not distinguishable from zero on 229 queries. Most of the remaining distance between the student and Jev disappears when Jev scores one passage at a time, as the student does. Students trained on teacher labels and on observed engine citations reach the same level, and labels from Jev alone match the two-teacher target at 0.6B. We release the model, the benchmark, the predictions of every reported scorer and the code.
31. 【2610.37973】Merge and Branch: Efficient Tree-Structured Supervised Fine-Tuning of LLMs for Recommendation
链接：https://arxiv.org/abs/2610.37973
作者：Yunqin Zhu,Yao Xie
类目：Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
关键词：text-rich interaction histories, short semantic identifiers, LLM-based generative recommendation, recommendation reads long, generative recommendation reads
备注： 12 pages, 8 figures
点击查看摘要
Abstract:LLM-based generative recommendation reads long, text-rich interaction histories but generates only short semantic identifiers (SIDs). This asymmetry makes supervised fine-tuning inefficient: encoding the history is expensive, yet each history usually provides supervision for only one target. We present Merge and Branch (MB) to reduce input cost and learn more from each encoded history. Merge compresses each historical item's SID and text into a small number of learned tokens. Branch uses the compressed history to predict multiple target SIDs. To share history computation while keeping these targets independent, MB organizes each example as a shallow, wide tree. We develop tree FlashAttention to execute this layout efficiently, processing the shared history once and propagating gradients from every branch for end-to-end training. A tree-structured key/value cache extends history sharing to constrained decoding. Experiments on public benchmarks show that MB outperforms strong recommendation baselines while substantially improving training and decoding efficiency.
32. 【2610.37962】Statistical Detection of Positional Bias in LLM-Based Recommender Systems: Methodologies and Empirical Evidence
链接：https://arxiv.org/abs/2610.37962
作者：Carlos Peláez-González,Francisco Herrera,Andrés Herrera-Poyatos
类目：Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
关键词：Large Language Models, Large Language, conversational recommender systems, Language Models, recommender systems
备注： 34 pages, 8 figures, 18 tables
点击查看摘要
Abstract:Large Language Models (LLMs) are increasingly used as recommendation generators in conversational recommender systems (RS) via retrieval-augmented generation (RAG). This flexible design introduces new sources of structural bias. This paper studies positional bias in top-1 and single-turn LLM-based RS: systematic dependence of the recommended item on the order in which candidates are presented in the prompt. We propose two complementary detection methodologies: (i) POB-ICM, based on contingency matrix analysis and Chi-square independence testing, and (ii) POB-UNBP, a sample-efficient approach derived from a uniformity null model under permutation-based evaluation, enabling stochastic decoding with formal statistical guarantees. Using a controlled RAG testbed for product recommendation, built on a real product catalog, a fixed prompt template and 2000 template-generated user queries with 20 permutations each (40,000 instances), both methodologies reject the null hypothesis of positional independence and uniformity in all evaluated state-of-the-art models. We introduce three complementary metrics -- permutation sensitivity, positional bias, and strong positional bias -- to quantify this effect, the latter targeting cases where the position displaces the recommendation towards clearly less relevant items rather than only choosing among comparably relevant ones. A robustness analysis across candidate-set size and decoding temperature shows that the bias is structural and model-dependent, not an artifact of deterministic decoding or candidate-set size. These results establish positional bias as a measurable, reproducible property of prompt-mediated recommendation pipelines in the evaluated setting and motivate routine auditing and governance controls (e.g., order randomization, standardized prompting) for deployed LLM-based RS. Source code is publicly released.
33. 【2610.37867】What Constitutes Food Similarity? Visual Representations and Retrieval in Indian Cuisine
链接：https://arxiv.org/abs/2610.37867
作者：Saransh Kumar Gupta,Akshadha Sharma,Lipika Dey,Partha Pratim Das,Ramesh Jain
类目：Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
关键词：food computing tasks, computing tasks, typically evaluated, Food, identity preservation
备注： 17 pages, 7 figures, 4 tables, 31 references
点击查看摘要
Abstract:Food similarity underlies retrieval, recommendation, substitution, and other food computing tasks, yet food representations are typically evaluated by identity preservation. We ask what happens to food similarity beyond identity: how different visual representations organize foods into relational neighborhoods, what culinary relationships these neighborhoods reveal or obscure, and how consistently these structures emerge across representations. Indian cuisine provides a challenging setting in which visually similar foods may have different culinary identities, while foods within a culinary family can vary substantially in appearance. Using an expanded Khana dataset and seven pretrained visual representations spanning convolutional, transformer-based, self-supervised, and image-text aligned models, we evaluate image-to-image retrieval and image-to-dish neighborhoods, examining culinary-family structure, contrast cases, and cross-representation agreement. We also aggregate image representations to compare dish-level visual similarity across models and against recipe-derived similarity for the same dishes. Image-level neighborhoods vary substantially across representations, including similarities across culinary-family boundaries. At the dish level, aggregation reveals common visual structure despite local variation, but this structure corresponds less strongly with recipe-derived similarity. These findings show that recognition performance alone cannot characterize relational structure: food similarity depends on representation, modality, and abstraction level rather than being a fixed property of the food collection. Understanding these structures is therefore a prerequisite to combining multiple food information modalities into a coherent notion of similarity.
34. 【2610.37830】Measuring Where Intent Gets Lost in Agentic Smart Contract Synthesis
链接：https://arxiv.org/abs/2610.37830
作者：Abhinav Goel,Agostino Capponi,Alfio Gliozzo,Chaitya Shah
类目：Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Information Retrieval (cs.IR); Multiagent Systems (cs.MA); Software Engineering (cs.SE)
关键词：contract, write software, plain-English description, AI-written contract, code
备注： ACM TheWebConf2027 8 pages main content 4 pages appendix
点击查看摘要
Abstract:AI agents can now write software from a plain-English description, but code that runs is not the same as code that does what was asked. This gap matters most for smart contracts, the programs behind Web3 applications that move money on a blockchain and cannot be fixed once deployed. We ask how to check whether an AI-written contract behaves as it should, and where the requested behavior gets lost. We design an evaluation built on execution rather than opinion. We replay the same transactions on an AI-written contract and on the original it should reproduce and measure how often they agree. We test every grading tool on planted mistakes, compare against an answer key written without the description, and audit our own scorer until known wrong answers cannot pass. Common methods, such as AI judges and standard analysis tools, miss obvious mistakes, and scoring choices matter too. Checking only the functions every system includes hid about half the difference between pipelines. We then propose an elicitation-first architecture. Before writing code, the agent asks the client a short checklist of questions, checks the contract against an independently written executable answer key, and returns disagreements to the client instead of silently rewriting code. Asking first matters far more than reviewing or patching afterwards, across three AI models, while automated security fixes often break requested behavior. A few well-chosen questions recover much of what a short description omits, and their value can be estimated without the original contract. To our knowledge, this is the first study in smart contract synthesis to gather missing requirements before generation and check the result against an independently written executable model. The lessons extend to any setting where AI agents turn requests into programs. We release our tools, 600 tasks and all results.
Comments:
ACM TheWebConf2027 8 pages main content 4 pages appendix
Subjects:
Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Information Retrieval (cs.IR); Multiagent Systems (cs.MA); Software Engineering (cs.SE)
Cite as:
arXiv:2610.37830 [cs.AI]
(or
arXiv:2610.37830v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.37830
Focus to learn more
arXiv-issued DOI via DataCite</p>
35. 【2610.37826】DeFT-RAG: Incrementally Decomposing Multi-Hop Questions into Self-Contained Sub-Questions with Reusable Answers
链接：https://arxiv.org/abs/2610.37826
作者：Dingyi Kang,Bingzhe Li
类目：Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
关键词：retrieving relevant passages, Retrieval-augmented generation, large language model, large corpus, large language
备注： 12 pages, 2 figures, 7 tables. Code: [this https URL](https://github.com/Dingyi-Kang/DeFT-RAG)
点击查看摘要
Abstract:Retrieval-augmented generation (RAG) lets a large language model (LLM) answer questions by retrieving relevant passages from a large corpus. However, RAG struggles with multi-hop questions, which require chaining facts spread across several documents through intermediate entities that the question never mentions. Graph-based RAG methods retrieve once from the entities a question names and so rarely recover the full evidence chain. Iterative RAG methods recover it step by step, but their accuracy is limited by the many tasks they ask the LLM to perform, such as planning the sub-questions, deciding their dependencies, rewriting them as answers arrive, and keeping track of what is known. We propose DeFT-RAG, a RAG framework built on three components. First, DeFT-RAG generates one self-contained sub-question at a time from the original question and the sub-questions answered so far with their answers (QA pairs). Each sub-question names every entity it asks about, which removes the need to plan, track dependencies or rewrite, and lets the LLM focus on a single fine-grained task at a time. Second, DeFT-RAG moves the remaining LLM work outside the LLM: rule-based updates to a QA table that keeps track of what is known, and yes/no decisions made by a fine-tuned lightweight decision model. Third, DeFT-RAG caches the QA pairs and reuses an answer when a later question asks a sub-question with the same meaning. On MuSiQue, 2WikiMultiHopQA and HotpotQA, DeFT-RAG achieves the best accuracy among thirteen baselines with both gpt-4o-mini and Qwen3.6 as the backbone, while giving the LLM fewer input tokens than any other iterative method. With a QA cache shared across 1,000 MuSiQue questions, 23.1% of sub-questions are answered from the cache, which reduces LLM calls by 12.9% without lowering accuracy.
36. 【2610.37728】Beyond Similarity Search: Knowledge Graph Metadata Enrichment for Enterprise Retrieval-Augmented Generation
链接：https://arxiv.org/abs/2610.37728
作者：Fakabbir Amin
类目：Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：organizational documentation responsibility, enterprise documents created, documentation responsibility, key challenge, systems is retrieving
备注：
点击查看摘要
Abstract:A key challenge of RAG systems is retrieving information from enterprise documents created by multiple teams over time, where it is not possible to enrich the information using the latest information and metadata available outside the scope of organizational documentation responsibility. We investigated whether using a knowledge graph to store metadata and frequently changing information would improve upon knowledge retrieval over similarity search over embeddings. To perform the experiment, we created a snapshot of the SEC EDGAR dataset for a quarter, comprising 167,050 filings from 10,511 companies and 508,714 embedded chunks of risk-factor text. To simulate the enrichment, we created a knowledge graph of 430,257 nodes and 561,144 edges carrying an accession number on every edge. To evaluate the gain obtained via a graph-based design, we evaluated three architectures: text-to-SQL over typed columns, dense-vector RAG with HNSW cosine search, and a two-stage GraphRAG design. All architectures share the same model and temperature to allow comparison. Results are scored programmatically by an LLM as judge over 20 questions divided across four categories, each question asked three times. GraphRAG solved 16 of 20 questions on all three attempts, compared to 10 for text-to-SQL and 3 for vector RAG. GraphRAG lost on structured aggregation queries. Vector RAG lost the single-document category it was expected to win, passing 3 of 15 where GraphRAG passed 15 of 15, because company attribution is not recoverable from cosine similarity at this scale. All six confident falsehoods in the benchmark were text-to-SQL conflating distinct people who share a surname, the failure that resolving identity to an identifier before reading any text prevents structurally; GraphRAG made none. We claim no advantage beyond this corpus, this model and this question set.
37. 【2610.37707】Relevance-Guided Residual Calibration for Spectral Compression of Dense Retrieval Embeddings
链接：https://arxiv.org/abs/2610.37707
作者：Yongkang Li,Panagiotis Eustratiadis,Evangelos Kanoulas
类目：Information Retrieval (cs.IR); Computation and Language (cs.CL)
关键词：smaller principal subspace, Spectral compression reduces, principal subspace, compression reduces, scoring costs
备注： Under Review
点击查看摘要
Abstract:Spectral compression reduces the storage and scoring costs of dense retrieval by projecting embeddings onto a smaller principal subspace. However, its corpus-derived weighting rules tie all retained directions to a shared spectral trend, without directly accounting for their individual contributions to relevance. To address this, we propose Relevance-Guided Residual Calibration (RRC), a closed-form method for refining spectral weights using relevance judgments. RRC estimates margin-based weights from the means and variances of positive--negative score contributions along each retained direction. It retains directions with positive estimated weights, decomposes their log weights into a fitted power-law trend and a residual, and adds the residual to the compressor's log weights on this support. This allows relevance judgments to adjust individual directions while preserving the compressor's overall spectral trend. RRC applies to PCA, Whitening, and SpecTemp, and incorporates the corrections into the existing transformation shared by documents and queries, without iterative model training or an additional inference stage. Experiments with three encoders on nine datasets covering conventional and reasoning-intensive retrieval show that SpecTemp + RRC achieves the highest mean effectiveness among tested compressors on the BEIR group and BRIGHT at every target dimension. On the five evaluated BRIGHT datasets, it improves over SpecTemp by an average of 2.34 nDCG@10 points at 768 dimensions. The gains also persist under scalar and product quantization, improving the trade-off between per-document storage and retrieval effectiveness. Our code is available at: this https URL
38. 【2610.37701】DuQUAD: A Dual-View Framework with Quality-Aware Evidence Pruning for Multi-Document Question Answering
链接：https://arxiv.org/abs/2610.37701
作者：Yeonjun Choi,Seyeon Jeong,Jong Wook Kim,Beakcheol Jang
类目：Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
关键词：Large language models, retrieval-augmented generation systematically, generation systematically underutilize, Large language, systematically underutilize evidence
备注：
点击查看摘要
Abstract:Large language models with retrieval-augmented generation systematically underutilize evidence in the middle of multi-document contexts, a phenomenon known as Lost in the Middle. Existing mitigation strategies, such as document reordering and attention steering, can reduce boundary bias but remain brittle in noisy settings, where inaccurate relevance estimation and spurious documents still obscure useful evidence in central positions. We propose DuQUAD, a multi-document QA framework that mitigates positional bias through dual-view reasoning and quality-aware evidence pruning. DuQUAD combines a Local Agent guided by a Structural Fusion Score with a Global Agent that reasons over the full document set, providing both targeted evidence recovery and broad contextual understanding. The framework further filters candidate evidence using explicit quality scoring and refines it at the sentence level to reduce noise. DuQUAD consistently outperforms strong Lost-in-the-Middle mitigation baselines, including hybrid retrieval and reranking, multi-agent reasoning, and context optimization methods, improving accuracy by 3.0% on average over the strongest competing baseline and gold evidence recall by 9.4%, while substantially reducing computational cost compared to existing multi-agent frameworks.
39. 【2610.37697】chATLAS: A Retrieval-Augmented Conversational Assistant for the ATLAS Experiment
链接：https://arxiv.org/abs/2610.37697
作者：Joe Egan,Ben Elliot,Samuel Van Stroud,Alex Sopio,Ki Ryeong Park,Daniele Dal Santo,Zhaohui Li,Runze Li,Jeremy Couthures,Joe George,Cary Randazzo,Juerg Beringer,Daniel Murnane,Gabriel Facini
类目：Information Retrieval (cs.IR); High Energy Physics - Experiment (hep-ex)
关键词：produce extensive internal, Large scientific collaborations, collaborations produce extensive, reuse existing knowledge, extensive internal documentation
备注： 20 pages, 6 figures
点击查看摘要
Abstract:Large scientific collaborations produce extensive internal documentation spread across heterogeneous systems, making it difficult for collaborators to locate and reuse existing knowledge. We present chATLAS, a deployed Retrieval-Augmented Generation (RAG) assistant developed for the ATLAS experiment at CERN. The system provides conversational access to more than 30,000 documents from several collaboration resources, material largely unavailable to general-purpose language models, using hybrid dense--sparse retrieval deployed on CERN infrastructure. A domain-specific benchmark built from automatically generated question--answer pairs is used to evaluate retrieval and generation performance and to guide design choices. We study the selection and domain-specific fine-tuning of an embedding model, contextual retrieval, ranking strategies, and generator choice. Fine-tuning combined with contextual retrieval more than doubles how often the correct document is returned as the top result on the most technically challenging documentation source. The deployed system also includes intent routing, safety handling, Mattermost integration, and an MCP server for future agentic workflows. chATLAS demonstrates that domain-adapted RAG can provide source-attributed, domain-specific access to collaboration knowledge, and offers a practical foundation for scientific collaborations and other organisations that rely on accurate, up-to-date retrieval from proprietary documentation.
40. 【2610.37695】After One-Click Video: The Inflow of Service-Type Content, the Failure of the Recommendation Pool Mechanism, and the BoLe System
链接：https://arxiv.org/abs/2610.37695
作者：Bo Zou,Chao Xu
类目：Information Retrieval (cs.IR); Computers and Society (cs.CY)
关键词：content, large language model, September, language model, large language
备注： 13 pages, 2 figures
点击查看摘要
Abstract:In September 2026, a new large language model was released. Soon after, a new way of making explainer videos became very popular: people asked the model to write a program, and the program produced the video. "One-click video" content then appeared in large amounts on many platforms. On the surface, this is a rise in creative power at the level of the individual. We argue that, at a deeper level, it changes the structure of the content industry. This judgment rests on two observations. First, this wave of content is service-type content. People watch it to meet a need, not to enjoy the content itself. The other kind is work-type content. AI makes both kinds faster to produce. But anyone can come in and make service-type content, while work-type content is limited by the number of people who have something to say. So the ratio between the two slowly becomes unbalanced. Second, free platforms choose content through layers of trial viewing, and this uses up user attention. The total amount of attention is fixed. The unbalanced ratio also dilutes good content. A simplified mathematical model shows that, as long as these conditions hold, the recommendation pool mechanism will fail once new content grows past a certain multiple. Based on this, we discuss one possible way out. Machines and entry rules judge whether service-type content passes the bar. Trusted people judge work-type content. We call this the BoLe system. At the end, we briefly imagine how the two kinds of content may develop from here.
计算机视觉
1. 【2610.40653】Micro-Edits: Computational Support for Fine-Scale Edits in Film
链接：https://arxiv.org/abs/2610.40653
作者：Nhan Tran,Hijung Valentina Shin,Abe Davis,Mackenzie Leake
类目：Human-Computer Interaction (cs.HC); Computer Vision and Pattern Recognition (cs.CV)
关键词：frame matters, high-level narrative, content-level edits shape, direct visual attention, film editing
备注： UIST 2026 Best Paper Honorable Mention. Project page: [this https URL](https://microedits.github.io/)
点击查看摘要
Abstract:In film editing, every frame matters. While macro, content-level edits shape the high-level narrative of a film, small local edits often control timing and direct visual attention. For example, editors can adjust the precise timing of clips to control the length of a meaningful silence, or cut to shots that highlight a specific character's reactions to important moments. These small, detailed edits are slow and tedious to make in standard editing tools because users have to map high-level narrative goals to low-level frame-by-frame adjustments. At the same time, these fine-scale adjustments, which we call micro-edits, can be nuanced and expressive, with different choices serving different creative goals. In this work, we start by conducting a formative study with professional editors to understand the structure of micro-edits and how they are used. We then derive a mathematical formalization for this structure to develop tools that leverage AI and optimization to make exploring micro-edits simpler and more efficient for users. Finally, we validate our tools in multiple user studies and a wide range of video examples.
2. 【2610.40649】Rethinking Neural Thickets Without Weight Perturbations
链接：https://arxiv.org/abs/2610.40649
作者：Nicolas Dufour,Elouan Gardès,Amil Dravid,Patrick Pérez,Loic Landrieu
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
关键词：Thickets hypothesis posits, Neural Thickets hypothesis, random Gaussian perturbations, random Gaussian, Neural Thickets
备注： Project page can be found here: [this https URL](https://kyutai.org/logit-psv)
点击查看摘要
Abstract:The Neural Thickets hypothesis posits that pretrained language models lie in dense neighborhoods of task-specific experts in weight space, which random Gaussian perturbations can uncover. We reproduce the reported gains, but show that they do not require this geometric explanation. Perturbing a single decoder block already recovers much of the improvement, while keeping the model entirely frozen and adding a random bias to its output logits matches or exceeds full-weight perturbations on most reasoning tasks. This suggests an alternative explanation based on output diversity, validation-based selection, and voting. We formalize this view as Perturb-Select-Vote (PSV), a general procedure that generates perturbed predictors, selects high-performing members on a small validation set, and aggregates their predictions. This interpretation connects Neural Thickets to established inference-time methods, including sampling, self-consistency, and verifier-guided selection. A first-order analysis further shows that small weight perturbations act primarily through their induced logit shifts. We extend this analysis to vision, where logit PSV substantially improves performance under domain shift, including on Earth observation, while full-weight perturbations provide little additional benefit. Together, these results suggest that the gains attributed to nearby weight-space experts can instead be obtained through output diversity, selection, and voting. Logit PSV realizes these benefits with a fully frozen model, providing a simple, training-free alternative to full-weight perturbation ensembles across language and vision.
Comments:
Project page can be found here: this https URL
Subjects:
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Cite as:
arXiv:2610.40649 [cs.LG]
(or
arXiv:2610.40649v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.40649
Focus to learn more
arXiv-issued DOI via DataCite (pending registration)</p>
3. 【2610.40647】Read, Allocate, Route: Role-Coupled Tokenization for Pixel-Space Generative Transformers
链接：https://arxiv.org/abs/2610.40647
作者：Jiongfan Yang,Weichen Fan,Haiwen Diao,Zhongang Cai,Ziwei Liu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
关键词：attention increasingly expensive, generative Transformers typically, capturing finer detail, image patch, increasingly expensive
备注： 22 pages, 10 figures; under review at ICLR 2027
点击查看摘要
Abstract:Pixel-space generative Transformers typically use one token per image patch, so capturing finer detail requires more tokens and makes global attention increasingly expensive. We observe that finer detail does not necessarily require more global communication: low-frequency structure benefits from global exchange, while high-frequency detail can be modeled locally. Based on this insight, we introduce SpecSlot, a pixel-space rectified-flow Transformer that decouples output detail from the length of the global attention sequence. SpecSlot expands each spatial anchor into multiple slots, each assigned a disjoint band of DCT coefficients; this separates anchor density from spectral subdivision. Only the lowest-frequency slot from each anchor participates in global attention, while all slots interact locally. This allows the model to increase spectral detail without enlarging the global sequence. A spatial head further adds a local correction to the spectral prediction, while the generative state, target, and loss remain in pixel space. On ImageNet-256, SpecSlot achieves a comparable final FID to a FLOPs-matched JiT control while converging faster. On ImageNet-512, it improves FID by 2.84 over the control with 3.9% fewer FLOPs, and comes within only 0.69 FID of JiT-B/16 at 38.2% lower cost. On text-to-image generation, SpecSlot surpasses JiT-B/8 on DPG-Bench by 1.06 at 40.6% lower forward cost.
4. 【2610.40643】Luggage re-identification: A study of strategies towards a real-world application
链接：https://arxiv.org/abs/2610.40643
作者：Quentin Dhôte,Romaric Audigier
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：lost luggage recovery, abandoned luggage recognition, surveillance for forensics, suspicious or abandoned, border management
备注： 7 pages, 4 figures, 5 tables. Accepted at The 2026 IEEE International Conference on Content-Based Multimedia Indexing (IEEE CBMI)
点击查看摘要
Abstract:Luggage re-identification (ReID) is a critical task in video surveillance for forensics, security or border management (suspicious or abandoned luggage recognition), as well as luggage management (lost luggage recovery). It enables the association of observations of the same luggage across disjoint camera views in complex environments such as airports, train stations and border crossing. While existing approaches primarily focus on appearance-based cropped image matching under relatively clean conditions, real-world scenarios are characterized by frequent occlusions, cluttered background and crowded scenes which significantly degrade performance. Yet, strategies leveraging the complementarity of multiple shots and views from different video streams is overlooked. In this paper, we study strategies for an end-to-end pipeline that operates from multiple video streams to detect and re-identify luggage. We evaluate different variants on a benchmark dataset acquired in real-world environments to address the questions of selecting and merging different views and shots of the same luggage as well as training on different environment datasets for an overall ReID performance improvement in practical applications.
5. 【2610.40640】SonoWorld4D: From One Video to a 4D Audio-Visual Scene via Diffusion Synchronization
链接：https://arxiv.org/abs/2610.40640
作者：Derong Jin,Xiyi Chen,Ming C. Lin,Ruohan Gao
类目：Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD)
关键词：Recent progress, minimal input, generation has extended, challenging problem, problem of generating
备注： Project Page: [this https URL](https://humathe.github.io/sonoworld4d/)
点击查看摘要
Abstract:Recent progress in visual scene generation has extended beyond static 3D scenes toward the more challenging problem of generating dynamic 4D scenes from minimal input. However, sound, despite being intrinsically coupled with dynamic objects and events, remains largely unmodeled in prior work. We introduce Video-to-4DAVScene, a new task that turns a video into an explorable, time-varying 4D audio-visual scene, and present the SonoWorld4D framework for this task. Given a monocular video, our method generates a 360 static background, reconstructs explicit motion trajectories of dynamic, sound-emitting foreground objects, and synthesizes a physically-grounded sound field consistent across multiple sources through dual-space audio diffusion synchronization. The resulting 4D scene supports real-time novel-view visual and spatial audio rendering. Evaluation on real-world dynamic scenes and a user study demonstrates SonoWorld4D generates 4D audio-visual scenes with perceptually realistic, physically grounded spatial audio consistent with scene geometry and visual content.
6. 【2610.40636】Probing Spatial State Representations in Visual Foundation Models
链接：https://arxiv.org/abs/2610.40636
作者：Baiqiao Yin,Zhongang Cai,Ruisi Wang,Qiwei Liang,Lei Yang,Ziwei Liu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：surrounding world, current view, Spatial, object persistence, Abstract
备注：
点击查看摘要
Abstract:An observer sees only part of the surrounding world, yet must keep track of what lies beyond the current view. Spatial perception therefore involves more than recognizing individual images: it calls for a persistent spatial state that is updated under self-motion and used to guide action. Do visual foundation models provide representations that support such a state? We study this question through four levels of spatial capability---spatial structure, object persistence, predictive spatial update, and actionability---using frozen-feature probes on ten representations under streaming egocentric observations of static indoor environments. The resulting capability profiles vary across tasks, network layers, and prefix lengths: strong readout of the visible scene does not ensure accurate object persistence or future-feature retrieval. Matched controls separate readout accuracy from reliance on visual history and motion. Native future prediction retains recoverable spatial structure, but a spatial reader trained on real features does not transfer to predicted features unchanged. Finally, goal-dependent candidate selection does not establish reliable continuous action utility: predicted actions respond to goal changes, yet supplying the correct goal does not reliably improve progress. Across the four levels, object persistence and actionability build on camera-pose readout, whereas feature prediction builds on visible-scene readout. Together, these findings motivate evaluating spatial representations not only by what can be decoded from them, but by whether that information is maintained beyond observation, updated predictively, and used for goal-directed action.
7. 【2610.40632】SymAnt: Grounding Long-Horizon Action Anticipation with Adaptable Symbolic Priors
链接：https://arxiv.org/abs/2610.40632
作者：Fnu Aryan,Yaqi Xie,Joseph Campbell,Woojun Kim,Kwonjoon Lee,Hossein Nourkhiz Mahjoub,Ehsan Moradi-Pari,Katia Sycara,Simon Stepputtis
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
关键词：intelligent systems increasingly, systems increasingly interact, interact with humans, future actions, robots and intelligent
备注： Accepted to WACV 2027. 18 pages (8-page main paper, references, and supplementary appendix), 6 figures, 9 tables. Project page: [this https URL](https://aryanmangal769.github.io/symant/)
点击查看摘要
Abstract:As robots and intelligent systems increasingly interact with humans, they must anticipate long sequences of future actions from only brief observations. While purely neural models typically require extensive observations to predict future actions and remain difficult to interpret and adjust, symbolic priors offer a structured alternative. To this end, we introduce SymAnt, a novel neuro-symbolic approach that unifies the expressiveness of neural networks with the structured reasoning of scene and knowledge graphs. Given a short video, SymAnt performs a joint-search over both graphs to construct a symbolic context, enabling unified reasoning over spatial relationships, object attributes, affordances, and associated actions. This symbolic context serves as a prior for an iterative refinement process that progressively generates and improves the sequence of future actions. By grounding predictions in symbolic context, SymAnt anticipates long-horizon actions from short observation windows in a parameter-efficient manner, while enabling users to directly edit the underlying knowledge to adjust predictions without further training. Through our experiments, we demonstrate the effectiveness of SymAnt, particularly when only short observation contexts are available, across five diverse datasets: Breakfast, 50 Salads, EPIC-KITCHENS, EGTEA Gaze+, and Assembly101.
8. 【2610.40616】Difficulty Calibration Across Time for Membership Inference in Diffusion Models
链接：https://arxiv.org/abs/2610.40616
作者：Michael Joseph,Yedid Hoshen
类目：Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
关键词：Modern diffusion models, Modern diffusion, trained on massive, massive datasets, include copyrighted
备注：
点击查看摘要
Abstract:Modern diffusion models are trained on massive datasets that often include copyrighted or private images. Membership inference attacks (MIAs) address this by testing whether a given image was in a model's training set. The standard approach scores an image by its reconstruction loss and flags low-loss images as members. This often fails because some images are simply easier to reconstruct than others, so a simple unseen image can score a lower loss than a complex training image. Difficulty calibration separates memorization from how difficult an image is to begin with. We show how to calibrate along two axes of time: training time and diffusion time. First, across training time, we introduce Reference-Model Calibration (RMC), which compares a fine-tuned model's loss against an earlier checkpoint. This cleanly isolates newly memorized data and gives near-perfect auditing, even without text captions. Since earlier checkpoints are not always available, we next present Temporal Self-Calibration (TSC), which calibrates across diffusion time using only a single model. Our key observation is that an image's difficulty persists across noise levels, whereas memorization is localized in time. By comparing the loss at a target timestep against neighboring timesteps from the same model, TSC attenuates image difficulty using the model's own trajectory. Empirically, our methods beat the state of the art by a wide margin, and unlike most existing attacks, generalize across different model parameterizations such as $x$-, $v$-, and $\epsilon$-prediction models. Code is available at this https URL.
9. 【2610.40614】L3G-World: Real-Time Infinite World Generation with Latent 3D Gaussian Memory
链接：https://arxiv.org/abs/2610.40614
作者：Jinhyeok Choi,JoungBin Lee,Jongmin Lee,Minkyeong Jeon,Jiyoung Kim,Kaeun Han,Yoonjin Yoon,Seungryong Kim
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Interactive world models, visual content conditioned, time remains challenging, models generate visual, generate visual content
备注： Project page: [this https URL](https://cvlab-kaist.github.io/L3G-World/)
点击查看摘要
Abstract:Interactive world models generate visual content conditioned on user actions, yet maintaining spatial consistency over long-horizon exploration in real time remains challenging. Existing approaches rely on either temporal context or explicit 3D memory, both of which scale poorly because memory grows with the length of the generated stream. Moreover, explicit 3D memories condition latent video generators by re-encoding RGB renderings, which precludes real-time generation. In this paper, we introduce a Latent 3D Gaussian Memory (L3G) that combines the efficiency of feed-forward reconstruction with a compact, rapidly renderable scene representation. Building on this memory, we propose L3G-World, in which a memory updater and a video diffusion transformer (DiT) alternate autoregressively to generate spatially consistent worlds. The memory updater spawns Gaussians in newly observed regions and refines existing ones, so that the memory grows with scene extent rather than stream length. Each Gaussian further stores a VAE latent feature, allowing the memory to be rendered directly into the DiT's latent space without VAE re-encoding. Experiments show that L3G-World achieves state-of-the-art generation quality and long-horizon spatial consistency, with up to $38.6\times$ fewer 3D primitives, $3.2\times$ faster memory updates, and $359\times$ faster memory rendering relative to the most efficient prior 3D-memory baseline for each metric, enabling real-time world generation at 9 FPS.
10. 【2610.40607】Fight World Model: Customizable Combat through Visual Interaction Proxies
链接：https://arxiv.org/abs/2610.40607
作者：Jiahua Li,Zixu John,Tom Zhong,Fuping Wu,Tianhao Xu,Ruicheng Zhang,Jun Zhou,Zunnan Xu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：extensible gameplay mechanics, enforce executable rules, Fight World Model, programmable game world, support extensible gameplay
备注： Tech Report
点击查看摘要
Abstract:A playable, programmable game world must maintain persistent state, enforce executable rules, and support extensible gameplay mechanics. Existing video world models can generate plausible combat, yet struggle to keep hit reactions and health updates consistent with game rules. We introduce Fight World Model (FWM), which separates rule-governed combat execution from video rendering. An executable engine maintains state and resolves interactions. A deterministic compiler encodes these outcomes and attempted actions in a spatially and temporally aligned interaction proxy. The proxy aligns action and skill windows, brief hit reactions, and persistent health by fighter identity and timing. These cues can be controlled separately to render interactions under modified rules. Identity-bound references preserve character appearance across side switches. We train FWM on FightWorld Corpus, spanning 52 skill effects, 187 character-appearance variants, and 15 arenas. Without retraining the renderer, FWM supports damage scaling, limited-hit immunity, lifesteal, and multi-stage skill mechanisms beyond the native game. It also allows character and arena customization. Experiments demonstrate the effectiveness of FWM, which outperforms existing game world models in action fidelity, hit reactions, health-state consistency, and character appearance preservation. Our project is available at this https URL.
11. 【2610.40599】HairDeck: Animation-Ready Stylized Hair Cards from a Single Image
链接：https://arxiv.org/abs/2610.40599
作者：Yuze He,Yixiang Wei,Wang Zhao,Chuhao Chen,Yanning Zhou,Yong-Jin Liu
类目：Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
关键词：artist-placed ribbons, animated by spring, spring bones, characters in games, games and avatars
备注： 27 pages, 23 figures
点击查看摘要
Abstract:The hair of stylized characters in games and avatars is built from hair cards: artist-placed ribbons whose geometry is the visible hair, animated by spring bones. Existing image-based methods produce realistic hair or a single fused mesh, while CHARM, the only generator of stylized hair cards, decodes a hairstyle autoregressively as a long sequence of control points. Observing that a card hairstyle is an unordered set of artist-defined primitives, we propose HairDeck, which generates it as one: each card is a single token, and a set diffusion transformer denoises all cards together. From a single image, HairDeck generates the cards and the spring-bone chains that drive them, yielding an animatable asset that exports to ready-to-use downstream formats. On CHARM's own test set, it lowers Chamfer distance by 44% and card-orientation error by more than half while running 33 times faster, and from single images it outperforms CHARM, image-to-3D pipelines and strand-based methods. Project page: this https URL
12. 【2610.40587】Beyond Ground Truth: A Heuristic Unsupervised Quality Assurance Engine for Diagnosing Calibrational Blindness in Dental Segmentation
链接：https://arxiv.org/abs/2610.40587
作者：Fatematuj Johura Mim
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：real-world clinical settings, reliability issues, Periapical Dental X-ray, dental X-ray, suffers from reliability
备注： Accepted at 11th IEEE Asia-Pacific Conference on Computer Science and Data Engineering (IEEE CSDE 2026), Cox's Bazar, Bangladesh. Copyright 2026 IEEE. 5 pages, 5 figures, 3 tables
点击查看摘要
Abstract:In real-world clinical settings, dental X-ray segmentation using deep learning methods suffers from reliability issues. Even when the segmentation quality is bad, deep learning models show high confidence in their predictions (calibrational blindness). Moreover, in real use, doctors do not have ground-truth masks to check their accuracy. To address this problem, we introduce a simple quality-checking engine called HUS-QA with three stages. We train it on the PRAD (Periapical Dental X-ray) dataset to build our SimCLR self-supervised ResNet-18 U-Net model, which generates the segmentation mask and then evaluates its quality. Our HUS-QA checks the segmentation mask quality without a ground-truth mask. Prediction confidence, edge sharpness, shape connectivity, and area-size penalty are the simple indicators considered in this research. Results show that standard confidence methods fail (r = -0.1394, p = 1.7864e-3, ROC AUC = 0.4275). By contrast, our HUS-QA shows a positive correlation with actual Dice scores (r = 0.4647, p = 3.7510e-28) and achieves a strong ROC-AUC of 0.8405. Therefore, HUS-QA serves as a practical tool for clinical quality control.
13. 【2610.40580】RADIOlogy: A Single-Pass Approach to Training-Free Zero-Shot Anomaly Detection
链接：https://arxiv.org/abs/2610.40580
作者：Joaquín Figueira,Giacomo D'Amicantonio,Egor Bondarau
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：detect small-scale anomalies, computationally expensive sliding-window, expensive sliding-window strategies, achieve fine-grained visual, fine-grained visual alignment
备注： Accepted to WACV 2027. Code available at [this https URL](https://github.com/jacomof/RADIOlogy)
点击查看摘要
Abstract:Current zero-shot anomaly detection methods often rely on complex prompt-learning techniques or computationally expensive sliding-window strategies to achieve fine-grained visual alignment to detect small-scale anomalies. This reliance stems from the fundamental limitation of standard CLIP architectures, where [CLS] token attention and spatial patch pooling struggle to capture subtle, localized anomaly features. While prompt-learning approaches aim to bridge this gap, they typically require a large number of ground-truth segmentation masks from auxiliary datasets, which are often unavailable in specialized domains. In this work, we show an emerging property of multi-head attention pooling in vision-language models: remarkable zero-shot anomaly localization and detection right out of the box. We translate this property into a single-pass anomaly detection technique, RADIOlogy, that does not require dense, windowed execution and significantly outperforms other training-free baselines. Instead, the method relies on a simple contrastive scoring approach that compares the average similarity between vision features and a suite of normal and abnormal prompts. Furthermore, we leverage the agglomerative features of RADIO models to enhance pixel-level anomaly segmentation using a dense anomaly refinement pipeline. Finally, we show how the vision-language family of training-free zero-shot anomaly detection techniques can be applied successfully to the medical domain with minimal modifications. Code is available at this https URL.
14. 【2610.40578】VeriCUA: Verification as Agentic Exploration for Computer-Using Agents
链接：https://arxiv.org/abs/2610.40578
作者：Qiushi Sun,Da Yu,Alexander Knop,Amer Sinha,Lingpeng Kong,Ben Kao,Ravi Kumar,Pritish Kamath
类目：Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：scaling computer-using agents, requires reliable task-completion, web environments requires, environments requires reliable, Building and scaling
备注： Work in progress
点击查看摘要
Abstract:Building and scaling computer-using agents (CUAs) across desktop, mobile, and web environments requires reliable task-completion verification for data curation, evaluation, and reinforcement learning. Typical automated approaches prompt vision-language models (VLMs) to judge trajectories or generate programs to inspect application state. However, both rely on partial views of complex digital environments. VLM judges are restricted to supplied observations, while generated programs depend on accessible state. These views can omit decisive evidence or leave completion judgments unsupported. Tailoring separate verifiers to different platforms and access settings adds further engineering cost. We present VeriCUA, a platform-agnostic framework for verification through agentic exploration. It treats verification as dynamic evidence gathering and assessment within a shared procedure for online and offline settings across platforms and observation modalities. After inspecting the available context to construct a task rubric, the verifier selects GUI or code checks for unresolved requirements and combines their findings under explicit evidence rules. Experiments show improved agreement with reference labels over the compared baselines across these settings, with the largest offline gains on hard cases and stronger rejection of false completion claims online. Further analyses connect these gains to complementary evidence and test-time exploration.
15. 【2610.40530】PhysCount: Do MLLMs Count Visual Appearances or Physical Entities?
链接：https://arxiv.org/abs/2610.40530
作者：Lidong Lu,Jiaxin Lv,Liangrui Dong,Xinyi Mao,Lidong Zou,Zhenquan Li,Tong Lu,Guo Chen
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：appearances, Images reveal appearances, Images reveal, physical objects, counting
备注：
点击查看摘要
Abstract:Images reveal appearances, but counting often asks about physical objects. A single entity may be visible directly, in several reflections, or through glass, so detecting every visible occurrence does not determine how many entities are present. Existing visual counting benchmarks rarely separate these two operations. We introduce PhysCount, a benchmark for evaluating whether multimodal large language models (MLLMs) can distinguish what appears in an image from what physically exists in the scene. PhysCount contains 546 real-world images and 4,595 human-written questions organized into 651 target-specific groups. For each group, annotators locate the relevant appearances, label how each is observed, and identify appearances of the same entity. Four complementary question families test the enumeration of appearances, the enumeration of physical entities, image-formation-specific counting, and appearance--entity correspondence. We measure both accuracy on each family and the proportion of groups answered entirely correctly. Across 40 MLLMs, counting visible appearances is substantially easier than recovering the identities that connect them. Even the strongest model answers every question correctly in only 13.36\% of groups. Performance declines when one entity produces many appearances or when its appearances cross observation modes. Higher image resolution improves access to visual evidence but brings limited gains in complete-group accuracy, whereas answering related questions jointly produces a marked improvement. These results expose a gap between recognizing what is visible and correctly recovering the physical entities behind those observations.
16. 【2610.40525】CrossWeave: Cross-Modal Scene and Instance Retrieval from Sparse 2D-3D Alignment
链接：https://arxiv.org/abs/2610.40525
作者：Aadith Warrier,Gnana Prakash Punnavajhala,Siddharth Tourani,Muhammad Haris Khan,Avinash Sharma,Madhava Krishna
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：instance retrieval aims, aims to retrieve, foundational mechanism, mechanism for spatial, spatial reasoning
备注： Project Page: [this https URL](https://crossweave3d.github.io/)
点击查看摘要
Abstract:Cross-modal scene and instance retrieval aims to retrieve a corresponding 3D environment or object given query 2D images, serving as a foundational mechanism for spatial reasoning in robotics and AR/VR. Existing methods achieve this by learning alignment only at the level of whole scenes or objects, which provides no supervision for the fine-grained local structure within them. In this work, we show that supervision on a set of sparse 2D-3D correspondences, without scene- or instance-level contrastive objectives, is sufficient to train a single model for both scene- and instance-level retrieval. Specifically, we propose $\textbf{CrossWeave}$, a novel framework for cross-modal retrieval that learns sparse patch-point alignment, which alone outperforms existing methods. An optimal transport-based aggregator over the aligned features gives further gains by clustering structurally redundant regions, freeing descriptor capacity for semantically distinct local features. CrossWeave achieves state-of-the-art performance on both scene- and instance-level retrieval on ScanNet and 3RScan with a $\textit{single model}$, whereas existing methods require separate models for each task. Our results suggest that whole-scene encoding is unnecessary for cross-modal retrieval and that sparse local correspondence supervision is a more effective and flexible alternative.
17. 【2610.40495】LucidLyra: Feed-Forward Generative 3D Scene Reconstruction and Joint Understanding via Multi-Task Distillation
链接：https://arxiv.org/abs/2610.40495
作者：Oliver Hahn,Rodrigo Marcuzzi,Sangeetha Grama Srinivasan,Alessandro Burzio,Sherwin Bahmani,Jiahui Huang,Stefan Roth,Shengyu Huang,Qunjie Zhou,Laura Leal-Taixé
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：makes generatively reconstructed, generatively reconstructed, framework that makes, makes generatively, feed-forward manner
备注： Project Page: [this https URL](https://research.nvidia.com/labs/sil/projects/lucidlyra)
点击查看摘要
Abstract:We present LucidLyra, a framework that makes generatively reconstructed 3D scenes directly queryable for semantics and instances in a feed-forward manner from video latents. While generative 3D reconstruction can synthesize virtual environments from a single image, downstream applications in robotics and physical AI require not only visual realism but also access to scene semantics and object identities. Existing 3D understanding approaches, however, require captured multi-view scenes, have limited generalization capabilities, or rely on expensive per-scene processing and optimization. To bridge this gap, our method distills complementary knowledge from a camera-controlled video diffusion model and frozen 2D vision foundation models into a unified model to jointly infer 3D Gaussians encoding geometry, appearance, open-vocabulary semantics, and class-agnostic instance cues. The resulting scenes support novel-view rendering, open-vocabulary segmentation, and object grouping without per-scene optimization, while training requires neither manual 3D annotations nor captured multi-view data. Our method outperforms baselines combining state-of-the-art 3DGS segmentation approaches with generative 3D scene reconstruction on curated synthetic and real-image-conditioned benchmarks, while preserving reconstruction quality. Furthermore, our latent formulation naturally extends to monocular video, reconstructing structured dynamic 3D scenes across time. By turning renderable but unstructured outputs into queryable 3D worlds, LucidLyra enables downstream applications including 3D scene editing, 3D asset mining, and robotic simulation.
18. 【2610.40487】MetaRefine: Score Refinement with Multi-Level Features for Image Assessment
链接：https://arxiv.org/abs/2610.40487
作者：Chuanzhi Xu,Chengkun Yue,Haolan Guo,Xuanhua Yin,Boyu Wei,Zihan Deng,Qiang Qu,Weidong Cai
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：composition provide important, provide important cues, aspect ratio, image original dimensions, aesthetic quality
备注：
点击查看摘要
Abstract:An image's original dimensions, aspect ratio, and composition provide important cues that can affect its technical and aesthetic quality. Image assessment methods use resizing, cropping, padding, or patch sampling to accommodate different input sizes. However, these operations can alter or obscure original geometric and compositional cues. Existing assessment models also encode rich multi-level visual evidence that can be further extracted and exploited to improve existing models. This motivates us to combine visual evidence with original-image geometry for score refinement without changing the original scoring pipelines. In this paper, we introduce MetaRefine, a metadata-aware score refinement architecture whose lightweight residual heads predict score corrections from geometry, multi-level visual features, and the frozen model's original score. Across 100 configurations on 21 datasets, MetaRefine improves PLCC in 88 configurations and SRCC in 85, with average relative gains of 7.55% and 7.10%, respectively. It reduces MAE in 90 configurations and RMSE in 89, with average relative reductions of 23.27% and 21.93%, respectively. The median parameter overhead is only 1.11%. MetaRefine can further improve state-of-the-art (SOTA) assessment models on multiple benchmarks.
19. 【2610.40480】AVPS-360: A Spatial Audio-Visual Benchmark for Instance Segmentation and Tracking of Sound-Producing Persons
链接：https://arxiv.org/abs/2610.40480
作者：Yifu Sun,Parthasaarathy Sudarsanam,Tuomas Virtanen,Archontis Politis
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：tracks individual sound-producing, individual sound-producing objects, segments and tracks, tracks individual, individual sound-producing
备注：
点击查看摘要
Abstract:Audio-visual instance segmentation (AVIS) segments and tracks individual sound-producing objects in videos. Existing benchmarks predominantly use monaural audio, which lacks inter-channel cues for spatially distinguishing concurrent sound sources. We introduce AVPS-360 (Audio-Visual Person Segmentation 360), a spatial audio-visual benchmark bridging AVIS and sound event localisation and detection (SELD). Built upon STARSS23, it pairs first-order Ambisonics audio with synchronised 360° video across 230 recordings totalling 10.5 hours. Recordings average 164.0 seconds. The dataset captures intermittent sound production in real-world interactions dominated by speech, alongside laughter, clapping, and footsteps. It provides 31,678 manually reviewed instance masks linked to source identities and sound-event labels. We also present a baseline that integrates spatial audio with visual features for frame-level segmentation and identity association across frames. Experiments on AVPS-360 show that existing AVIS methods struggle with detection and identity association in long recordings with intermittent sound production, while our baseline substantially improves performance in both aspects.
20. 【2610.40478】Accurate MRI-free PET Quantification and Reliability Assessment via CT-to-MRI Synthesis and Conformal Prediction
链接：https://arxiv.org/abs/2610.40478
作者：Jungwoo Seo,Seung Yun Choi,Kun Ho Lee,Seun Jeon,Byoung Seok Ye,Jiook Cha
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：amyloid PET commonly, PET commonly relies, Standardized uptake, PET quantification, MRI-free PET quantification
备注： 28 pages, 8 figures, including supplementary material. Submitted to Medical Image Analysis
点击查看摘要
Abstract:Standardized uptake value ratio (SUVR) quantification of amyloid PET commonly relies on a subject-specific T1-weighted MRI, yet a usable MRI is not always available. We therefore combined CT-to-MRI synthesis with conformal prediction for MRI-free PET quantification with reliability assessment. A slice-consistent Brownian Bridge Diffusion Model was strengthened with multi-dataset training and data-source conditioning to enrich the anatomical prior learned from MRI, and with expanded multi-slice CT context to exploit the through-plane continuity of subject-specific anatomy in CT. Synthetic MRI was used for FreeSurfer-based anatomical parcellation and PET quantification, while deep-ensemble uncertainty was calibrated with conformal prediction to estimate a plausible range for the corresponding real MRI-based SUVR. On $^{11}$C-PiB PET from OASIS-3, the proposed method achieved the highest synthesis and parcellation fidelity among the compared methods. Mean cortical SUVR from the proposed synthetic MRI agreed closely with the real MRI-based reference (MAE $0.016 \pm 0.015$, Pearson $r = 0.9987$) and yielded the same amyloid-positivity classification as the real MRI in all 87 test subjects at the prespecified cut-off. Conformal prediction achieved empirical coverage close to the prespecified levels, and at 90% target coverage, 85 of 87 subjects had prediction intervals that supported an unambiguous amyloid-positivity decision. Zero-shot evaluation on an independent external cohort showed that the proposed CT-to-MRI model could support PET quantification without retraining, although conformal calibration required target-cohort recalibration to recover the intended coverage. These findings support CT-to-MRI-based MRI-free PET quantification as a promising approach for providing accurate SUVR measurements and an explicit assessment of their reliability.
21. 【2610.40467】InfantNet: A Large Scale Dataset for Infant Body Pose and Shape Estimation
链接：https://arxiv.org/abs/2610.40467
作者：Zhipeng Hui,Jiahao Wang,Xun Dong,Guofeng Zhang,Xingrui Wang,Jiawei Peng,Qihao Liu,Xiaoding Yuan,Yi Zhang,Junjie Oscar Yin,Wufei Ma,Qinfan Lyu,Sara N. Lupolt,Keeve Edward Nachman,Adam Kortylewski,Alan Yuille
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：developmental monitoring, medical diagnosis, Infant pose, pose, Infant
备注：
点击查看摘要
Abstract:Infant pose and shape estimation is essential for applications in childcare, developmental monitoring, and medical diagnosis. However, existing methods and datasets are largely designed for adults, and direct transfer to infants fails due to substantial differences in body proportions, articulation limits, and frequent self-occlusion. To address this gap, we introduce InfantNet, the largest real-image infant dataset to date, comprising 108,902 RGB images of infants aged 6-18 months. Each image is annotated with 2D keypoints, and a curated subset of 11,642 images additionally includes 3D pose and shape annotations with full SMIL parameters, verified through multi-view rendering and reprojection-error filtering. We use an iterative annotation pipeline to ensure high fidelity across both 2D and 3D labels. InfantNet establishes a large-scale, comprehensive benchmark for infant 2D keypoint detection and 3D pose-and-shape recovery. Baseline experiments demonstrate that state-of-the-art adult pose estimators do not generalize well to infants, whereas fine-tuning on InfantNet yields a consistent improvement. The gains are even more pronounced for 3D pose and shape estimation, where SMIL-based estimation demonstrates robust performance with accuracy comparable to results reported for SMPL-based models. By releasing the InfantNet dataset and benchmark, we provide a vital resource for advancing infant pose analysis and related healthcare applications.
22. 【2610.40455】A Unified Benchmark for Aerial-Enhanced Autonomous Driving: The KITScenes Aerial Dataset
链接：https://arxiv.org/abs/2610.40455
作者：Daniel Lengerer,Richard Schwarzkopf,Jonas Merkert,Fabian Immel,Rupert Polley,Frank Bieder,Jan-Hendrik Pauls,J. Marius Zöllner,Klaus Bogenberger,Christoph Stiller,Carsten Markgraf
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：exposing road structure, Aerial imagery, Aerial, autonomous driving, driving by exposing
备注： Accepted at the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2027)
点击查看摘要
Abstract:Aerial imagery provides complementary spatial context for autonomous driving by exposing road structure beyond the immediate perception range of onboard sensors. Yet map construction from aerial imagery, aerial-assisted online HD mapping, and cross-view localization are typically studied on separate data and benchmarks. We introduce KITScenes Aerial, an extension of the KITScenes Multimodal dataset that augments synchronized onboard sensor data, vehicle poses, and rich Lanelet2 map annotations with spatially associated aerial imagery and higher-resolution neural representations, providing a common basis for a variety of research tasks. Our mapping evaluation focuses on all 53 KITScenes map classes across different model designs, spatial ranges, and aerial resolutions, while retaining compatibility with the widely used three-class setting. Representative baselines show that aerial context is particularly beneficial for fine-grained and extended-range mapping, more than doubling 53-class mAP at the larger mapping extent. Higher-resolution aerial imagery yields a further 4.3 mAP (+24 %) improvement, with gains extending to challenging map elements such as traffic lights. We additionally evaluate cross-view localization under geographically disjoint conditions, where an adapted fine-grained feature-matching baseline achieves a mean translation error of 5.13 m. KITScenes Aerial thus enables unified study of aerial and ground-level information across complementary mapping and localization tasks. Project page: this https URL
23. 【2610.40453】PhysInsert: Physically Plausible Object Insertion with Geometric Priors and Bidirectional Object-Scene Interactions
链接：https://arxiv.org/abs/2610.40453
作者：Yue Gong,Le Zhang,Rui Chen,Song Chang,Xiaochao Qu,Luoqi Liu,Ting Liu,Yebin Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：physically plausible reference-based, plausible reference-based object, present PhysInsert, surrounding scenes, reference-based object insertion
备注： 51 pages, including supplementary material. Project page: [this https URL](https://arthuring.github.io/PhysInsert/)
点击查看摘要
Abstract:We present PhysInsert, a framework for physically plausible reference-based object insertion that jointly adapts objects and their surrounding scenes. Existing methods often rely on masked inputs, generation-based pairs, or training pairs with limited viewpoint variation, leaving pose and illumination adaptation and bidirectional object-scene interactions insufficiently constrained. We construct the Insert World dataset with strictly aligned Blender-rendered counterfactual pairs and object references across viewpoints and lighting conditions. Building on these data, Geometry-Guided Adaptation couples a VGGT geometry stream with image and text streams under target-scene feature supervision to guide spatially consistent insertion. Complementary 2D Object-Scene Spatial Supervision jointly supervises object occupancy and environmental changes while grounding spatial instructions in object regions through image-text attention. We introduce PhysInsert-Bench with dedicated cross-environment and cross-view cases requiring illumination and pose adaptation beyond appearance copying. Experiments on PhysInsert-Bench and DreamBooth demonstrate improved image fidelity; a user study further shows more accurate placement and improved lighting integration, perspective, and pose plausibility.
24. 【2610.40432】Soil Moisture Estimation in Precision Agriculture: A Knowledge Distillation Approach
链接：https://arxiv.org/abs/2610.40432
作者：Mohammed Rakib,Vrinda Nagula,Arunkumar Bagavathi
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：Accurate soil moisture, Accurate soil, influencing irrigation, precision agriculture, harvesting strategies
备注： Accepted by MWAIS 2025
点击查看摘要
Abstract:Accurate soil moisture estimation is critical for precision agriculture, influencing irrigation, fertilization, and harvesting strategies. Traditional methods primarily rely on meteorological and soil property data, while recent approaches incorporate visual modalities such as aerial and geospatial imagery. However, these high-resolution images are expensive and computationally demanding to analyze. In this work, we propose Soil Moisture Distillation (SMD), a Knowledge Distillation (KD)-based multimodal learning framework that leverages unimodal teachers to enhance multimodal soil moisture estimation. SMD surpasses not only traditional multimodal fusion techniques, including MIS-ME, but also outperforms unimodal models, demonstrating the benefits of guided unimodal knowledge transfer. Additionally, we conduct a comparative analysis of soil patches and crop images, showing that crop images provide superior visual cues, leading to improved prediction accuracy. SMD achieves a 2% improvement in MAPE for soil patches and 1.2% for crops over state-of-the-art baselines, highlighting the effectiveness of leveraging both meteorological and visual data through knowledge distillation for enhanced soil moisture forecasting.
25. 【2610.40421】Uncertainty-aware Feedback Loop Diagnosis: integrating reinforcement learning and conformal prediction for reliable wafer defect classification
链接：https://arxiv.org/abs/2610.40421
作者：Aleksandr Dekhovich,Carlas Smith,Oleg Soloviev
类目：Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
关键词：low operating costs, Semiconductor manufacturing, operating costs, high precision, precision and low
备注：
点击查看摘要
Abstract:Semiconductor manufacturing is a complex, multi-step process that requires both high precision and low operating costs. Therefore, detecting defects in a high-volume setting is an important component of wafer map analysis for reliable production. However, most state-of-the-art deep learning approaches require a full scan of the surface, which hinders the development of high-throughput inspection systems. To address this problem, we propose Uncertainty-aware Feedback Loop Diagnosis (UFLD), a novel algorithm that employs reinforcement learning for adaptive measurement acquisition and provides a prediction uncertainty level after each step, stopping the process once sufficient information is obtained. Unlike existing methods that rely on ensemble models or auxiliary confidence estimation deep learning models, UFLD utilizes a more parameter-efficient conformal prediction framework to determine the optimal stopping time. The proposed UFLD reduces acquisition time by about $40\%$ and achieves $98.4\%$ inspection accuracy, without requiring an additional learned confidence-estimation network. We validated our approach on the MixedWM38 dataset, demonstrating its advantages over existing adaptive acquisition procedures and standard convolutional neural networks.
26. 【2610.40401】Continuous-Action World Models: Learning Cross-Embodiment Spatiotemporal Dynamics Priors
链接：https://arxiv.org/abs/2610.40401
作者：Liangyu Li,Qingwen Liu
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：Continuous-Action World Models, introduce Continuous-Action World, World Models, Continuous-Action World, cross-embodiment spatiotemporal dynamics
备注： 12 pages, 6 figures
点击查看摘要
Abstract:We introduce Continuous-Action World Models (CoAWM), a framework for learning cross-embodiment spatiotemporal dynamics priors that support robot action generation across embodiments and execution frequencies. At its core is a shared latent function space that captures the spatial and temporal structure of robot actions. We learn this space through two stages. First, function pretraining brings trajectories from different embodiments and sampling frequencies into a common representation, learning shared dynamics priors through action reconstruction and consistency across temporal sampling views. By accommodating heterogeneous robot trajectories, this stage broadens the range of data sources available for pretraining and facilitates scaling to larger and more diverse datasets. Second, we connect the pretrained function space to a world-model backbone, enabling observations, language instructions, and the current robot state to condition the prediction of a latent action function. A single decoder takes any query time and returns the corresponding end-effector action. By separating shared dynamics priors from task-conditioned action selection and temporal sampling, our approach provides a unified foundation for transferring robot experience across embodiments and control frequencies.
27. 【2610.40373】HiFOR: Hierarchical Foresight with Action-Guided World Grounding for Vision-Language-Action Model
链接：https://arxiv.org/abs/2610.40373
作者：Rongxing Ding,Hongyu Qu,Kehong Wu,Xudong Wang,Fan Zhang,Xinguang Xiang,Xiangbo Shu
类目：Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
关键词：integrating visual perception, demonstrated promising capabilities, visual perception, demonstrated promising, promising capabilities
备注：
点击查看摘要
Abstract:Vision-Language-Action (VLA) models have demonstrated promising capabilities in robotic manipulation by integrating visual perception, language understanding, and action generation. Recent advances have further incorporated future prediction to enhance the understanding of environmental dynamics and task progression. However, existing methods often model short-term dynamics and long-term task objectives separately, while directly incorporating predicted futures into action policies without sufficiently considering their relevance to current action requirements. To address these limitations, we propose HiFOR, a hierarchical foresight framework with action-guided world retrieval. Specifically, Hierarchical Foresight Prediction jointly predicts short-term state evolution, semantic milestones, and final goal states through a Hierarchical Dependency Mask (HDM), establishing dependencies across temporal horizons. Building upon these predictions, Structure-Aware Latent World Modeling organizes future latent representations along the horizon, view, and spatial dimensions to facilitate fine-grained retrieval. Furthermore, Action-Guided Factorized World Retrieval adaptively retrieves relevant future information from the structured latent world based on evolving action representations during denoising, providing targeted guidance for action generation. Extensive experiments on two simulation benchmarks and six real-world robotic manipulation tasks demonstrate the effectiveness of our approach.
28. 【2610.40339】Wizard: Manipulate Any World in Real Time via Agentic Control
链接：https://arxiv.org/abs/2610.40339
作者：Yuzhe Wu,Zijun Lin,Yeying Jin
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Recent advances, increasingly realistic, realistic and interactive, interactive virtual environments, Recent
备注： Project Page: [this https URL](https://wyzardwu.github.io/wizard)
点击查看摘要
Abstract:Recent advances in video world models have enabled the generation of increasingly realistic and interactive virtual environments, opening new possibilities for player-customizable gameplay. Agentic frameworks further advance this direction by decoupling explicit game logic and controllable world representations from visual generation. However, existing approaches support only simple game settings and predefined motion patterns, generalize poorly to unseen environments and characters, and offer limited real-time interactivity. In this paper, we introduce Wizard, an agentic game world model that enables real-time, precise control over character motions while generalizing zero-shot across novel scenes, characters, and actions. To achieve this, we construct a diverse dataset spanning 750 scenes and 178 characters, incorporating skeleton and geometry information as structured representations. We leverage these representations as intermediate proxies to condition a video generation model, together with an adaptive masking mechanism that improves alignment between the conditioning signals and generated video rollouts. Furthermore, we introduce an agentic framework that explicitly maintains and updates game states, allowing players to freely configure their virtual worlds and exercise fine-grained control over character motions while preserving visual quality and gameplay consistency. By combining agentic control with structured proxy-based generation, our framework unlocks new possibilities for player-driven game creation with real-time, generalizable, and mechanically consistent world simulation.
29. 【2610.40333】RFW-A: Agentic Procedural Generation of 400K Articulated Assets over 500 Categories
链接：https://arxiv.org/abs/2610.40333
作者：Yuanbo Lyu,Changjian Pan,Saining Zhang,Yi Yang,Peishuo Li,Zhiheng Li,Jialu Liu,Junhao Chen,Jiawei Weng,Yupeng Gao,Weiqing Xiao,Wenyi Li,Jiarui Guo,Rui Zhang,Jingrun Du,Weiyi Lin,Zenan Meng,Hu Wei,Lin Qu,Zongzheng Zhang,Hao Zhao,Chaojian Li
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：scale remains difficult, High-quality articulated assets, High-quality articulated, embodied perception, remains difficult
备注： 37 pages, 19 figures. Project Page: [this https URL](https://brain-labbb.github.io/RFW_A_web/)
点击查看摘要
Abstract:High-quality articulated assets are essential for embodied perception, simulation, and manipulation, yet constructing them at scale remains difficult. Manual annotation and procedural modeling require substantial human effort. Recent agentic systems automate asset construction through program synthesis, but repeat costly reasoning and refinement for every new asset. Our key observation is that assets within a category can share construction logic despite variations in structure and appearance. Agentic synthesis and procedural generation are therefore complementary: agents can construct reusable programs, while procedural execution reuses the resulting logic across many samples. We therefore present RFW-A, which combines the two in a new articulated-asset generation paradigm. Starting from existing articulated programs, RFW-A organizes reusable components and their dependencies into category-level generators. Agents test and refine these generators across sampled configurations and motion states. Once constructed and validated, the generators produce diverse assets through procedural execution with no further model calls or token charges. The estimated API cost of generator authoring averages $10.62, amortizing to $0.05 per asset over 200 assets, versus $1.24 for the sampled per-asset agentic synthesis baseline. Using RFW-A, we built RFW-A-400K, containing over 400K sim-ready articulated assets across more than 500 categories. Experiments show high structural and kinematic validity and stable execution across physics simulators. Beyond asset validity, training with RFW-A-400K improves part-level articulation estimation on seen and unseen categories, while policies trained in simulation with retrieved and program-edited assets transfer zero-shot to real articulated objects.
30. 【2610.40327】Radiology Guideline-derived Decision Graph for Efficient Clinical Decision Support
链接：https://arxiv.org/abs/2610.40327
作者：Xuewen Zhou,Zhen Chen
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Radiological risk stratification, Radiological risk, essential downstream tasks, medical imaging, risk stratification
备注：
点击查看摘要
Abstract:Radiological risk stratification and management planning are essential downstream tasks in medical imaging. Automating these tasks is challenging because clinical guideline criteria are distributed across prose, tables, and cross-referenced definitions. We present Guideline-derived Decision Graph (GDGraph), a framework that decouples guideline compilation from case evaluation. During compilation, a large language model partitions guidelines into section-based chunks, extracts local decision subgraphs with linked definitions, and aggregates related branches into an executable decision graph. During inference, a fast System One model answers localized clinical questions while deterministic traversal resolves Boolean logic, turning opaque single-pass predictions into a traceable decision path without task-specific training. Across 347 cases spanning renal cyst, adnexal mass, and thyroid nodule guidelines, GDGraph improves System One accuracy from 57.3% to 90.8%, outperforming retrieval-augmented GPT-4o (80.1%) and graph-guided GPT-4o with chain-of-thought (82.7%) at substantially lower latency and cost. The source code will be available.
31. 【2610.40309】xtual Response-Aware Guidance for Zero-Shot Adversarial Robustness of Vision-Language Models
链接：https://arxiv.org/abs/2610.40309
作者：Yubo Cui,Xianchao Guan,Zheng Zhang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Vision-language models, exhibit strong zero-shot, exhibit strong, strong zero-shot generalization, Vision-language
备注：
点击查看摘要
Abstract:Vision-language models (VLMs) exhibit strong zero-shot generalization but remain vulnerable to adversarial attacks. Existing adversarial fine-tuning methods often use pre-trained model guidance to preserve image--text alignment while improving robustness. Our analysis of visual representation drift reveals complementary limitations of two representative forms of guidance. Matching visual features provides broad coverage by constraining all directions equally, but lacks semantic emphasis. Matching prediction distributions prioritizes semantically relevant directions, but remains limited by the candidate texts and prediction probabilities. These limitations can hinder preservation of pre-trained image--text relationships and degrade zero-shot performance. To address these issues, we propose Textual Response-Aware Guidance (TRAG), which constrains visual representation drift using the pre-trained text encoder's responses to local variations in input embeddings. Specifically, TRAG constructs anchor-level constraints from the response directions and magnitudes around semantic anchors, then aggregates them using pre-trained image--anchor relevance, yielding an image-adaptive constraint. Extensive experiments across architectures and tasks demonstrate state-of-the-art clean and robust performance. Further analysis shows that TRAG helps preserve pre-trained image--text relationships.
32. 【2610.40302】Self-training for cross-centre coronary vessel segmentation in X-ray angiography using pre-contrast frame subtraction
链接：https://arxiv.org/abs/2610.40302
作者：Danial Moafi,Aida Moafi,Mostafa Aryafar,Ashkan Sami,Maryam Aryafar
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Deep learning models, Deep learning, segmentation in X-ray, X-ray angiography, coronary vessel segmentation
备注： 7 pages, 2 figures, 3 tables
点击查看摘要
Abstract:Deep learning models for coronary vessel segmentation in X-ray angiography often perform poorly on data from centres that were not represented in training. Apart from differences in equipment, the labels of the largest public dataset (ARCADE) are incomplete, as only the segments defined in the SYNTAX score are annotated. In this study, we evaluated whether self-training with pseudo-labels can improve performance on external data without additional manual annotation. Pseudo-labels were generated from two sources, namely the predictions of an nnU-Net model trained on ARCADE and a contrast map obtained by subtracting a pre-contrast frame from the same sequence. The nnU-Net predictions were extended along the contrast map, and pixels detected only in the subtraction image were assigned an ignore label and excluded from the loss. Six models were trained and tested on ARCADE and on two external datasets (XCAD and XACV). All first-round models initialised with the baseline predictions performed better than both the baseline and a control model fine-tuned for the same number of epochs. The model trained on the XCAD pool improved the macro-averaged Dice score by 2.7 points (95% CI 2.3-3.2), while the control decreased it by 0.6 points. A second round of self-training did not lead to further improvement. When an external model trained on 50 fully annotated images was used to generate the initial labels, the XCAD Dice score increased from 0.623 to 0.745 without any XCAD labels, although performance on XACV did not improve because large main vessels were missed. These results suggest that the benefit of self-training depends largely on the sensitivity of the model used to generate the initial pseudo-labels.
33. 【2610.40284】LoRe-MIM: Lost-or-Replaced Masked Image Modeling for Self-Supervised Representation Learning in Whole-Body PET/CT
链接：https://arxiv.org/abs/2610.40284
作者：Amartya Bhattacharya,Bashirul Azam Biswas,Biratal Raj Wagle,Michael S. Leapman,Matthew E. Maeder,James B. Yu,Indrani Bhattacharya
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：positron emission tomography, hold significant clinical, Deep learning models, whole-body positron emission, computed tomography
备注： 25 pages, 9 figures
点击查看摘要
Abstract:Deep learning models for whole-body positron emission tomography/computed tomography (PET/CT) hold significant clinical potential for disease detection, treatment planning, and response assessment. However, the time and cost associated with expert annotations remain major bottlenecks in developing robust and generalizable models. Although self-supervised learning (SSL) has emerged as a promising approach to mitigating annotation scarcity, its application to whole-body PET/CT remains relatively underexplored due to the limited availability of large-scale datasets and the challenges associated with learning complementary cross-modal representations. To address these limitations, we curate a heterogeneous, multi-tracer, multi-institutional cohort comprising 6,531 whole-body PET/CT scans and propose LoRe-MIM (Lost-Or-Replaced Masked Image Modelling), a novel self-supervised framework designed to learn multimodal representations through masked image reconstruction. Specifically, LoRe-MIM introduces a masking strategy in which image regions are either replaced with zeros (Lost) or substituted with corresponding complementary imaging modality (Replaced), encouraging the model to capture both modality-specific characteristics and cross-modal relationships. We evaluate the learned representations across multiple downstream tasks, including multimodal PET/CT image reconstruction, patch-based classification into cancer vs. non-cancer, whole-body lesion segmentation, label-efficient training, and robustness assessment. Experimental results demonstrate that LoRe-MIM consistently outperforms the corresponding state-of-the-art baselines across these tasks, highlighting its potential as a scalable and generalizable self-supervised learning framework for whole-body PET/CT analysis.
34. 【2610.40276】askFrame: Connecting Task-Level Planning with Frame-Level Control for Game Agents
链接：https://arxiv.org/abs/2610.40276
作者：Guangzhi Wang,Junfu Pu,Yiran Wang,Xingyilang Yin,Honghao Chen,Cheng Cheng,Lingen Li,Yinan Zhou,Jiarui Wu,Yinuo Jiang,Hongbo Jin,Xiaodong Cun,Ying Shan
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：changing visual environments, Building general game, rapidly changing visual, Building general, visual environments
备注：
点击查看摘要
Abstract:Building general game agents requires both high-level task reasoning and low-latency action control in rapidly changing visual environments. These capabilities favor models with substantially different computational profiles, making them difficult to realize within a single model. In this work, we present TaskFrame, a hierarchical planner-controller framework that connects task-level planning with frame-level keyboard-and-mouse control. A vision-language task planner decomposes a high-level goal into an ordered sequence of atomic instructions and is optimized through supervised learning and dependency-aware GRPO. The hidden states of special goal tokens that terminate each atomic instruction provide continuous goal embeddings that are cached and reused across control steps. A causal Transformer controller combines the embedding of the active goal with recent observations to predict discrete keyboard and mouse-button actions and continuous camera motion at every game step. We train the system in three stages: planner training, controller pretraining, and joint alignment. Experiments in Minecraft show that TaskFrame is competitive with general-purpose vision-language models on instruction decomposition, and that in closed-loop evaluation the agent achieves a higher combat success rate than the evaluated baselines.
35. 【2610.40212】Benchmarking color judgment in multimodal LLMs with CIE-computable ground truth
链接：https://arxiv.org/abs/2610.40212
作者：Shuwei Yue,Kaida Xiao,Zhihua Wang,Qiao Chen
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Multimodal large language, Multimodal large, colorimetric ground truth, image quality, ground truth
备注：
点击查看摘要
Abstract:Multimodal large language models (MLLMs) are now used as zero-shot judges of image quality, but their color judgments are rarely tested with colorimetric ground truth. We present a five-task benchmark based on the CIE colorimetric framework: AWB pair judgment, skin-tone bias direction, CCT estimation, CIE Delta E 2000 pair ordering, and color naming under changing illuminants. We test eight MLLM endpoints, including four Qwen3-VL variants, GLM-4.5V, GPT-5.4, Claude-Opus-4.7, and Gemini-3.1-Pro. The models often solve coarse binary AWB judgments, but they remain weak on numeric CCT estimation and on some fine color-difference judgments. On ColorChecker-24 stimuli, the three closed-weight frontier models reach CCT MAE of 715-1056 K, compared with 1358-2133 K for the five open-weight models. A non-ColorChecker control shows that this advantage is partly template-dependent. The results suggest that current MLLMs have useful color words and coarse visual priors, but their color judgments are not reliably calibrated to CIE-computable color differences.
36. 【2610.40199】Decision-Level Error Attribution and Depth-Aware Safety Evaluation of Monocular RGB Human Tracking for Autonomous Robot Navigation
链接：https://arxiv.org/abs/2610.40199
作者：Abdulahi Taiwo,Chijioke Cyriacus Ekechi
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
关键词：Monocular RGB tracking, errors affect binary, affect binary stop, Monocular RGB, geometric accuracy metrics
备注： 10 pages, 7 figures, 6 tables
点击查看摘要
Abstract:Monocular RGB tracking is useful for robots operating near people, but perception errors affect binary stop/proceed decisions in ways that geometric accuracy metrics do not capture. We examine how much decision error arises from the safety rule versus perception. We prove that lateral-only distance is a lower bound on planar distance: with accurate positions, it cannot miss dangers identified by the planar rule, but it produces unnecessary stops. We introduce a two-factor Shapley attribution separating rule and perception contributions, and further distinguish missed detections from localization errors. Experiments use ETH and UCY pedestrian videos with simulated stationary robots, disjoint calibration and evaluation pedestrians, and time-block bootstrap intervals. Across 898 pedestrians in three scenes, the lateral rule missed no dangers on annotated positions but achieved only 5.6-31.9% stop precision. In two scenes with RGB perception, calibrated ground-plane localization achieved 0.59 m and 0.51 m planar RMSE, compared with 1.72 m and 1.39 m for a geometric baseline and 0.81 m and 0.68 m for calibrated Depth Anything V2. Detection recall of 0.43 and 0.26 made missed detections the leading source of missed danger. A two-standard-deviation safety margin with ground-plane localization matched or improved on the lateral rule's missed-danger rate in all eight scene-threshold settings and reduced unnecessary stops in seven. All robot interactions are simulated, and outcomes are offline decision classifications, not demonstrations of collision avoidance.
37. 【2610.40191】Video Generation Meets Agent Harness: Agentic Video Generation, Agentic World Modeling and Recursive Self-Improvement
链接：https://arxiv.org/abs/2610.40191
作者：Liaoyuan Fan,Yuhao Zheng,Dongxing Mao,Hangyu Ran,Siyuan Hu,Weijia Mao,Rui Zhao,Weijia Wu,Guian Fang,Yanzhe Chen,Zeyu Zhu,Yuxin Jiang,Yiren Song,Yuchao Gu,Junchao Yi,Qi Hu,Dan Kong,Alex Jinpeng Wang,Kevin Qinghong Lin,Mike Zheng Shou,Wei Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：feedback enable coherent, controllable multi-shot videos, enable coherent, controllable multi-shot, Video
备注： website: [this https URL](https://agentic-videogen.github.io/) github: [this https URL](https://github.com/agentic-videogen/awesome-agentic-videogen)
点击查看摘要
Abstract:Video generation is shifting from end-to-end synthesis toward agentic generation, where planning, tool use, and feedback enable coherent, controllable multi-shot videos. In parallel, video world models provide action-conditioned observations and predicted futures, serving as interactive environments for agents. We study their intersection through the agent harness: an executable runtime that constructs context, exposes tools, maintains state, routes feedback, and manages budgets around video models. This report examines three directions. First, agentic video generation, where harnesses coordinate scripting, shot planning, asset management, production, and editing, alongside the evaluation challenges of these systems. Second, agentic world modeling, where harnesses mediate observation, control, execution, and state. We discuss hybrid designs in which game engines or agent-written programs maintain world state while video models render observations, and how agents and world models can mutually evaluate and improve one another. Third, recursive self-improvement (RSI) in video model development, distinguishing output revisions, updates to harness state, generator parameters, and evaluators, and changes to the improvement procedure itself. We identify how these improvements feed back into subsequent development. Across these directions, we analyze essential agent capabilities, including audiovisual perception, long-horizon planning, memory, code-based interaction, feedback-guided control, skill reuse, and multi-agent coordination. We conclude with challenges in autonomous video production, lifelong memory, programmable hybrid control, autonomous model development, provenance, cost accounting, and evaluation linking video quality and task performance to sustained improvements in models and harnesses.
38. 【2610.40190】AdMuse: Skill-Guided Creative Advertising Image Generation
链接：https://arxiv.org/abs/2610.40190
作者：Minyan Luo,Yuxin Zhang,Meng Zhao,Xincan Wang,Fuzhang Wu,Tong-Yee Lee,Oliver Deussen,Fan Tang,Weiming Dong
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：attract attention, outstanding creative advertising, outstanding creative, advertising, product
备注：
点击查看摘要
Abstract:An outstanding creative advertising image should both attract attention and communicate a product's positioning. How to choose a persuasive conceptual bridge before rendering begins is challenging for advertising generation. We present AdMuse, a skill-guided multi-agent framework that distills high-quality advertisements into reusable creative skills for cross-domain advertising strategy. Rather than retrieving isolated exemplars, AdMuse abstracts recurring bridge patterns into a skill library that can guide new product briefs. At generation time, AdMuse coordinates skill routing, multi-candidate ideation and selection, and VLM-guided visual refinement: diverse skills propose alternative concepts, relative judgment selects the most promising direction, and vision-language feedback helps realize the selected intent in the final image. This design treats advertising creativity as a structured decision process rather than a single prompt-writing step. Experiments across diverse product and brand positionings show that AdMuse produces visually compelling advertisements that faithfully communicate product positioning.
39. 【2610.40175】Beyond CCT: Multi-channel SPD recovery for per-shot camera color correction
链接：https://arxiv.org/abs/2610.40175
作者：Shuwei Yue,Hongke Wei,Qiao Chen
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Correlated Color Temperature, Modern cameras correct, Color Temperature, Correlated Color, cameras correct color
备注：
点击查看摘要
Abstract:Modern cameras correct color via two or three pre-calibrated 3x3 matrices interpolated by Correlated Color Temperature (CCT). This collapses the illuminant SPD to a scalar. LEDs, fluorescents, and mixed sources sit off the Planckian locus, and metameric illuminants with the same chromaticity demand different optimal CCMs that a CCT pipeline cannot distinguish. We recover the illuminant SPD instead using non-negative least squares against a basis dictionary, conditioned on the camera white point and a multi-channel ambient-light sensor (ALS). A closed-form 3x3 CCM is then fit per shot. On a synthetic benchmark of 300 illuminants and 3,100 reflectances, mean angular error drops from 1.89 to 1.20 degrees, with a 1.7x tail reduction on the LED-like off-Planckian group (n=45) where metameric ambiguity is tangible. The gain holds under realistic ALS imperfections and observation noise, and the same off-Planckian advantage carries over to 1,684 real measured SPDs.
40. 【2610.40166】EventSeeColor: Learning to Reconstruct Color from Monochrome Events
链接：https://arxiv.org/abs/2610.40166
作者：Chuanzhi Xu,Langyi Chen,Haoxian Zhou,Zelin Zhang,Haodong Chen,Qiang Qu,Xiaoming Chen,Weidong Cai
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：challenging lighting conditions, cameras capture visual, capture visual brightness, lighting conditions, capture visual
备注：
点击查看摘要
Abstract:Event cameras capture visual brightness changes under challenging lighting conditions. However, traditional event-based image reconstruction targets grayscale intensity and produces images without scene color. Color distinguishes surfaces with similar intensity. Reconstructing color from events can reveal appearance differences that reconstructed grayscale images cannot express. Although monochrome event cameras do not directly measure color, their event-triggering patterns depend on surface reflectance, illumination, and the sensor's spectral response. These patterns may contain cues that support color reconstruction. In this paper, we present EventSeeColor, which uses event-reconstructed grayscale and aligned events to jointly guide a diffusion colorizer. Grayscale-derived semantic context supports color inference, while region-balanced chroma supervision and local RGB refinement improve color distribution, brightness, and texture. In the combined evaluation on two real event datasets, EventSeeColor achieves the best structural similarity, learned perceptual scores, and chromatic-region color difference among the external methods.
41. 【2610.40153】Hierarchical Step Decomposition for Training-free Online Video Step Grounding
链接：https://arxiv.org/abs/2610.40153
作者：Long Zhou,Luca Zanella,Paolo Rota,Massimiliano Mancini,Elisa Ricci
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Large Multimodal Models, aims to identify, Hierarchical Step Decomposition, Video Step Grounding, HSD
备注： WACV 2027
点击查看摘要
Abstract:Given a set of procedural steps, Video Step Grounding (VSG) aims to identify which steps occur in a video and localize when they are performed. Existing methods are either offline, requiring access to the full video before making predictions, or online, but relying on fixed-length video segments producing noisy, over-segmented outputs. In this work, we propose Hierarchical Step Decomposition (HSD), a training-free framework for online VSG that leverages the semantic understanding capabilities of Large Multimodal Models. HSD addresses two central challenges in online grounding: constructing semantically coherent temporal segments from a streaming video, and bridging the semantic gap between coarse procedural step descriptions and visual observations. To construct elementary temporal units, HSD incrementally groups incoming segments into semantically consistent events, using an anchor caption that is continuously validated against the visual stream. Based on these events, we perform zero-shot procedural step classification and selectively refine uncertain predictions by decomposing candidate steps into finegrained substeps that capture more nuanced visual evidence. By operating on semantically coherent events rather than every fixed-length segment, HSD reduces redundant computation and produces more stable predictions. Experiments on HT-Step, Ego4D Goal-Step, and CrossTask show that HSD substantially outperforms state-of-the-art methods, overall achieving better step recognition, more accurate temporal localization, and faster inference.
42. 【2610.40137】NeurEO: dissecting Earth observation embeddings with computational neuroscience
链接：https://arxiv.org/abs/2610.40137
作者：Thijs L van der Plas,Jacob JW Bakermans,Ioannis N Athanasiadis
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：rich Earth observation, complex geospatial tasks, Earth observation data, Earth embeddings, enable downstream models
备注： Originally published at the EurIPS 2025 REO workshop
点击查看摘要
Abstract:Earth embeddings are compressed representations of rich Earth observation data that enable downstream models to tackle a wide range of complex geospatial tasks. Yet, this increase in performance comes at the cost of decreased interpretability, as the semantic meaning of Earth embeddings is unknown. To address this challenge, we borrow methods from the field of computational neuroscience, which aim to interpret activity in (biological) neurons. Here, we adapt and apply these methods to assign meaning to AlphaEarth embeddings. We analyse embedding features both as stand-alone data, equivalent to spontaneous neural activity, and in relation to land cover and elevation correlates, equivalent to stimulus-response neural activity. We demonstrate correlational and multi-scale structure in the embedding features. Embedding features can display highly specific and directional tuning to land cover and elevation, such as a `North vs South woodland edge' response. When we remove the main trends of combined land cover responses (like water versus land tuning), we discover fine-grained representational details with increased effective dimensionality. Together, our analyses start to unravel what these Earth embeddings represent, which is critical for developing trustworthy downstream applications.
43. 【2610.40131】VideoMemBench: How Well Do MLLMs Remember in Video Conversations?
链接：https://arxiv.org/abs/2610.40131
作者：Zhengdong Hu,Chao Wang,Kaituo Feng,Huanjin Yao,Mingfei Lu,Hehe Fan,Yi Yang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Multimodal large language, large language models, Multimodal large, making extended multi-turn, process hour-long videos
备注：
点击查看摘要
Abstract:Multimodal large language models (MLLMs) can now process hour-long videos, making extended multi-turn dialogue over a single video a natural mode of use. Sustaining such a dialogue requires more than answering each question in isolation: the model must retrieve and reuse information established earlier in the conversation, both what it has seen and what it has said, an ability we term in-context video memory. Existing video QA benchmarks leave this ability unmeasured: their questions are mutually independent, never reference one another, and each is answerable from the video alone, so nothing ever requires carrying information across turns. We introduce VideoMemBench, in which each sample pairs a single video with a long multi-turn dialogue, whose turns are connected by typed cross-turn dependencies: an Episodic Dependency refers back to a visual episode described in an earlier turn, probing memory of what the model has seen, while an Answer Dependency builds on the model's own earlier answer, probing memory of what the model has said. Every dependency is verified considering two properties: 1) necessity, meaning the question cannot be answered without its earlier turns, and 2) non-leakage, meaning the question does not restate the information from its earlier turns. To enforce these properties, we design a construction pipeline with three LLM roles: a planner, a generator, and a judger, with each dialogue reviewed by human experts. VideoMemBench comprises 326 videos with 2,476 turns, of which 1,387 depend on earlier turns, including multi-parent dependencies and long-range call-back. Among over 20 evaluated open- and closed-source MLLMs, Claude Opus 5 performs the best with an overall score of only 45.7\%, highlighting the challenge of in-context video memory. Our dataset and evaluation are released.
44. 【2610.40124】It Takes Two to Probe, Thousands to Retrieve: Putting Vision--Language Probes to the Retrieval Test
链接：https://arxiv.org/abs/2610.40124
作者：Ivana Beňová,Michal Gregor,Marián Šimko
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：text matching probes, text matching, tested for fine-grained, fine-grained understanding, correct item
备注： 12 pages, 2 figures, 3 tables
点击查看摘要
Abstract:Vision--language models (VLMs) are tested for fine-grained understanding with image--text matching probes, a model must prefer the correct image or caption over handcrafted negative. In retrieval, the correct item must beat thousands of alternatives. We measure how these two evaluation methods relate by using two probing benchmarks, \svo{} and SugarCrepe, for retrieval tasks. For 100 image and caption queries we judge the top-10 retrievals from full dataset of six VLMs, with human annotators and an LLM assessor. We than compare how well models rank pool of items on the same queries, while changing only where the negatives comes from: the benchmark, a random draw, or the judged-incorrect retrievals of other models. On \svo{} the accuracy of retrieving a positive over handcrafted negatives is significantly higher than for retrieved ones (the correct item is preferred in 85\% against 77\% of the time). Using pool of ten retrieved negatives leads to same accuracy as when one uses pool of 200 to 430 random negatives. On SugarCrepe the order is reversed: its handcrafted negatives are harder than the retrieved ones. Incomplete relevance labels explain most of the low retrieval scores on \svo{}, where the top-ranked item is correct for 54--90\% of the queries under human labels against 6--21\% under the benchmark's labels. However, we detect no association between passing the probe for a query and retrieving a correct item first. The LLM assessor o3 agrees with individual annotators ($\kappa$ 0.57--0.61) slightly less than they agree with each other (0.64--0.68).
45. 【2610.40121】Learning a Motion Manifold to Constrain Ambiguity in Unsupervised 4D Medical Image Interpolation
链接：https://arxiv.org/abs/2610.40121
作者：Haojin Li,Huazhu Fu,Heng Li,Jiang Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：medical image interpolation, image interpolation reconstructs, interpolation reconstructs intermediate, temporally sparse observations, medical image
备注：
点击查看摘要
Abstract:Unsupervised 4D medical image interpolation reconstructs intermediate anatomical states from temporally sparse observations without intermediate supervision. Distinct motion fields may attain comparable endpoint agreement while producing different intermediate anatomy, giving rise to motion ambiguity. This ambiguity is difficult to constrain when many independently adjustable motion variables must be estimated from the observed endpoints. We develop a framework that learns a motion manifold to restrict candidate motion fields while retaining sample-specific variation. Cross-Sample Spectral Basis Learning extracts recurring motion structure from endpoint-derived dense estimates and encodes it in a shared spectral basis. Sample-Conditioned Cayley Basis Reorientation uses predicted motion coordinates to rotate the effective motion directions for each endpoint pair without increasing the number of independent coordinates. Their composition defines a nonlinear family of complete velocity fields across endpoint pairs, accommodating anatomical and motion-state differences within a restricted representation. Integrating each predicted field supports bidirectional reconstruction at arbitrary intermediate times. Theoretical analysis establishes sufficient conditions under which endpoint observations distinguish nearby motions within the learned family. Experiments on ACDC and 4D-Lung demonstrate state-of-the-art performance among the evaluated methods trained without intermediate volumes and competitive results against methods using intermediate supervision.
46. 【2610.40119】SynthFaces: Balanced Large Scale Human Dataset
链接：https://arxiv.org/abs/2610.40119
作者：Sergey Karpukhin,Denis Bobkov,Kirill Morozov,Aibek Alanov
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：offer imbalanced representations, sensitive personal information, critically hindered, inherent biases, biases and privacy
备注： accepted at BMVC 2026 github: [this https URL](https://github.com/ControlGenAI/SynthFaces)
点击查看摘要
Abstract:The pursuit of fair and equitable AI in human-centric tasks is critically hindered by the inherent biases and privacy risks of real-world datasets, which offer imbalanced representations of identity and often contain sensitive personal information. Consequently, state-of-the-art generative models trained on this skewed data exhibit unfair performance, particularly in identity-preserving generation. In this work we propose a pipeline for controllable generation of dataset with diverse faces for generative models training and finetuning. We achieve this by expanding foundational Arc2Face diffusion model with a flow matching generative model in the manifold of identity embeddings. We demonstrate that such approach achieves both goals of generative fidelity and label diversity not present in real human-centric datasets. As a result we collect and further contextualize a highly balanced identity-centric dataset of humans named SynthFaces. By design, our released images and identities are synthetic, reducing direct reliance on real personal images; we empirically check identity separation with a nearest-neighbor non-leakage metric. We purpose it for training and tuning of fair models for various downstream tasks, and showcase its usability in finetuning an existing adapter-based method for personalized generation.
47. 【2610.40115】SelfThinker: Self-Evolving Reasoning via Reliability-Verified Intrinsic Visual Thoughts
链接：https://arxiv.org/abs/2610.40115
作者：Zhangquan Chen,Ruqi Huang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Unified multimodal models, set of weights, offering an opportunity, generate and understand, single set
备注： 25 pages, 12 figures
点击查看摘要
Abstract:Unified multimodal models can both generate and understand images within a single set of weights, offering an opportunity to use image synthesis as training supervision for visual comprehension. However, existing visual chain-of-thought methods rely on human-annotated helper images during training and require slow image generation at test time. We present SelfThinker, the first self-evolving framework that turns a unified model's self-generated visual thoughts into stronger understanding without human image annotations. Given a visual problem, SelfThinker samples candidate helper images from its generation pathway and uses its frozen understanding pathway to score each candidate via conditional pointwise mutual information. High-utility visual thoughts then refine the generator via flow matching and serve as privileged teacher context for on-policy self-distillation. At inference, SelfThinker reasons directly from the original image in text without generating intermediate images. Although trained only on small synthetic visual-search and navigation datasets, SelfThinker transfers broadly to natural-image and spatial benchmarks, raising the seven-benchmark average of ThinkMorph-7B from 52.7 to 64.0 (+11.3 points). Controlled image-perturbation probes further show that SelfThinker surpasses ThinkMorph-7B by up to 40.4 points and its pretrained BAGEL-7B backbone by up to 23.2 points across fine-grained and spatial perception tasks. Our code is available at this https URL.
48. 【2610.40094】RepConTUR: A Report-Conditioned Volumetric Multimodal Model for Abdominal Tumor Segmentation
链接：https://arxiv.org/abs/2610.40094
作者：Seunghwan Lee,Sangjoon Park,Jongmin Lee,Yoon Jo Kim,Wonyoung Cho,Jin Sung Kim,Han Joo Chae
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：predominantly false positives, increase editing time, radiotherapy planning, limited by errors, predominantly false
备注： Submitted to CVPR 2027
点击查看摘要
Abstract:Tumor delineation is required for radiotherapy planning, yet automatic segmentation is limited by errors, predominantly false positives, that distort dose and increase editing time. The radiology report accompanies the scan and localizes disease at the suborgan level, but existing text-promptable models condition on short referring phrases, and most operate in 2D. We present RepConTUR, a single native-3D early-fusion multimodal model that conditions volumetric abdominal tumor segmentation on the narrative report. We extend AbdomenAtlas 3.0 to a suborgan-level promptable segmentation dataset and generate report-based VQA pairs, yielding four report-derived tasks on the same annotations: referring segmentation (RES), report-grounded segmentation (GSEG), report generation (RRG), and visual question answering (VQA). We train RepConTUR, comprising a 3D CT encoder, an early-fusion language model, and a native-3D segmentation module, jointly on the four tasks. Across liver, kidney and pancreas, RepConTUR attains the highest Dice similarity coefficient (DSC), instance-wise precision, and detection F1 among the text-promptable baselines, and adding the report raises mean precision while leaving recall comparable. Against two vision-only specialists, nnU-Net and the report-supervised R-Super, RepConTUR attains the highest value on every in-domain metric for liver and pancreas and the highest mean DSC, recall, precision, and detection F1 over the four out-of-distribution public cohorts. Ablations indicate that conditioning on the narrative report raises detection F1 from 0.499 to 0.528 within the same model, that joint GSEG training raises RES F1 from 0.450 to 0.482, and that the native-3D segmentation module raises GSEG F1 from 0.139 to 0.528 over its 2D counterpart. Code is available at this https URL.
49. 【2610.40088】Neural Blind Deconvolution for Image Scanning Microscopy
链接：https://arxiv.org/abs/2610.40088
作者：Eliot Guez,Nathan Buskulic,Lisa Cuneo,Luca Calatroni,Giuseppe Vicidomini
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Image scanning microscopy, Deep Image Prior, scanning microscopy, obtain a super-resolved, blind Deep Image
备注： 4 pages, 2 figures
点击查看摘要
Abstract:Image scanning microscopy (ISM) acquires one scanned image per detector element, which are then fused into a single image by multi-image deconvolution to obtain a super-resolved image. Its accuracy, however, depends on point spread function (PSF) models that sample- and system-induced aberrations often invalidate. Since differences between the scanned images encode aberration information, we propose a training-free blind Deep Image Prior (DIP) framework that jointly estimates the specimen and the detector-dependent PSFs. The PSFs are parametrised through the detector geometry and shared excitation and emission PSFs, reducing the unknowns while preserving the physical ISM model, and a Gaussian pre-initialisation stabilises the blind optimisation. On experimental tubulin data, blind DIP recovers finer filamentous structures than multi-image deconvolution and DIP with fixed theoretical PSFs, while adapting the PSFs to the measurements.
50. 【2610.40085】Morphology-Aware Channel-Decoupled Mamba Network for Medical Image Segmentation
链接：https://arxiv.org/abs/2610.40085
作者：Chenchen Wang,You Zhou
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：treatment planning, Accurate multi-organ segmentation, Accurate multi-organ, computed tomography, computer-aided diagnosis
备注：
点击查看摘要
Abstract:Accurate multi-organ segmentation in computed tomography (CT) is essential for computer-aided diagnosis, treatment planning, and disease monitoring. However, substantial variations in organ shape and size, together with low tissue contrast and ambiguous boundaries, complicate the modeling of heterogeneous anatomical structures. To address these challenges, we propose a Morphology-Aware Channel-Decoupled Mamba Network (MCDNet), which integrates adaptive local structural enhancement with channel-decoupled long-range modeling. Specifically, we introduce a Morphology-Aware Channel-Decoupled State-Space (MCSS) block that divides feature channels into separate processing groups, allowing differentiated structural representations to be learned within an efficient state-space framework. Within each group, a Morphology-Aware Enhancement Module (MAEM) combines deformable convolutions with channel attention to adapt local feature extraction to anatomical shape variations. Group-wise state-space modeling then captures long-range spatial dependencies, and channel gating adaptively integrates the resulting features. Experiments on the Synapse abdominal CT dataset demonstrate the effectiveness of MCDNet, achieving an average Dice Similarity Coefficient of 81.61% for multi-organ segmentation.
51. 【2610.40073】Uncertainty-aware Coarse-to-Fine Main Pancreatic Duct Segmentation for Width Measurement in 3D MRCP
链接：https://arxiv.org/abs/2610.40073
作者：Zipei Zhang,Jaakko Sahlsten,Mikko Saukkoriipi,Johanna Ronkainen,Mona Lehto,Akseli Aittoniemi,Irina Rinta-Kiikka,Kimmo Kaski
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：magnetic resonance cholangiopancreatography, pancreatic cancer analysis, volumetric magnetic resonance, ductal dilatation diagnosis, main pancreatic duct
备注：
点击查看摘要
Abstract:Assessment of the main pancreatic duct (MPD) from volumetric magnetic resonance cholangiopancreatography (MRCP) scans is an essential task for ductal dilatation diagnosis and pancreatic cancer analysis. However, manual assessment is affected by reader variability in which automated analysis may assist. Nevertheless, fully automated and uncertainty-aware MPD analysis from volumetric MRCP remains underexplored area of research. In this study, we propose an automatic and uncertainty-aware deep learning-based pipeline for MPD segmentation and maximum-width analysis from MRCP scans. A coarse localisation model first identifies the ductal region-of-interest, followed by a fine model that segments the MPD within the region. The dataset consists of 256 MPD-segmented MRCP scans from 69 patients. A randomly selected holdout test subset of 14 patients with 55 MRCP volumes was used for evaluation. We employed a five-model cross-validation ensemble for improved performance and uncertainty estimation. The best MPD segmentation performance was 0.634 DSC, 0.744 precision, and 0.593 recall. The mean absolute error for maximum MPD width prediction was 0.355~mm and dilated MPD ($\geq 5$~mm) was detected with 75.0\% sensitivity and 74.4\% specificity. These novel results suggest that automated MPD segmentation can be used to aid MRCP-based duct measurements.
52. 【2610.40068】BRACHIO: Broad Representations Across Centres for Head-and-neck Imaging Oncology (HECKTOR 2026)
链接：https://arxiv.org/abs/2610.40068
作者：Roxanne De Maeyer,Ine Dirks,Yannick Brackenier,Simon DeKeyser,Jef Vandemeulebroucke,Vicky De Ridder
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：nodal tumour volumes, predict recurrence-free survival, benchmarks machine learning, teams delineate primary, challenge benchmarks machine
备注： HECKTOR Challenge 2026, MICCAI, head and neck cancer, PET imaging
点击查看摘要
Abstract:The HECKTOR 2026 challenge benchmarks machine learning for head and neck cancer, where teams delineate primary and nodal tumour volumes, perform TN staging, and predict recurrence-free survival (RFS). This report describes the submissions for the HECKTOR challenge of team nucli. As the three tasks span both dense and global information from multicentric PET/CT scans, we utilise self-supervised pre-trained encoders, training a 2D DINOv2 encoder (PET-DINOv2-MIPs) on HECKTOR 2026's PET scans. Our final test submission ensembles a DynUNet trained from scratch with a SwinUNETR whose Swin trunk was initialised from a SimMIM-pre-trained checkpoint for segmentation, reaching a mean Dice of 0.5409. For TN staging, GTVp/GTVn geometric features from the segmentation masks are fit to stage probabilities by logistic regression (LR) and fused with PET-DINOv2-MIPs deep features (PCA + LR) via a second-level LR, reaching balanced accuracies of 0.3962 (T-stage) and 0.6432 (N-stage). For the RFS prognosis task, we fine-tune the last two blocks of the PET-DINOv2-MIPs ViT-S encoder jointly with an MLP head under a CoxPH loss, concatenating clinical covariates with the imaging features, reaching a test Harrell's C-index of 0.6577 and a weighted challenge score of 0.5802. The prognosis result is comparable to the best entry of the previous edition and consistent with our internal cross-validation, while our segmentation result is well below it and limits the staging results that depend on it.
53. 【2610.40061】Localizing Concepts in Visual Autoregressive Models
链接：https://arxiv.org/abs/2610.40061
作者：Nanxiang Jiang,Yang Liu,Yuanhao Wang,Min Xu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Understanding the internal, safe and controllable, autoregressive models, Understanding, visual autoregressive models
备注：
点击查看摘要
Abstract:Understanding the internal routing of visual concepts is crucial for the safe and controllable adaptation of generative models. While concept localization has been widely studied in diffusion models, the emerging paradigm of next-scale visual autoregressive models remains largely unexplored. In this paper, we introduce LoCo, the first model-agnostic framework to localize where and when specific concepts emerge within autoregressive models. Driven by the native coarse-to-fine nature of next-scale generation, our method precisely maps conceptual knowledge across three distinct dimensions: Layer, Scale, and Position. To systematically localize and evaluate concept routing without context bias, we propose LoCoBench, a comprehensive dataset spanning 10 diverse categories. Extensive probing on image and video autoregressive models like Infinity, HunyuanImage-3.0, and InfinityStar shows that the localized positions are both interpretable and causally related to concept emergence. Building on these insights, we apply our localization method to three key applications: concept erasure, model personalization, and adversarial concept injection. Experiments demonstrate that our targeted intervention achieves state-of-the-art performance, substantially reducing computational overhead while preserving benign utility. Overall, our findings offer insights into how conceptual knowledge is routed during autoregressive generation, introducing a practical pathway for more interpretable, efficient, and secure adaptation.
54. 【2610.40046】EPC-RawDC: Equivariant Physics Constrained 3D Diffusion with Raw Data Consistency for CBCT to CT Image Synthesis
链接：https://arxiv.org/abs/2610.40046
作者：Alzahra Altalib,Chunhui Li,Alessandro Perelli
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Medical Physics (physics.med-ph)
关键词：Cone beam computed, inaccurate Hounsfield units, Hounsfield units limit, beam computed tomography, Cone beam
备注： 10 pages, 4 figures, 8 tables
点击查看摘要
Abstract:Cone beam computed tomography (CBCT) is widely used for image guided and adaptive radiotherapy, but scatter, noise, reconstruction artifacts, and inaccurate Hounsfield units limit its quantitative use. Recent diffusion approaches for CBCT to CT synthesis remain largely image driven, with limited use of the underlying acquisition physics. We propose EPC-RawDC, a conditional 3D latent diffusion framework that combines volumetric generative modeling with two complementary constraints derived from the same scanner specific CBCT forward model. First, geometry equivariant projection constraint exploits the rotational equivariance of the CBCT forward operator by supervising the synthesized CT through correspondingly transformed reference projections. Second, raw data consistency uses noisy projection measurements of the individual acquisition as scan specific supervision during training. The framework was evaluated on two independent clinical cohorts with distinct scanner geometries and projection sources. For JUST, the physics constraints used measured clinical raw CBCT projections acquired with an Elekta system, whereas for SynthRAD2025 the projection data were simulated from the reference CT using a Varian geometry with Poisson photon statistics. EPC-RawDC achieved an SSIM of 0.97 in both cohorts and improved PSNR over DDPM by 2.66 dB on SynthRAD2025 and 2.10 dB on JUST, while reducing mean MAE by 20.3% and 19.0%, respectively. On SynthRAD2025, projection consistency and equivariant errors were reduced by 55.0% and 41.4% relative to DDPM. These results demonstrate that combining a volumetric diffusion prior with scanner geometry and scan specific projection measurements improves quantitative CBCT to CT synthesis across both simulated and measured projection domains. Code available after publication at this https URL.
55. 【2610.40029】Segment Anything Model Based Context Guided Mamba Unet for Medical Image Segmentation
链接：https://arxiv.org/abs/2610.40029
作者：Chenchen Wang,You Zhou
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：fine-grained structural precision, global contextual understanding, State Space Models, treatment planning, crucial for clinical
备注：
点击查看摘要
Abstract:Medical image segmentation is crucial for clinical diagnosis and treatment planning, requiring both global contextual understanding and fine-grained structural precision. While the Segment Anything Model (SAM) demonstrates strong generalization capabilities, its direct application to medical images often lacks domain-specific adaptability. Meanwhile, State Space Models (SSMs)-based methods, such as Mamba, efficiently capture long-range dependencies but may struggle with preserving fine anatomical details. To address these challenges, we propose a segment anything model based context guided mamba unet, SAMBA-UNet, a novel framework that integrates SAM's context knowledge with Mamba's hierarchical feature learning to enhance both global representation and local detail refinement. Specifically, we introduce the Hierarchical Context Enhancement Module (HCEM), an attention-based fusion module that deeply incorporates the multi-level features of SAM and Mamba to improve detail perception and segmentation precision. Additionally, we employ LoRA and Adapter-based fine-tuning to adapt SAM to medical imaging tasks, further enhancing its domain-specific segmentation capability and robustness. We conducted a comprehensive study on several medical segmentation datasets, including Synapse, ISIC17, and ISIC18, evaluating our method across multiple metrics. Experimental results demonstrate that SAMBA-UNet achieves state-of-the-art performance, effectively balancing generalization, segmentation accuracy, and computational efficiency. Our code will be released upon acceptance.
56. 【2610.40018】DINOv2 for PET: A Pre-Training Domain Benchmark Across Classification, Segmentation, and Prognosis
链接：https://arxiv.org/abs/2610.40018
作者：Roxanne De Maeyer,Yannick Brackenier,Vicky De Ridder,Simon DeKeyser,Maarten Larmuseau,Jef Vandemeulebroucke
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：dard for self-supervised, image DINO features, DINO features transfer, DINO, producing features
备注： 8 pages + references, MICCAI 2026, CaPTion workshop
点击查看摘要
Abstract:DINO (self-DIstillation with NO labels) has become a stan- dard for self-supervised pre-training, producing features that generalise across datasets and downstream tasks. In medical imaging it has been adopted mainly for radiology (X-ray, CT and MRI), where even natural- image DINO features transfer well. In the literature, these features have been reported to transfer worse to functional imaging such as PET. Yet PET-specific DINO models have barely been introduced, and even less validated across multiple tasks and datasets. We present PET-DINOv2-2D, the first DINOv2 model pre-trained on 2D PET, and benchmark it against open-source encoders: the origi- nal DINOv2 (pre-trained on the LVD-142M natural-image dataset) and OmniRad (pre-trained on CT, MRI and ultrasound). Using the same lightweight heads for every backbone, and despite far less pre-training data, PET-DINOv2-2D matches or outperforms both on three down- stream PET tasks: normal/abnormal classification, lesion segmentation and recurrence-free survival prognosis. Its advantage is largest when few labels are available, suggesting that PET pre-training gives more transferable, label-efficient features. Code is made available at this https URL.
57. 【2610.40007】Greenscapes: Optimized Vegetation Configurations in Cities
链接：https://arxiv.org/abs/2610.40007
作者：Patryk Bartkowiak,Marcin Marlewski,Paweł Wójcik,Jae Joong Lee,Dominik L. Michels,Bedrich Benes,Sören Pirk,Wojciech Pałubicki
类目：Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV)
关键词：serving functional roles, providing acoustic screening, shaping the microclimate, carbon sequestration, central role
备注： 18 pages (11 main paper + 7 supplementary material), 15 figures, 5 tables. Project page: [this https URL](https://greenscapes-project.github.io/) Code: [this https URL](https://github.com/greenscapes-project/greenscapes)
点击查看摘要
Abstract:Vegetation plays a central role in cities, serving functional roles such as balancing shade and sky exposure, shaping the microclimate, carbon sequestration, and providing acoustic screening. Furthermore, aesthetic choices in landscape styles (formal, modernist, picturesque) are important for citizens' well-being and should be considered in the design of urban landscapes. In this paper, we present a novel method for placing vegetation, ranging from shrubs to large trees, in 3D urban scenes. Our method uses individual lots as input and employs a conditioned Vision-Language Model (VLM) to generate a top-down rendering of vegetation, placing large trees, small trees, and shrubs on lawn areas. This rendering is then converted into a 3D scene by detecting individual plants and selecting appropriate 3D assets. To refine the proposed configurations, we define an optimization loop that keeps the VLM-generated positions fixed after a deterministic feasibility repair and uses CMA-ES to optimize per-instance scale transformations along with discrete vegetation asset substitutions. To define the objective function for the optimization, we evaluate the generated 3D scenes with efficient surrogate models of shade and path-based acoustic exposure. Furthermore, we render in-scene views at each optimization step to compute view-dependent perceptual descriptors, enabling the generation of both aesthetic and functional vegetation layouts for urban scenes.
58. 【2610.40002】How Much Should a VLA Trust Its Reasoning? Step-Level Guidance as a Measured Action-Value Function
链接：https://arxiv.org/abs/2610.40002
作者：Dabin Seo,Eunhyek Joa
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：policies write, wrong trace yields, Reasoning, reasoning trace, VLA
备注： 6 pages, 2 figures, 4 tables
点击查看摘要
Abstract:Reasoning vision-language-action (VLA) policies write a reasoning trace and then decode the action from it. A wrong trace yields a wrong action, and even a correct-looking trace can mislead the plan. Existing remedies ask whether the reasoning is right. We ask how strongly it should steer. CoC-CFG applies classifier-free guidance between the policy's context and the same context without the reasoning span. One weight per Euler step attenuates or amplifies the reasoning, and w=1 recovers the original policy. We treat the choice of this weight as a Markov decision process whose action values are measured, not bootstrapped. With the expert frozen and the noise fixed, applying a weight for one step and finishing the plan under the current controller returns that weight's value exactly. Q-adaptive guidance fits these values with small critics and improves the controller conservatively. In the Alpasim driving simulator it outperforms every fixed weight and online RL (SAC, GRPO) in open loop, and in closed loop it passes 262 of 300 scenes, against 252 for the original policy. On held-out frames of the Waymo end-to-end driving benchmark, its first three stages outperform every fixed weight, four adaptive-guidance methods, SAC and GRPO. On the hidden test set, a frozen Alpamayo 1.5 with rule-based camera and vehicle conversions and Q-adaptive guidance sets a new state of the art among published methods: 8.135159 rater feedback score, against 8.079 for the best published method.
59. 【2610.39955】Mind The Gap (MTG-AD): Dataset and Fine-grained Evaluation for End-to-End Inline Audio Descriptions
链接：https://arxiv.org/abs/2610.39955
作者：Subhashini Venugopalan,Jenny Tan,Taylor Roper,Jimmy Tobin,Anton Kast,Alicia Martin,Sam Sepah,Amy Pavel
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
关键词：Multimodal Large Language, Audio descriptions, Large Language Models, making visual media, visual media accessible
备注： NeurIPS 2026 Evaluations Datasets Track, camera ready
点击查看摘要
Abstract:Audio descriptions (AD) are essential for making visual media accessible to blind and low-vision (BLV) users. While Multimodal Large Language Models (MLLMs) offer a scalable solution for generating audio descriptions on-demand, their performance on ``inline'' audio descriptions -- which must fit within existing silences in a video -- remains under-explored, particularly for diverse user-generated content (UGC). We present a comprehensive investigation into MLLM-generated inline audio descriptions. First, we introduce MTG-AD, a professionally annotated dataset of 36 hours of videos, each averaging $\sim$5 mins. and 8k+ descriptions in total, co-created by paired blind and sighted AD professionals to provide a high-quality benchmark for this task. Second, we propose MTG-AD-Eval, an evaluation framework based on audio description expert guidelines that provides fine-grained, reference-free actionable metrics for model improvement. Crucially, our framework evaluates the end-to-end task: it renders the generated scripts into audio tracks to assess their fit within the original video's natural silences, minimizing disruptive overlaps. Our analysis across frontier MLLMs reveals that while models accurately describe visual events and substantially reduce speech overlaps, gaps persist in narrative pacing and flow. Finally, we empirically validate that our automated and reference-free metrics correlate significantly with human quality ratings from sighted raters and 7 BLV participants, finding that while users narrowly prefer human-authored audio descriptions, they find MLLM-generated audio descriptions highly beneficial for comprehension. We provide actionable system design guidance and release MTG-AD at this https URL and MTG-AD-Eval at this https URL.
60. 【2610.39948】UniRA: Unified Representation Alignment and Auto-Encoding for Efficient Diffusion Generation
链接：https://arxiv.org/abs/2610.39948
作者：An Lanji,Dawei Liu,Jin Li,Haoran Xu,Mei Chen,Yu Tian
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：high-fidelity image synthesis, high computational cost, image synthesis, computational cost, slow training convergence
备注： 7 pages, 5 figures; submitted to ICML2027;
点击查看摘要
Abstract:Diffusion transformers (DiTs) have become the dominant architecture for high-fidelity image synthesis, yet they suffer from slow training convergence and high computational cost. Two recent lines of work address this from complementary angles: Representation Alignment (REPA) accelerates DiT training by aligning intermediate features with a pretrained external representation model (e.g., DINOv2), while Representation Auto-Encoders (RAE) replace the pixel-space VAE with a compact representation-space tokenizer learned from self-supervised features. However, these approaches are developed and applied in isolation: REPA still relies on a pixel-space VAE for latent diffusion, and RAE does not exploit the alignment signal that accelerates the diffusion backbone. We present \textbf{UniRA}, a unified framework that couples representation alignment and representation auto-encoding into a single objective. UniRA learns a representation-space tokenizer whose output serves simultaneously as the diffusion target and the alignment anchor, so the tokenizer, the diffusion backbone, and the external representation model are co-optimized. We give a theoretical result showing that this joint optimization preserves the representation quality of the external model while accelerating diffusion convergence, and we derive a principled weighting that balances the reconstruction, alignment, and generation objectives. On ImageNet class-conditional generation at 256*256 and 512*512, UniRA matches or surpasses state-of-the-art DiTs (SD3, REPA, RAE) in FID while reducing training steps by up to 5*. We further show that the unified representation space supports improved zero-shot classification and linear probing, confirming that generation and understanding benefit from a shared, aligned latent space.
61. 【2610.39946】ElasticFormer: Detecting Objects in HRW Shots via Elastic Computing Vision Transformer
链接：https://arxiv.org/abs/2610.39946
作者：Wenxi Li,Jingchen Huang,Chenyang Lyu,Moran Liu,Haozhe Lin,Guiguang Ding,Yuchen Guo
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：brought High-Resolution Wide, High-Resolution Wide shots, Recent advances, High-Resolution Wide, Wide shots
备注： Published at CVPR 2026. 14 pages including supplementary material. Code: [this https URL](https://github.com/Purecoo00/ElasticFormer)
点击查看摘要
Abstract:Recent advances in gigapixel-level imaging have brought High-Resolution Wide shots to the forefront of research. However, these images present significant challenges: extreme sparsity of foreground, gigapixel-level resolutions and diverse target counts. This makes traditional close-up detectors inaccurate and slow as they are overwhelmed by the background. Although previous research has explored sparse backbones, their fixed sparsity patterns lack the adaptability required to handle diverse target numbers. To address this, we introduce ElasticFormer, a sparse backbone that dynamically allocates computational resources based on foreground proportion. After scoring windows based on variance, proposed ElasticSelector module will predict the foreground proportion for top-k selection. The mechanism guides the model to select target-containing windows, scaling resources in areas where objects are clustered. We introduce a novel loss function combined with the 3-phase training strategy for ElasticSelector, allowing it to function properly when bounding box annotations are missing. A WSOD study is carried on PASCAL VOC 2007 to evaluate its extensibility. Further, ElasticNet is created to verify its backbone-agnostic nature. In experiments on the PANDA gigapixel benchmark, ElasticFormer reduces backbone FLOPs by 80% while achieving a significant improvement in AP50 when compared to fixed-ratio sparse methods.
62. 【2610.39937】BiHashFormer: Hash-Driven Dual-Branch Transformer for Efficient Object Detection in HRW Shots
链接：https://arxiv.org/abs/2610.39937
作者：Jingchen Huang,Wenxi Li,Chenyang Lyu,Moran Liu,Haozhe Lin,Yuchen Guo
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：brought High-Resolution Wide, High-Resolution Wide shots, Recent advances, High-Resolution Wide, Wide shots
备注： Accepted at NeurIPS 2026. 34 pages including supplementary material and the NeurIPS checklist. Code: [this https URL](https://github.com/Purecoo00/BiHashFormer)
点击查看摘要
Abstract:Recent advances in gigapixel-level imaging have brought High-Resolution Wide shots to the forefront of research. However, these images present significant challenges: extreme sparsity of foreground, gigapixel-level resolutions and interleaving of foreground and background. This causes traditional detectors and attention mechanisms to be hindered by background, resulting in inefficiency and inaccuracy. To tackle this problem, we propose BiHashFormer, a dual sparsified branch transformer built on hashing. RoI Selector will predict the foreground proportion for top-k selecting target-containing windows. The compression branch processes windows selected and refines attention into HashAttention by discarding half of the key-value pairs, reducing cost and improving efficiency. The compensation branch uses HashMiner to perform low-cost hash searches in the remaining regions, recovering features of missed objects. Experiments confirm that the selection relationship in HashAttention is one-way and reveal the preference of different queries for keys. In experiments on the gigapixel benchmark PANDA, BiHashFormer reduces 37.5% backbone FLOPs while improving AP50 to 81.0%. Moreover, this parallel dual-branch structure allows the model to avoid inference delays caused by inter-branch dependencies, while achieving much greater stability than single-branch models, specially with low rates of RoI selection.
63. 【2610.39936】MLLM-GS: Multimodal Large Language Model Prior-Guided Fast Novel View Synthesis in Extreme Environments
链接：https://arxiv.org/abs/2610.39936
作者：Zhiliang Wu,Xinye Zheng,Kun Li,Fei Wang,Hao Guo,Yanyan Wei,Guanghui Yue,Beier Zhu,Wei Zhou,Xun Yang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：multi-view observations affected, aims to reconstruct, unseen viewpoints, generate clear, clear and photorealistic
备注：
点击查看摘要
Abstract:Novel view synthesis (NVS) in degraded scenes aims to reconstruct 3D scenes from multi-view observations affected by various degradations, and to generate clear and photorealistic images at unseen viewpoints. Despite recent advances, existing methods often rely on degradation-specific models, limiting their generalization across diverse extreme environments. Generative models provide rich visual priors, but their outputs often suffer from pixel misalignment and color deviations, which can propagate into the reconstructed 3D scene and compromise reconstruction quality. Furthermore, the high computational and memory costs of 3D reconstruction hinder the practical application of NVS in extreme environments. To this end, we propose MLLM-GS, a multimodal large language model (MLLM) prior-guided framework for fast NVS in extreme environments. MLLM-GS consists of two branches: MLLM prior-guided view enhancement and fast 3D scene reconstruction. The former leverages the rich visual priors of a frozen MLLM to enhance views under different severe degradations and introduces a lightweight correction network to reduce pixel shifts and color distortions in MLLM outputs. With supervision from only a few paired clean views, this branch produces reliable enhanced views for all degraded observations, providing pseudo-GT views for subsequent 3D reconstruction. The latter improves 3DGS efficiency by optimizing rasterization, Gaussian memory layout, and parameter updates, reducing redundant computation, memory-access overhead, and optimization costs. Extensive experiments on multiple degraded scenes demonstrate that MLLM-GS effectively handles various severe degradations and achieves high-quality NVS with significantly lower reconstruction time and memory consumption. Project Page: this https URL
64. 【2610.39926】Spectral Invariant Fusion and Refinement for Domain Generalized Semantic Segmentation
链接：https://arxiv.org/abs/2610.39926
作者：Yansong Bai,Yunzhi Zhuge,Xinzhuo Yu,Lu Zhang,Jin Zhan,Pingping Zhang,Huchuan Lu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：requires accurate dense, requires accurate, access to target-domain, accurate dense prediction, Domain generalized semantic
备注：
点击查看摘要
Abstract:Domain generalized semantic segmentation is challenging because it requires accurate dense prediction on unseen domains without any access to target-domain data during training. Existing methods improve transferability through data diversification, invariant learning, or direct use of foundation models, but they often struggle to simultaneously suppress domain-sensitive interference and preserve fine-grained spatial cues. In this work, we propose SIFR, a frequency-aware dual-stream framework that tackles domain shift at both the fusion and decoding stages. SIFR uses spectral modeling as a unified principle to regulate both heterogeneous prior interaction and dense decoder refinement. Instead of directly fusing frozen CLIP and DINOv2 features, SIFR performs spectral-guided invariant fusion to filter out domain-sensitive responses while preserving complementary semantic and structural information. In addition, SIFR incorporates a spectral decoder refinement module that stabilizes multi-scale decoder features through low-frequency statistics-based normalization and adaptive frequency selection, thereby reducing the propagation of residual domain bias in dense prediction. This design enables SIFR to jointly improve domain robustness and segmentation precision under severe distribution shifts. Extensive experiments on multiple DGSS benchmarks, including synthetic-to-real, real-to-real, and adverse-weather settings, demonstrate consistent improvements over previous state-of-the-art methods.
65. 【2610.39924】PhysAg3D: Bridging 3D Generative and Agentic Models for High-Fidelity and Physically-Interactive Assets
链接：https://arxiv.org/abs/2610.39924
作者：Bi'an Du,Fanpeng Meng,Jiemin Fang,Chen Yang,Pufan Li,Daizong Liu,Qi Tian,Wei Hu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：requires realistic appearance, Generating interactive, single image requires, image requires realistic, realistic appearance
备注：
点击查看摘要
Abstract:Generating interactive 3D assets from a single image requires realistic appearance, fidelity to the depicted object, and plausible physical behavior. Yet detailed geometry alone does not determine how an object should move or deform. We introduce PhysAg3D, which couples 3D generative models with agentic reasoning and tool use. A vision-language agent first infers semantic parts, candidate interactions, and physical priors from the image; a 3D generator then produces a textured whole-object mesh from the same image. A hierarchical interaction representation links these revisable priors to mechanisms and the generated surfaces they control, organizing interaction construction and refinement. Guided by this representation and observations of the mesh, the agent uses geometric and physical tools to construct parts, joints, deformable regions, and surface bindings while retaining useful generated geometry and materials. Rendered interactions and execution measurements guide targeted revisions within a fixed budget. The framework accommodates both rigid articulation and compliant deformation. Quantitative and qualitative evaluations on real photographs and external benchmarks demonstrate detailed, image-aligned assets with plausible interactions. Ablations support bounded refinement and source-surface preservation. Project Page: this http URL.
66. 【2610.39913】okenRepel: Generating Diverse Image Sets Without Compromising Quality
链接：https://arxiv.org/abs/2610.39913
作者：Qingtao Yu,Zhen Wang,Ning Lai,Boyang Hu,Hongdong Li,Dylan Campbell
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：image sets generated, present an inference-time, inference-time approach, approach that enhances, compromising the average
备注： NeurIPS 2026
点击查看摘要
Abstract:We present an inference-time approach that enhances the visual diversity of image sets generated by text-to-image models, without compromising the average image quality. Our method leverages the intrinsic geometry of hidden representations in pretrained models to diversify images generated from the same prompt. First, we introduce a token repulsion mechanism that increases the angle between feature tokens across samples. It takes a negative Riemannian gradient step on a von Mises-Fisher mixture distribution of tokens on the unit hypersphere. Second, to preserve the quality of each output image, we propose a token density preservation mechanism that projects the repulsion-induced prediction change onto a half-space aligned with the model's score direction. This encourages the updates to remain within the high-density regions learned by the generative model. We further extend this mechanism to the output of each intermediate layer, enabling finer-grained iterative guidance. We evaluate our method on three types of pretrained text-to-image models. Experiments show substantial gains in visual diversity under identical prompts, while maintaining a comparable level of image fidelity.
67. 【2610.39889】sDMD: Sinkhorn Distribution Matching Distillation
链接：https://arxiv.org/abs/2610.39889
作者：Jaihoon Kim,Lingxiao Li,Soshi Shimada,Roshni Cooper,Minhyuk Sung,Yiwei Zhao
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：Distribution Matching Distillation, Distribution Matching, distills multi-step diffusion, multi-step diffusion models, Matching Distillation
备注： Project Page: [this https URL](https://sinkhorn-dmd.github.io/)
点击查看摘要
Abstract:Distribution Matching Distillation (DMD) distills multi-step diffusion models into few-step generators by minimizing reverse KL divergence. Despite its strong performance, DMD guides samples by local density alone, which introduces two limitations: teacher scores can be unreliable in regions of low teacher density, and the mode-seeking behavior of reverse KL can collapse the student onto a subset of the teacher modes. We first connect DMD to Wasserstein gradient flow, showing that its gradient follows from a fixed-point formulation with integral KL (IKL) as the energy functional. This perspective motivates augmenting IKL with an optimal transport term that addresses these limitations by guiding outlier samples toward the target and promoting mode coverage. We realize this term using Sinkhorn divergence estimated from finite batches. In practice, however, errors in the estimated Sinkhorn velocity can perturb in-domain samples for which IKL guidance is already accurate. To this end, we introduce sDMD, which adaptively activates the Sinkhorn correction based on discriminator outputs, focusing on outlier samples where IKL may be unreliable. With 2--4 sampling steps on FLUX.1-dev, Qwen-Image-20B, and Wan2.1-14B, sDMD outperforms distillation baselines on GenEval, DPG, WISE, and VBench while improving sample diversity over DMD2.
68. 【2610.39880】What Survives Label Downsampling? Exact Region Survival for Coarse Segmentation Targets
链接：https://arxiv.org/abs/2610.39880
作者：Zhiyuan Zhang,Yukun Ou,San Tek Choi
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Segmentation networks, label maps downsampled, nearest-neighbor resampling, networks that predict, maps downsampled
备注： 25 pages, 13 figures, including supplementary material
点击查看摘要
Abstract:Segmentation networks that predict at an output stride are trained against label maps downsampled by strided sampling, nearest-neighbor resampling or majority pooling. These methods label each coarse grid cell on its own, so a region that wins no cell is absent from the target and receives no positive label. We present MatchDown, which builds the coarse target from a minimum-cost matching between regions and the grid cells they occupy. It keeps as many regions as any one-region-per-cell target can, certifies how many losses are unavoidable, and with priority tiers (MatchDown-P) keeps every evaluated region (a region that a metric scores) of at least one cell's area. On COCO, Cityscapes and BBBC038 at factors 2 to 16, MatchDown-P loses exactly the minimum number of regions, whereas the downsampling used by StarDist, nnU-Net and connectomics pyramids loses four to five times as many on COCO at factor 8. Keeping a region in the target and reconstructing its shape are different problems: we show that no region lost by majority pooling can be reconstructed above IoU 0.5 by upsampling any coarse target, yet on BBBC038 a star-convex head trained on MatchDown targets segments 35% of the nuclei that majority pooling loses above IoU 0.5, against 15% when trained on majority targets and 24% on StarDist's strided target. On two nucleus datasets, with thresholds cross-validated for count error, this head has instance accuracy at IoU 0.5 (Acc50) 2.6 to 4.2 points above the strided target and at most 2.8 points below cell-center sampling, and a lower count error than the strided and cell-center targets on BBBC039. On BBBC039, an affinity head, which outputs one label per cell, lowers count error relative to majority targets by 9.3 nuclei per image at 7.2 points lower Acc50. MatchDown runs once, before training, in place of the label-downsampling step.
69. 【2610.39868】Parametric Gaussian Noise Mixer (PGNM): Unified Synthesis of Camera Noise and Film Grain
链接：https://arxiv.org/abs/2610.39868
作者：Mingyang Song(ETH Zürich)
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：generate noisy images, noisy images, synthesis is commonly, commonly approached, neural networks trained
备注： Technical report. 49 pages: 9-page main paper and 40-page supplementary material; 29 figures, 33 tables
点击查看摘要
Abstract:Noise synthesis is commonly approached through physical models of image formation or neural networks trained to generate noisy images. We explore a middle ground: an explicit statistical model fitted directly from paired clean and noisy images. A normalized convolution correlates a Gaussian field across pixels and color channels, and a lookup maps each Gaussian value to a residual given the clean intensity. Because the field stays standard Gaussian, the two components are fitted separately, without reproducing the physical process or training a neural generator. With one formulation for camera noise from SIDD and simulated film grain from Newson, the model has the lowest average distribution and correlation errors on SIDD among the evaluated methods and remains competitive on film grain. A denoiser trained on its noise comes closest to one trained on real noise, and a 1080p frame renders in 10 ms on GPU.
70. 【2610.39861】M-PFT: Angular Margin for Calibrated Prompt-Tuning of Fine-Tuned Vision-Language Models
链接：https://arxiv.org/abs/2610.39861
作者：Shihab Aaqil Ahamed,Mohamed Ansaff,Nang Khin Myat Lay Phyu,Udaya Sampath K. Perera Miriya Thanthrige,Chamira U. S. Edussooriya,Ranga Rodrigo,Muhammad Haris Khan
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：adapting large-scale vision-language, large-scale vision-language models, pre-trained weights frozen, weights frozen, parameter-efficient technique
备注： Accepted at WACV 2027
点击查看摘要
Abstract:Prompt-tuning has emerged as a parameter-efficient technique for adapting large-scale vision-language models (VLMs) to downstream tasks while keeping the pre-trained weights frozen. However, prompt-tuned VLMs often suffer from a base-to-novel dual miscalibration problem, where models typically show underconfidence on base classes and overconfidence on novel classes. Current VLM calibration approaches address this dual miscalibration either separately or jointly, but lack a geometric understanding of the hyperspherical embedding space and thus fail to exploit angular discriminative properties to adjust distorted decision boundaries. To address this, we propose M-PFT, a novel angular-margin approach for prompt tuning that introduces two complementary objectives to promote class discriminability and semantic anchoring. This enlarges the angular margin by 1) minimizing the maximum cosine similarity between tunable class embeddings to reduce embedding collapse and 2) minimizing the angular distance between each tunable embedding and its corresponding zero-shot embedding to preserve semantic geometry. Extensive experiments across 9 prompt-tuning methods with various backbones on 15 diverse datasets show that our approach reduces the average calibration error on novel classes, while remaining competitive on base classes and maintaining comparable accuracy. Further analyses reveal that M-PFT effectively adjusts distorted decision boundaries induced by prompt tuning while preserving the pre-trained embedding space, improving calibration with negligible computational overhead.
71. 【2610.39857】Open-set multi-view seahorse re-identification from underwater photograph sequences
链接：https://arxiv.org/abs/2610.39857
作者：Tristan Fauvel,Sylvie Louisy,Patrick Louisy,Thomas Fauvel
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Automated animal re-identification, facilitate minimally invasive, Automated animal, studies at scale, minimally invasive
备注：
点击查看摘要
Abstract:Automated animal re-identification from photograph sequences can facilitate minimally invasive Capture-Mark-Recapture studies at scale. Yet, photographs taken during an observation session can vary greatly in quality, not always providing the necessary information for accurate re-identification. In this paper, we present new methods for taxonomic classification, sex determination, and open-set re-identification from image sequences. We train and evaluate our models on a dataset of 35,914 photographs of 474 individual seahorses observed in the wild. We demonstrate that our approach, by combining information from multiple photographs, yields state-of-the-art performance.
72. 【2610.39848】LiDAR-SHIELD: A Fully Dockerized Adversarial Robustness Assessment Framework for LiDAR-Based 3D Perception Models
链接：https://arxiv.org/abs/2610.39848
作者：Alexandros Gkillas,Ioulia Kapsali,Christos Anagnostopoulos,Aris S. Lalos
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：perception underpins autonomous, underpins autonomous driving, safely requires stress-testing, models safely requires, perception underpins
备注：
点击查看摘要
Abstract:LiDAR-based 3D perception underpins autonomous driving and robotics, yet deploying these models safely requires stress-testing them against adversarial inputs before real-world use. Although state-of-the-art adversarial attacks exist, their implementations are often fragmented across individual repositories, each tied to a specific architecture or task and requiring manual environment reconstruction and model-specific integration code. We present LIDARSHIELD, a unified and fully Dockerized framework that reverses this workflow: rather than installing the target model inside the toolkit, LIDARSHIELD exposes any model through three lightweight HTTP endpoints, namely /info, /forward, and /gradient As a result, connecting a new model or dataset requires only a single loader function or adapter. The framework supports both 3D semantic segmentation and 3D object detection through the same engine-service protocol, integrating a task-aware suite of state-of-the-art attacks, including PGD and Square Attack for segmentation, and point perturbation, point removal, point injection, and LiDAttack for detection. This enables systematic exposure of corner cases and vulnerabilities across heterogeneous LiDAR backbones. The LiDAR-Shield codebase is publicly available at this https URL.
73. 【2610.39835】Aligning Medical Vision Language Models at the Capability Boundary with Entropy Adaptive Rubric Grounding
链接：https://arxiv.org/abs/2610.39835
作者：Hongyu Bai,Jiaqi Zeng,Chunzheng Zhu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：requires verifiable links, current medical vision, medical vision language, vision language models, reasoning requires verifiable
备注： Accepted at ACM Multimedia Asia 2026 (MMAsia 2026). 10 pages, 4 figures
点击查看摘要
Abstract:Medical diagnostic reasoning requires verifiable links between visual evidence and clinical knowledge, yet current medical vision language models are aligned with objectives that reward the wrong behavior. Uniform supervised fine tuning lets easy cases dominate updates while clinically critical boundary cases saturate prematurely. Sparse rewards based only on final-answer correctness cannot assess anatomical grounding or clinical logic, inviting reward hacking through answer shortcuts. We propose MedCBR, a two-stage alignment framework coupling boundary-aware preference optimization with rubric-grounded reinforcement learning. First, we construct counterfactual boundary preferences at the model's capability frontier using empirical pass rates and preference pairs that share reasoning prefixes but diverge at decisive diagnostic branches. Difficulty Weighted Boundary Preference Optimization (DBPO) concentrates learning on this calibrated boundary region. Second, we distill reusable clinical rubrics from boundary trajectories and activate them according to prediction entropy. Entropy Gated Grounded Policy Optimization (EGPO) rewards answer correctness, anatomical grounding, and reasoning fidelity while suppressing rubric noise on reliable samples. A reference policy anchored to the boundary-aligned model couples population-level boundary priors with instance-level uncertainty, making diagnostic reasoning more accurate and clinically grounded. Experiments on six medical benchmarks show consistent improvements over strong general, specialized, and reinforcement learning baselines, reaching 61.2% average accuracy with the largest gains on challenging differential diagnosis tasks. Ablations show super-additive gains: combining the two stages exceeds the sum of their individual improvements.
74. 【2610.39812】GROME: GROunding Factorized MEmory into Video World Models
链接：https://arxiv.org/abs/2610.39812
作者：Fang Yu,Xiaoxue Wu,Haiyu Zhang,Yangfu Li,Botong Zhao,Yu Qiao,Tianfan Xue,Yue Lu,Xinyuan Chen
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：observed content falls, video world models, world models, Local Memory Retrieval, consistency is essential
备注：
点击查看摘要
Abstract:Long-horizon consistency is essential for video world models, yet it becomes difficult to maintain once observed content falls outside the context window. Existing memory mechanisms either retrieve history as implicit context, which remains constrained by context capacity and entangles static and dynamic information or consolidate history into explicit global representations, which struggle to model evolving dynamics and accumulate geometric errors. To overcome these limitations, we introduce \textbf{GROME}, a novel factorized memory framework that augments video world models for long-horizon coherence across both static scenes and dynamic entities. GROME operates via three synergistic components: 1)~\emph{Static-Dynamic Factorization} decouples historical evidence into static and dynamic components, preserving static scene geometry and dynamic entity states separately; 2)~\emph{Local Memory Retrieval} selects spatially aligned static history to eliminate geometric drift alongside the latest dynamic entity states to ensure temporal continuity; 3)~\emph{Training-Free Memory Injection} seamlessly grounds the factorized memory into video world models via their native conditioning pathways. Experiments on long-horizon revisit scenarios demonstrate that GROME substantially enhances scene persistence and dynamic entity fidelity over existing methods, all while preserving the visual quality and controllability of the frozen backbone without additional training. Beyond self-rollouts, GROME readily accommodates external observations as plug-and-play memory, unlocking flexible and controllable video continuation and scene exploration.
75. 【2610.39778】SCOPE: Semantic COdec Pre-Editing for Object-Preserving Compression
链接：https://arxiv.org/abs/2610.39778
作者：Silvia Dani,Leonardo Galteri,Marco Bertini
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：machine vision systems, downstream task accuracy, vision systems, machine vision, optimized for human
备注：
点击查看摘要
Abstract:Image compression is increasingly used before machine vision systems, yet standard codecs are mainly optimized for human perception rather than downstream task accuracy. We propose SCOPE (Semantic COdec Pre-Editing), a conditional image pre-editor that modifies images before compression to improve the bitrate and detection accuracy trade-off. The model is conditioned on the codec family and compression severity, allowing a single trained network to adapt to multiple compression settings without retraining or loading codec-specific weights. Instead of uniformly preserving the whole image, the pre-editor learns to protect task-relevant regions while simplifying less important background content to improve compressibility. The resulting rate-accuracy curves show consistent gains across different codec families and detector architectures, together with strong transferability to unseen learned and wavelet-based compression families.
76. 【2610.39773】Quantum-Inspired Convolutional Networks for High-Resolution Pseudo-Label Generation from Multimodal Earth Observation Data
链接：https://arxiv.org/abs/2610.39773
作者：Abhishek Singh,Giovanni Cuozzo,Alexander Jacob,Claudia Notarnicola
类目：Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
关键词：Accurate land-cover mapping, multimodal Earth observation, Earth observation, Accurate land-cover, costly to obtain
备注：
点击查看摘要
Abstract:Accurate land-cover mapping from multimodal Earth observation (EO) data remains a challenge in regions where high-resolution annotations are scarce or costly to obtain. This work introduces a quantum convolutional encoder-decoder network for pseudo-label generation using Sentinel-1 SAR and Sentinel-2 multispectral imagery, guided only by coarse MODIS-derived land-cover maps. The proposed framework uses quantum-inspired convolutional blocks to enhance feature extraction and cross-sensor fusion, enabling the model to infer a fine-scale semantic structure from heterogeneous inputs. Experiments were conducted on the DFC2020 dataset, which provides a quadruple of Sentinel-1 SAR, Sentinel-2 multispectral, low-resolution MODIS labels and high-resolution reference land-cover maps. The proposed training framework first learns a robust representation of multimodal inputs through the reconstruction branch before any supervision from low-resolution labels is applied. This self-supervised stage enables the model to capture a fine-scale spectral-spatial structure, which is then used to generate high-resolution pseudo-labels. These pseudo-labels are subsequently evaluated through classification metrics on the test data, demonstrating that the proposed approach produces more spatially coherent and class-consistent compared to classical baselines. The results highlight the potential of quantum-inspired deep learning to improve label generation in low-annotation remote sensing scenarios, offering a scalable path towards high-quality land-cover products for EO applications.
77. 【2610.39727】Cross-Encoder Anchor Refinement for Zero-Shot Anomaly Detection
链接：https://arxiv.org/abs/2610.39727
作者：Kaisong Ji,Ziyang Xu,Chang Bian,Yu Zhou,Wenyu Liu,Xinggang Wang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：localizing subtle defects, auxiliary training domains, transfer anomaly discrimination, Zero-shot anomaly detection, subtle defects
备注： 15 pages, 6 figures
点击查看摘要
Abstract:Zero-shot anomaly detection aims to transfer anomaly discrimination from auxiliary training domains to unseen categories while localizing subtle defects. A key challenge is to construct discriminative references that combine contextual representations with complementary local evidence. We present Cross-Encoder Anchor Refinement (CEAR), which connects two visual encoding paths for the same image at the anchor stage, before normal and anomalous prototypes are constructed. An independent convolutional encoder pretrained for industrial product recognition builds an image-specific memory shared across selected vision transformer (ViT) layers. Spatial anchors read from ViT patch features query this memory through cross-attention and undergo residual refinement. Image-conditioned normal and anomalous queries then gate anchor aggregation to construct prototypes for anomaly scoring. Across 13 industrial and medical datasets, CEAR achieves the best performance in 57 of 68 evaluations spanning image-level detection and pixel-level localization. Notably, CEAR surpasses all other state-of-the-art methods in 41 of 44 pixel-level comparisons. Across 11 localization datasets, mean gains in pixel-level maximum F1 score and average precision are 4.45 and 3.79 percentage points, respectively. This study shows how cross-encoder interaction at the anchor stage can integrate complementary visual representations for fine-grained anomaly localization across domains.
78. 【2610.39721】WorldProbe: When Video World Models Fall Apart under Long-Horizon Spatial Exploration
链接：https://arxiv.org/abs/2610.39721
作者：Minqi Wang,Jie Peng,Hejun Dong,Ruchang Yao,Puyi Wang,Siyuan Huang,Jihai Zhang,Yafu Li,Qi Dou,Xiaoye Qu,Yu Cheng
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：future visual observations, time and space, Video world models, goal of synthesizing, explored across time
备注： 19 pages, 9 figures
点击查看摘要
Abstract:Video world models generate future visual observations from camera or action inputs, with the goal of synthesizing persistent 3D worlds that can be explored across time and space. Existing benchmarks typically evaluate short interactions or rely on camera trajectories fixed before generation. The performance of video world models under long-horizon exploration with extensive camera motion remains under-explored. We introduce WorldProbe, a diagnostic benchmark that uses a goal-conditioned multimodal agent to adapt camera actions online during generation. WorldProbe comprises 164 test cases across four probing families. Across 10 state-of-the-art models, we generate more than 1,600 rollouts totaling approximately 800 minutes, with horizons up to 180 seconds. Our evaluation across four dimensions reveals both the degradation of local measures and the emergence of long-range inconsistencies. We find that recent models sustain visual and content quality for longer, while long-range memory and evolution remain weak. These results expose a gap between locally stable generation and persistent world consistency that is largely missed by existing benchmarks. Code and benchmark resources are available at this https URL.
79. 【2610.39715】NIF: Normal-Intersection Fields for Geometrically Consistent Oriented Box Refinement
链接：https://arxiv.org/abs/2610.39715
作者：Shaokai Zheng
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Accurate oriented detection, standard regression head, oriented detection requires, regression head predicts, follow object boundaries
备注： 17 pages, 6 figures, 5 tables
点击查看摘要
Abstract:Accurate oriented detection requires a box to follow object boundaries, yet a standard regression head predicts its center, size, and angle from a pooled region representation. Small errors in these coupled parameters can displace long boundaries even when the object is correctly classified. We propose NIF, a refinement module that replaces an additional box regressor with normal-intersection fields. Given a coarse oriented box, the module observes local features at 20 boundary anchors and predicts signed distances along their outward normals. These observations admit an exact affine representation in five geometric variables. A differentiable, reliability-weighted least-squares fit projects the predicted field onto this representation and analytically recovers a refined rectangle. Tangential pooling preserves the normal position profile, while shared projection limits the added network to 0.132M parameters. On DOTA-v1.0, NIF reaches 78.00%, 79.11%, and 79.02% mAP with Oriented R-CNN, LSKNet-S, and Strip R-CNN, respectively, achieving the highest single-scale SOTA result in Table 1. It reaches 68.19% on DIOR-R, improves matched FAIR1M/DIOR-R baselines by 0.30/0.68%, and raises HRSC mean IoU from 0.867 to 0.888 while reducing field error by 24.5%. Across four different detector frameworks with the detectors frozen, relative AP75 improves by 1.57-4.18% and COCO-style mAP by 0.42-3.74%, without detector retraining. Code is available at this https URL.
80. 【2610.39704】Confidence Filtering in Video-Based Gait Assessment: Measurement Error versus Selection Shift
链接：https://arxiv.org/abs/2610.39704
作者：David Regev,Eran Beeri Bamani
类目：Computer Vision and Pattern Recognition (cs.CV); Quantitative Methods (q-bio.QM); Methodology (stat.ME)
关键词：Video-based gait assessment, high-confidence pose observations, Video-based gait, pose observations, gait assessment
备注： 6 pages + 2-page appendix, 4 figures, 2 supplementary figures, 2 tables. Submitted to the 2027 IEEE International Conference on Rehabilitation Robotics (ICORR)
点击查看摘要
Abstract:Video-based gait assessment often keeps only high-confidence pose observations, which can lower measurement error but change the movement assessed. We compared confidence selection with budget-matched random selection in a Parkinson-related cohort (n = 96) and an NF2-related replication cohort (n = 18). In the Parkinson-related cohort, confidence selection produced a small but consistent shift in the observed movement summary (+0.05 SD, 95% CI 0.03 to 0.07), while held-out Mini-BESTest error stayed flat (-0.04 points, -0.13 to 0.05); a policy preserving motion-group proportions reduced the shift, while NF2 showed no shift and its predictor did not beat a constant baseline. Because this shift mixes measurement error with selection shift, we decompose and simulate both on the cohort's window structure. Filtering lowers local error when error depends on confidence, but biases the endpoint once confidence tracks movement more than error. In the Parkinson-related data, confidence was only weakly tied to movement (rho = 0.14). There, filtering is harmless if well-tracked windows are unbiased. If instead systematic error only redistributes within participants, filtering can add up to 0.58 degrees. Unchanged clinical error from such small-cohort predictors therefore cannot show that quality filtering preserves the assessed movement.
81. 【2610.39701】DEQ-Stereo: Deep Euilibrium Video Stereo Matching via Adaptive Convergence
链接：https://arxiv.org/abs/2610.39701
作者：Kiljoon Han,Kyumin Hwang,Wonheok Choi,Jeonghoon Kim,Sunghoom Im
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Iterative refinement models, update schedules incur, schedules incur heavy, incur heavy computational, heavy computational costs
备注： ACCV2026
点击查看摘要
Abstract:Iterative refinement models have significantly improved stereo matching accuracy, but their rigid, fixed update schedules incur heavy computational costs and a strict trade-off between speed and accuracy. This challenge is magnified in video streams, where traditional online methods resort to reducing feature channels or relying on expensive temporal warping. To address this, we propose \textbf{DEQ-Stereo}, a novel video stereo matching framework based on the Deep Equilibrium (DEQ) paradigm. Instead of unrolling a fixed number of iterations, DEQ-Stereo reformulates the refinement as an implicit fixed-point problem, adaptively terminating once convergence is met. To ensure fast and stable intra-frame convergence, we introduce the \textbf{Convolutional Simplified Gated Unit (ConvSGU)}, which eliminates unnecessary non-linearities by modeling the latent state as a gated, element-wise sum of inputs. Furthermore, to exploit cross-frame temporal coherence, we propose \textbf{Temporal State Transfer (TST)}, a warm-start strategy that propagates the converged hidden state to the next frame without requiring explicit pose or warping priors. Extensive evaluations demonstrate that DEQ-Stereo delivers state-of-the-art performance with a superior accuracy-to-latency trade-off.
82. 【2610.39676】Event-SAMA: Language-Grounded Video Object Segmentation on Neuromorphic Event Streams
链接：https://arxiv.org/abs/2610.39676
作者：Sergei Fedorchenko,Max Zehnder,Nikola Zubić,Rong Zou,Davide Scaramuzza
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：microsecond temporal resolution, high dynamic range, lacks event-stream datasets, cameras offer microsecond, offer microsecond temporal
备注： 15 pages
点击查看摘要
Abstract:Event cameras offer microsecond temporal resolution, high dynamic range and no motion blur, but referring video object segmentation (RVOS) lacks event-stream datasets and models. We introduce Event-SAMA-239K, a synthetic-event conversion of the SAMA-239K grounded video-chat corpus and all additional SAMA training data: 30,650 video clips and 392,162 images. Optical-flow, FILM or saccade upsampling followed by a linear event generator preserves all mask, referring-expression and dialogue annotations. We also introduce Event-SAMA, adapting SAMA-4B with a two-channel polarity patch embedding initialised from pretrained RGB weights and a zero-initialised event adapter. End-to-end training yields 24.37 $\mathcal{J}\\mathcal{F}$ on MeViS valid_u and 22.33 $\mathcal{J}\\mathcal{F}$ on the converted ReVOS valid subset, establishing the first event-native RVOS baseline (RGB SAMA-4B: 55.4 on MeViS). Frozen RGB SAMA-4B with pseudo-RGB renderings or E2VID reconstructions of the same events scores 13.23 and 15.94 $\mathcal{J}\\mathcal{F}$. Without retraining, Event-SAMA works on real event-camera recordings when the accumulation window matches their event density. A seed-controlled ablation with a measured noise floor of 0.47 $\mathcal{J}\\mathcal{F}$ finds gains from diverse training sources over MeViS alone (+4.99), RGB-mean event-stem initialisation (+4.33), and a trainable SAM2 mask decoder (+2.76); the event adapter, partial ViT unfreezing and event polarity are within noise at this budget. An earlier, partially converted corpus snapshot trained with the identical recipe scores 1.38 $\mathcal{J}\\mathcal{F}$ higher on MeViS and 1.12 higher on ReVOS. Both are single runs that also differ in schedule length, so we draw no conclusion about corpus scale.
83. 【2610.39664】EditAgent: Rethinking Agentic Image Editing
链接：https://arxiv.org/abs/2610.39664
作者：Haiyang Xu,Tianyu Wang,Jialu Li,Jiteng Mu,Yuanjun Xiong,Jianming Zhang,Zhuowen Tu,Mingze Xu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Agentic image editors, decompose complex requests, Agentic image, image editors decompose, editors decompose complex
备注：
点击查看摘要
Abstract:Agentic image editors decompose complex requests into steps, but when a step fails, they mostly retry with the same narrow set of tools and a judge that scores each output by eye directly. We argue that such agents are bounded by two spaces: the action space of what the agent can do, and the observation space of what its judge can see. On our expert-annotated benchmark, a general-purpose editor yields no usable result in over 40% of the cases, and no judge is reliable across edit types. We present EditAgent, which expands both spaces. It routes each step to general editors and specialized methods, and attaches to each candidate a verification record. With these records, off-the-shelf VLM judges outperform finetuned reward models. Candidates are generated in parallel and judged in one listwise call: experts prefer this to sequential refinement in 78.5% of decided comparisons, at no more than half the judge calls. EditAgent improves two open-weight editors on three benchmarks and serves as a data engine for training stronger editors.
84. 【2610.39663】AB-Attention: Accurate Block Attention for Accelerating Video Diffusion Transformers
链接：https://arxiv.org/abs/2610.39663
作者：Bin Xu,Mo Zhang,Zhengfeiyang Zhang,Xinfeng Yang,Yixing Ma,Wei Wang,HeXin Wang,Pengfei Zhou,Wangbo Zhao
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Video diffusion transformers, generate high-quality videos, diffusion transformers generate, transformers generate high-quality, Video diffusion
备注： 20 pages, 5 figures, 3 tables
点击查看摘要
Abstract:Video diffusion transformers generate high-quality videos, but the quadratic cost of 3D full attention dominates their inference. Dynamic block-sparse attention reduces this cost by computing exact attention only on key/value blocks selected from current activations. However, its fidelity degrades under aggressive sparsity: blocks are scored from coarse summaries or formed by costly global clustering, pruned blocks vanish from the softmax, and whole tiles are kept or dropped even when only a few of their keys matter. To address these limitations, we propose Accurate Block Attention (AB-Attention), a training-free method that accurately approximates dense attention at high sparsity. Its Intra-Block Clustering Selector (IBCS) estimates block importance faithfully yet cheaply by clustering queries only within each query block, without permuting tokens. Sampling with Inclusion-Probability Weighting (SIP) then samples blocks in proportion to these estimates and reweights them by inverse inclusion probability, so that one pass estimates dense attention rather than truncating it. Finally, its kernel gathers key/value blocks much smaller than a tile into full tensor-core tiles, refining selection at no extra tensor-core cost. On Wan2.1 and HunyuanVideo, AB-Attention skips 89--92\% of attention, whereas the baselines compute 1.8--3.6$\times$ as much, yet it still improves PSNR by up to 4.4~dB over the strongest baseline and achieves up to 2.23$\times$ end-to-end speedup. Our code is available at this https URL.
85. 【2610.39649】Look at Where You Think: Localization-Head Alignment for Fine-Grained Perception in Thinking VLMs
链接：https://arxiv.org/abs/2610.39649
作者：Yuan Wang,Wei Wu,Hongning Wang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Thinking vision-language models, miss small details, vision-language models reason, Thinking vision-language, high-resolution images
备注： 23 pages, including references and appendices
点击查看摘要
Abstract:Thinking vision-language models reason for thousands of tokens, yet still miss small details in high-resolution images. Zooming methods pay for extra inference passes, and outcome rewards never tell the model where to look at. In Qwen3.5-4B, we find that the bottleneck is selection, not seeing: a few attention heads, which we call localization heads, pick the evidence out of thousands of visual tokens; ablating four of them costs 13.2 points more than ablating four random heads, and their focus fades as the reasoning grows. We therefore propose Localization-Head Alignment (LHA), which makes attention a training signal: whenever the reasoning states a visual fact, these heads learn to look at the focal region rather than the background, as an auxiliary loss in both SFT and RL training. On benchmarks of V*, HR-Bench 4K and TreeBench, LHA lifts Qwen3.5-4B from 74.1% to 76.6% accuracy, the best single-pass 4B result we evaluate, ahead of multi-pass zooming and released 7B models. Alignment alone adds 1.8 points over otherwise identical training while shortening reasoning by 20%, and the gains carry over to Qwen2.5-VL-3B.
86. 【2610.39648】Annotation-Free Representation Learning for Polyp Retrieval in Colonoscopy Videos
链接：https://arxiv.org/abs/2610.39648
作者：Luca Parolari,Lamberto Ballan,Carlo Biffi,Loic Le Folgoc
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：reliable visual assessment, complementary views acquired, inherently comparative, reliable visual, imaging conditions
备注： Under submission at ISBI 2027
点击查看摘要
Abstract:Colonoscopy is inherently comparative: reliable visual assessment of polyps often depends on complementary views acquired under different viewpoints and imaging conditions. Retrieving such views requires polyp tracklet representations, which recent contrastive methods learn from temporal cues but still build from manually annotated bounding boxes. We propose an annotation-free framework that constructs tracklet views directly from raw colonoscopy videos, enabling temporal self-supervised learning on full-procedure videos, including frames without visible polyps. On REAL-Colon, our method achieves state-of-the-art results on tracklet retrieval and re-identification, outperforming both self-supervised and supervised baselines. We then introduce few-shot polyp retrieval, a task closer to clinical use, where a few observations of a target polyp are used to retrieve its remaining tracklets: with only two examples, our method surpasses most competing methods given ten. Finally, we show that a modest number of unlabeled procedures suffices for competitive performance, making on-site model development more practical. Code is available at this https URL.
87. 【2610.39646】FlexTry-On: A Time-Efficient Framework for Multi-Scenario Virtual Try-on
链接：https://arxiv.org/abs/2610.39646
作者：Phuong Dam,Quyen Nguyen Duc,Son Hoang Minh,Daeyoung Kim
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：achieved remarkable progress, Virtual try-on, virtual try-on scenarios, diverse virtual try-on, recent years
备注：
点击查看摘要
Abstract:Virtual try-on has achieved remarkable progress in recent years, yet most existing methods are designed for a specific scenario, such as cloth-to-person or person-to-person try-on, and often require separate conditioning strategies and processing pipelines for different settings. We present FlexTry-On, a unified and time-efficient diffusion-based framework for diverse virtual try-on scenarios, including cloth-to-person, person-to-person, cross-domain, and in-the-wild settings. FlexTry-On adopts a common conditioning formulation, where a text-aware condition extraction strategy converts heterogeneous garment sources into a unified representation, and a dual-conditioning diffusion transformer jointly models garment appearance and person identity. Extensive experiments demonstrate competitive performance on both cloth-to-person and person-to-person benchmarks while maintaining fast inference. Moreover, the same framework can be directly applied to unseen cross-domain scenarios without requiring separate architectures or scenario-specific processing pipelines. These results highlight the potential of unified and efficient virtual try-on frameworks as a practical step toward more flexible and widely applicable virtual try-on systems.
88. 【2610.39639】Where Do Vision Language Models Fail at 3D Reasoning?
链接：https://arxiv.org/abs/2610.39639
作者：Khai Le-Duc,Nguyen Kim Hai Bui,An Ngo,Nam D. Hoang,Yueqi Song,Jiarui Liu,Thai Quoc Hoang,Fang Wu,Chris Ngo,Terry Jingchen Zhang,Mona T. Diab,Ran Xu,Silvio Savarese,Xiao-Ming Wu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Vision-language models, individual judgments, Vision-language, model, spanning individual judgments
备注： Preprint, 150 pages
点击查看摘要
Abstract:Vision-language models can get individual judgments right while struggling to recover and consistently use the exact properties of a 3D object. To characterize this gap, we introduce BedBench, a controlled benchmark of thirteen procedurally generated tasks spanning individual judgments and composite answers. Across twelve open-weight and four frontier models, per-color counting exposes a pronounced weakness: humans achieve 93.3% accuracy, compared with 7.2% for the best model. Targeted fine-tuning raises one model's accuracy from 3.0% to 68.6%, but reveals a further difficulty: its answers about the most common colors remain inconsistent with its own per-color counts. Supplying those counts as text improves its judgments, showing that recovering a property does not mean using it consistently. Reasoning, scaling, and added visual cues each improve individual judgments, but none makes exact answers reliable. To locate the difficulty, we probe Qwen3.5-27B: exact per-color counts remain recoverable at the language model's input, yet the model's answers rarely recover them (65.0% for the probe vs. 3.0% for the model). Image-removal controls further show that some apparent successes require no visual evidence. These findings motivate evaluating 3D reasoning by exact correctness, consistent use, and image dependence, rather than accuracy alone. Code and dataset are publicly available: this https URL
89. 【2610.39633】Intrinsic Distribution Matching
链接：https://arxiv.org/abs/2610.39633
作者：Deyuan Liu,Yansen Han,Kaiwen Zheng,Jun Xie,Jiacheng Liu,Zhenglin Cheng,Peng Sun,Jintao Zhang,Jianguo Li,Tao Lin
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：One-step and few-step, generation must accurately, opportunities for refinement, accurately render, render the details
备注：
点击查看摘要
Abstract:One-step and few-step text-to-image generation must accurately render the details specified in the prompt despite limited opportunities for refinement. Distribution matching distillation (DMD) uses score differences to bring the generated distribution closer to a reference distribution. Following prior work, we estimate this difference with internal velocity branches, avoiding a separate fake-score network. We ask whether an explicit moment constraint improves this learned score supervision. A decomposition of the reverse-KL field into a Gaussian moment component and a non-Gaussian residual identifies statistics that can be constrained explicitly. We introduce Intrinsic Distribution Matching (IDM), which retains the full DMD correction and matches feature means and per-coordinate standard deviations using the generator's exponential-moving-average (EMA) copy, requiring no external feature encoder. Experiments use full-parameter and low-rank adaptation (LoRA) tuning on Z-Image-Turbo 6B and Qwen-Image 20B. At two network evaluations, IDM reaches 94.25% and 90.20% on Chinese and English LongText-Bench, the highest scores among the compared methods at this budget. Controlled ablations show gains from the added moment loss, while generator EMA features perform competitively with an external DINOv2 encoder across the five reported metrics. Code is available at this https URL.
90. 【2610.39619】FutureWorld: A Plug-and-Play Causal Next-State Predictor for Interactive Video World Exploration
链接：https://arxiv.org/abs/2610.39619
作者：Guangyuan Li,Kelu Yao,Shuolin Xu,Xinlei Yu,Wei Xing,Lei Zhao,Peng-Tao Jiang,Xiaogang Xu,Xiaobin Hu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：follow user actions, follow user, maintaining scene consistency, maintaining scene, reuse generated latents
备注： 17 pages, 6 figures
点击查看摘要
Abstract:Interactive video world models aim to follow user actions while maintaining scene consistency over extended interactions. History-based methods reuse generated latents to preserve scene content, but three limitations of this reuse contribute to error recirculation. 1) The latent generated at each interaction is retained as history, potentially containing blurred details, geometric distortions, or semantic drift. 2) Although retrieved historical content matches the current scene, it can still contain errors that are carried into the next chunk. 3) New outputs then become history for later interactions, potentially reinforcing the same errors through repeated reuse. To limit this cycle, we aim to predict the future scene state from the current generated latent and the next action before generating the next interaction's video chunk. This future-state prior provides guidance for the next chunk, with the aim of reducing the errors carried forward from history. To this end, we propose a plug-and-play Causal Next-Chunk Predictor (C-NCP) with cross-model adaptability and prediction-guided modulation. We integrate C-NCP into four base generators and evaluate the augmented models across three benchmarks for interactive long-video generation and action-conditioned world modeling. Extensive experiments demonstrate reduced error accumulation and improved long-horizon consistency across diverse generation settings. The project page is at this https URL.
91. 【2610.39617】UniTele: Learning Device-Conditioned Degradations for Unified Smartphone Telephoto Super-Resolution
链接：https://arxiv.org/abs/2610.39617
作者：Yuchen Zhang,Aojie Li,Wei Cheng,Xianfang Zeng,Runxing Liu,haixin hao,Gang YU,Xiaoyun Yuan
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：mobile phones suffers, conventional handcrafted pipelines, suffers from complex, conventional handcrafted, Super-telephoto imaging
备注：
点击查看摘要
Abstract:Super-telephoto imaging on mobile phones suffers from complex, device-dependent degradations that are poorly captured by conventional handcrafted pipelines. Collecting large-scale aligned training pairs across different phones and diverse scenes, however, is both costly and difficult to scale. We therefore study whether device-specific degradations can be learned from only a small number of physical captures and transferred for training a unified super-resolution (SR) model. Our key idea is to factor degradation learning into two learnable components: one captures the device-specific imaging characteristics under well-resolved conditions, while the other captures the additional information loss and artifacts that arise in super-telephoto imaging. We build a paired benchmark across three phone brands (Huawei, Apple, and vivo) by photographing the same content at two display sizes and zoom settings. The larger displayed image is captured at the phone's maximum optical zoom setting, while the smaller image is photographed directly at the phone's highest native zoom setting through the phone's native imaging pipeline. The benchmark contains 519 training and 60 held-out pairs. We use the benchmark to learn device-specific degradation, then apply it to external images to synthesize thousands of additional pairs to train our UniTele SR model. Relative to the pretrained Qwen and FLUX backbones before adaptation, the full UniTele training recipe substantially improves paired restoration fidelity and perceptual similarity on the held-out benchmark and yields markedly more stable and visually faithful restorations.
92. 【2610.39609】DiFuse: Fusing Native Diffusion Representations for Spatial Reasoning in Unified Multimodal Models
链接：https://arxiv.org/abs/2610.39609
作者：Ruixin Fan,Weining Wang,Jiahui Sun,Tailai Chen,Jing Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：routinely condition generation, remains largely asymmetric, capabilities remains largely, jointly support visual, representations routinely condition
备注： 20 pages, 6 figures
点击查看摘要
Abstract:Unified multimodal models (UMMs) jointly support visual understanding and image generation, yet information flow between these two capabilities remains largely asymmetric: understanding representations routinely condition generation, whereas generative representations are rarely reused for multimodal reasoning. In this work, we investigate whether intermediate representations from a UMM's native diffusion branch can enhance spatial reasoning without compromising general multimodal understanding. We propose DiFuse, which selectively incorporates intermediate diffusion representations into the understanding pathway of UMMs. Through systematic layer-wise analysis, we identify which diffusion representations are most informative and where they can be integrated with minimal interference to semantic processing. Guided by these findings, DiFuse spatially aligns diffusion features with visual tokens and injects them into selected multimodal decoder layers via residual fusion. Extensive experiments on UniWorld-V1 show that, DiFuse consistently improves spatial reasoning over a matched SFT baseline while maintaining broadly comparable performance on general multimodal benchmarks. Further analyses show that these gains depend critically on the choice of diffusion representations, injection positions, and the preservation of spatial correspondence between diffusion features and visual tokens. Results on OmniGen2 further demonstrate the transferability of DiFuse across unified multimodal architectures. Overall, our findings show that native diffusion representations can provide complementary spatial cues for multimodal reasoning when selectively integrated in a structure-preserving manner.
93. 【2610.39602】Evaluating Cultural Knowledge Editing in Vision-Language Models
链接：https://arxiv.org/abs/2610.39602
作者：Malvina Nikandrou,Georgios Pantazopoulos,Ioannis Konstas,Alessandro Suglia
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：recognize culture-specific concepts, underrepresented regions, Current Vision-Language Models, recognize culture-specific, knowledge
备注：
点击查看摘要
Abstract:Current Vision-Language Models (VLMs) frequently fail to recognize culture-specific concepts, especially from underrepresented regions. Effectively injecting this long-tail knowledge into existing models remains a challenge. In this work, we investigate knowledge editing as a targeted alternative by introducing CulturalBind, a benchmark comprising 3.2K images of 1K concepts from 6 domains, designed to evaluate culture-aware cross-modal knowledge as well as concept-level knowledge editing in VLMs. Pre-edit evaluation reveals that textual and visual knowledge is decoupled: models often demonstrate text-based knowledge of cultural concepts but fail to identify them from images. We then evaluate editing methods in single and sequential scenarios. Although current methods successfully correct individual visual recognition errors, they exhibit low cross-modal portability, failing to integrate edits into the model's broader parametric knowledge. Furthermore, performance degrades rapidly under sequential editing, highlighting the need for comprehensive adaptation approaches that better consolidate multimodal knowledge over time.
94. 【2610.39601】GC-LIF: Spiking Group-Equivariant Neural Network With Group-Coupled Membrane Dynamics
链接：https://arxiv.org/abs/2610.39601
作者：Nhan Trong Luu,Duong Trung Luu,Thang Cong Truong
类目：Neural and Evolutionary Computing (cs.NE); Computer Vision and Pattern Recognition (cs.CV)
关键词：Spiking neural networks, acquire rotation invariance, promise event-driven, low-power inference, data augmentation
备注： Under review
点击查看摘要
Abstract:Spiking neural networks (SNNs) promise event-driven, low-power inference, but on vision tasks they still acquire rotation invariance only through data augmentation. Group-equivariant CNNs build such symmetries into the architecture, yet a faithful spiking analogue is missing: a leaky integrate-and-fire (LIF) neuron placed after a $G$-equivariant convolution can break equivariance through its membrane recurrence. We give a necessary and sufficient condition for this not to happen. The layer is equivariant at every time step if and only if the neuron parameters $(\beta, v_{\mathrm{th}})$ are constant along the joint orbits of the group action on the (group, space) index. We solve it in closed form, and it covers any elementwise stateful neuron. Its sufficiency half specialises a known result for equivariant recurrent networks; its necessity half rules out the heterogeneous per-neuron parameters common in SNNs. The per-channel parameterisation that satisfies it, Group-Coupled LIF (GC-LIF), is combined with $p4$ and $p4m$-equivariant convolutions and a spike-rate group max-pool into SGE-Net. On MNIST, KMNIST, Fashion-MNIST, CIFAR-10 and CIFAR-100, evaluating each checkpoint on the clean test set and all its group-transformed copies, SGE-Net trained without rotations is invariant under $C_4$ and $D_4$ and outperforms a plain SNN trained with rotations in nine of ten dataset-group pairs, whereas non-equivariant baselines lose up to $72$ percentage points in the worst case. Operator-level tests confirm the theory before and after training, and an analytical estimate places SGE-Net at $1.7\times$ the arithmetic energy of a standard CNN and $2.3\times$ below its ANN counterpart at eight time steps.
95. 【2610.39592】Operator-Aware Posterior Sampling for Multivariate Weather Downscaling
链接：https://arxiv.org/abs/2610.39592
作者：Carlo Saccardi,Rudolf Saathof,Geethu Joseph,Justin Dauwels
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：reconstructing high-resolution meteorological, Statistical downscaling, high-resolution meteorological fields, task of reconstructing, Bayesian inverse problem
备注：
点击查看摘要
Abstract:Statistical downscaling is the task of reconstructing high-resolution meteorological fields from coarse inputs. A principled approach formulates downscaling as a Bayesian inverse problem, using a diffusion prior over high-resolution fields and incorporating the coarse field only at inference through a likelihood. This decouples the prior from the coarse source, allowing one pretrained model to adapt to different sources. However, two challenges limit this flexibility: (1) when coarse and target fields come from different reanalysis systems, no fixed coarsening operator can capture their relationship; and (2) guidance gradients can differ by orders of magnitude across variables, making a single guidance strength ineffective. This paper addresses both challenges within the guidance loop, leaving the prior unchanged. We learn the forward operator from paired data and normalize the gradient per channel. We show that this normalization corresponds to steepest descent under a block-max norm, allowing a single guidance strength across variables. Our code is publicly available at this https URL.
96. 【2610.39585】actoHand: Learning Geometry-Based Dense Hand-Surface Interaction Fields from Video
链接：https://arxiv.org/abs/2610.39585
作者：Ziheng Xi,Weibo Li,Jiayi Yu,Sipeng Zheng,Wanpeng Zhang,Haiweng Xu,Hao Luo,Chuan Mao,Zongqing Lu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Understanding how humans, hand surface, humans interact, representation of physical, world requires
备注： 23 pages, including references and appendices
点击查看摘要
Abstract:Understanding how humans interact with the world requires a representation of physical interactions at the hand surface. Tactile sensing captures these interactions, but collecting tactile data at scale requires specialized hardware. We introduce TactoHand, a model that predicts three dense hand-surface interaction fields---contact, proximity, and 3D force---from monocular video. We derive geometry-based contact and proximity labels and motion-derived force labels from eligible training data within a curated collection spanning approximately 50 million source frames. TactoHand uses a temporal transformer to aggregate visual evidence across frames, while three learned queries decode the corresponding hand-surface fields. A differentiable physics layer further constrains force predictions using observed object motion. Across in-domain and out-of-domain benchmarks, TactoHand improves contact and proximity estimation over prior methods. In addition, physics supervision reduces position and rotation errors in motion rollouts, and the predicted contact and proximity fields improve downstream hand-object reconstruction. Together, these results show a scalable, low-cost path to learning hand-surface interaction fields without collecting tactile measurements. Code will be released at this https URL.
97. 【2610.39584】DOSE: Dominance-Guided Ordinal Severity Estimation from Sparse Relative Annotations
链接：https://arxiv.org/abs/2610.39584
作者：Tushir Sahu,Hideaki Hayashi,Hajime Nagahara
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：estimation commonly relies, severity estimation commonly, Medical severity estimation, Sparse Relative Annotations, Ordinal Severity Estimation
备注： Accepted to WACV 2027. 8 pages, 4 figures. Adjust the counts if you add the supplementary material
点击查看摘要
Abstract:Medical severity estimation commonly relies on absolute grade annotations, which are fundamentally ordinal, yet existing methods often treat the grades as independent categories. Learning-to-rank methods, which can be trained using relative annotations based on severity comparisons of image pairs, can provide continuous severity rankings but cannot directly predict discrete severity grades. We propose Dominance-guided Ordinal Severity Estimation from Sparse Relative Annotations (DOSE), a framework for learning ordinal medical severity from sparse pairwise comparisons without using absolute annotations. DOSE aggregates the observed comparisons through a global comparison graph to estimate a dominance score for each image. Differences between dominance scores set cumulative margins in the embedding space, so that pairs with larger estimated severity differences are separated more strongly, and also generate pseudo-pairs that supervise a scalar severity head alongside the observed comparisons. The learned continuous severity axis is converted into discrete grades via label-free score calibration, with tie-anchored segmentation as the main calibration method. In the experiment, DOSE consistently outperformed pairwise-ranking baselines on ACNE04, IDRiD, and LIMUC under different pair budgets. Code is available at this https URL
98. 【2610.39553】BEACON: Distilled Begin/End-Aware Structured Decoding for Continuous Sign Language Segmentation
链接：https://arxiv.org/abs/2610.39553
作者：Guohou Zhang,Kalin Stefanov
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：segmentation-based continuous sign, segmentation-based continuous, untrimmed video, lexical units, continuous sign language
备注： Accepted at the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2027. 12 pages, 4 figures, including supplementary material
点击查看摘要
Abstract:In segmentation-based continuous sign language recognition, an untrimmed video is first partitioned into lexical units; segmentation is therefore a means to an end---downstream recognition. However, segmenters are usually judged by local boundary detection rather than by the quality of the lexical segments and the recognition they enable. We present BEACON, a begin/end-aware segmenter that combines four-class Outside/Begin/Inside/End (BIOE) labeling, constrained Viterbi decoding over eight legal transitions, and boundary knowledge distilled from a binary teacher into a shared student. Under strict no-cut evaluation with validation-based model selection, BEACON improves segment-F1 by 14.58/14.59 points and boundary-F1 by 2.17/1.22 points, and reduces the downstream word error rate (WER) by 9.10/4.15 points relative to the state-of-the-art in the PHOENIX-14/BSL Corpus. To examine which segmentation metric better reflects downstream recognition, we evaluated five distinct architecture families. Segment-F1 aligns more consistently with lower WER than boundary-F1 in both corpora (Spearman $\rho=-1.0/-0.9$ versus $-0.9/-0.8$). These results support evaluating sign language segmentation in terms of segment quality and downstream recognition utility rather than local boundary accuracy alone.
99. 【2610.39551】Comprehensive empirical evaluation of feature extractors in computer vision
链接：https://arxiv.org/abs/2610.39551
作者：Murat Isik
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：Binary Robust Invariant, Invariant Scalable Keypoints, Robust Invariant Scalable, Binary Robust Independent, Accurate Invariant Scalable
备注： 26 Pages, 9 Figures, 6 Tables
点击查看摘要
Abstract:Feature detection and matching are fundamental components in computer vision, underpinning a broad spectrum of applications. This study offers a comprehensive evaluation of traditional feature detections and descriptors, analyzing methods such as Scale Invariant Feature Transform (SIFT), Speeded-Up Robust Features (SURF), Binary Robust Independent Elementary Features (BRIEF), Oriented FAST and Rotated BRIEF (ORB), Binary Robust Invariant Scalable Keypoints (BRISK), KAZE, Accelerated KAZE (AKAZE), Fast Retina Keypoint (FREAK), Dense and Accurate Invariant Scalable descriptor for Yale (DAISY), Features from Accelerated Segment Test (FAST), and STAR. Each feature extractor was assessed based on its architectural design and complexity, focusing on how these factors influence computational efficiency and robustness under various transformations. Utilizing the Image Matching Challenge Photo Tourism 2020 dataset, which includes over 1.5 million images, the study identifies the FAST algorithm as the most efficient detector when paired with the ORB descriptor and Brute-Force (BF) matcher, offering the fastest feature extraction and matching process. ORB is notably effective on affine transformed and brightened images, while AKAZE excels in conditions involving blurring, fisheye distortion, image rotation, and perspective distortions. Through more than 2 million comparisons, the study highlights the feature extractors that demonstrate superior resilience across various conditions, including rotation, scaling, blurring, brightening, affine transformations, perspective distortions, fisheye distortion, and salt-and-pepper noise
100. 【2610.39539】ROAD: Modern Adversarial Training for Simple One-Step Distillation at Scale
链接：https://arxiv.org/abs/2610.39539
作者：Xiaoou Li,Shan You,Tao Huang
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：turns pretrained diffusion, distillation turns pretrained, One-step distillation turns, pretrained diffusion models, One-step Adversarial Distillation
备注： 37 pages, 12 figures, 7 tables
点击查看摘要
Abstract:One-step distillation turns pretrained diffusion models into fast generators, but strong results often require substantial training machinery and compute. Pure adversarial distillation offers a simpler route, yet has struggled at modern text-to-image scale. To address this limitation, we propose ROAD (Representation-grounded One-step Adversarial Distillation), a one-stage framework that distills text-to-image models with one adversarial objective and a lightweight representation-space critic. First, we replace the classical convolutional head with a Lipschitz-controlled Transformer head with token-level text conditioning, which matches the semantic tokens of modern visual encoders. Second, we observed a fixed random feature mixing leads to long-horizon degradation. We attribute it to generator-critic co-adaptation around one realization-specific view, and periodically re-draw the mixing to keep the randomization effective. Third, for our dense and vector-valued logits, we introduce Jacobian R1 which penalizes the full input-logit Jacobian, overcoming pooled R1's limitation of neglecting most of logit sensitivities. A single adversarial objective then suffices: in one step, ROAD reaches 11.03 FID on SDXL and 13.73 on SD3.5-M on COCO-10K, surpassing the multi-step teachers at a fraction of distribution matching's training compute.
101. 【2610.39534】MOSAIC-Flow: Orthogonal-Axis and Status-Conditioned Fusion for Missing-Sequence Completion in Brain Tumor MRI
链接：https://arxiv.org/abs/2610.39534
作者：Pengli Zhu,Zhihao Peng,Yixuan Yuan
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：brain tumor characterization, comprehensive brain tumor, obscure sequence-specific contrast, three-dimensional tumor detail, fine-grained three-dimensional tumor
备注： 5 pages, 4 figures
点击查看摘要
Abstract:Missing-sequence completion is important for comprehensive brain tumor characterization, but cross-modal fusion can obscure sequence-specific contrast and lose fine-grained three-dimensional tumor detail. To address this challenge, we introduce Modality-aware Orthogonal-axis Status-conditioned Attentive Integration for Completion (MOSAIC-Flow), a unified framework for missing-sequence completion that learns a target-conditioned velocity field while keeping the seven acquired volumes fixed throughout the transport. Specifically, modality-aware stems independently encode each sequence, and orthogonal-axis refinement captures axial, coronal, and sagittal dependencies while retaining the three-dimensional representation. Status-conditioned prototype fusion uses compact modality and availability tokens to route and modulate pooled queries from the fused visual representation, enabling target-adaptive conditioning without dense voxel-to-voxel attention. A conditional flow matching objective transports target-channel noise toward the single missing sequence, while four Heun trajectories are averaged to produce the completion. Evaluation across structural, diffusion, and contrast-enhanced MRI in UCSF-PDGM demonstrates consistent whole-brain and tumor-region fidelity across all eight single-target completion tasks, with the strongest performance on structural sequences. Code is available at this https URL.
102. 【2610.39530】MorphSIG: Subject-Driven Image Generation via Decoupled Anchoring and Feature Transport
链接：https://arxiv.org/abs/2610.39530
作者：Hailong Yan,Yongrui Zhang,Xiangtao Zhang,Le Zhang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：reformulates subject-driven image, pseudo-video feature transport, Guided Feature Transport, Progressive Decoupled Anchoring, Pseudo-flow Guided Feature
备注：
点击查看摘要
Abstract:We propose MorphSIG, a training-free framework that reformulates subject-driven image generation (SIG) as pseudo-video feature transport. MorphSIG contains two core modules: Progressive Decoupled Anchoring, which uses frequency-domain decoupling and structural scrambling during early denoising to break structural locking and enable a gradual transition from free composition to identity alignment; and Pseudo-flow Guided Feature Transport, which constructs a latent-space pseudo-flow field to mimic Image-to-Video temporal coherence and transport subject semantics precisely. To address stylistic homogeneity and the lack of ground truth in existing benchmarks, we further introduce a benchmark with diverse stylized subjects. Extensive experiments show that MorphSIG significantly improves pose diversity and subject fidelity over baselines without additional training. Overall, MorphSIG bridges static generation and dynamic propagation, providing a cost-effective paradigm for high-fidelity subject consistency.
103. 【2610.39520】BRIDGE: Learning Brain-Gut Metabolic Evidence for Alzheimer's Disease Diagnosis
链接：https://arxiv.org/abs/2610.39520
作者：Pengli Zhu,Ruxi Chang,Tianliang Yao,Yu Jiang,Shengyuan Liu,Yiyang Lin,Xiaoyi Duan,Shuang Wu,Yixuan Yuan, theAlzheimer's Disease Neuroimaging Initiative
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Alzheimer disease, largely analyze cerebral, analyze cerebral glucose, cerebral glucose metabolism, existing models largely
备注： 10 pages, 6 figures
点击查看摘要
Abstract:Alzheimer's disease (AD) diagnosis may benefit from gut metabolic information associated with brain-gut interactions, yet existing models largely analyze cerebral glucose metabolism alone. Although whole-body fluorodeoxyglucose positron emission tomography (FDG-PET) can provide paired brain-gut observations, routine brain examinations rarely include gut imaging, limiting direct use of this information for diagnosis. In this paper, we present BRIDGE (Brain and Gut Representation Integration via Distribution-Guided Enhancement), a two-stage framework for transferring brain-gut metabolic knowledge to brain-only diagnosis. Specifically, BRIDGE integrates a conditional brain-to-gut latent transport module and a disease-evidence-guided cerebral pathway to capture subject-specific cross-organ correspondence and disease-relevant cerebral representations. For the cerebral pathway, a large language model (LLM) helps organize literature-derived descriptors into human-approved semantic anchors that guide feature learning. Additionally, we introduce a counterfactual-guided residual fusion mechanism, which uses matched and non-matched virtual-gut evidence to supervise a confidence gate and dynamically regulate the contribution of transported cross-organ information. At inference, BRIDGE exploits complementary gut-related context using brain PET alone, without measured gut PET. Extensive experiments across public and private cohorts demonstrate that BRIDGE outperforms brain-only and cross-organ baselines, achieving AUC gains of 6.0% and 5.2% over its brain-only counterpart for CN/MCI and CN/AD diagnosis, respectively. Notably, its ability to transfer complementary brain-gut metabolic context while preserving brain PET as the sole inference input highlights its potential for clinically practical diagnosis. Our code is available at this https URL.
104. 【2610.39517】FFAI: Flow Matching with Feedback Training and Adaptive Initialization for CT Reconstruction
链接：https://arxiv.org/abs/2610.39517
作者：Jigang Duan,Heran Wang,Genwei Ma,Xing Zhao
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：limited measurement information, distinct degradation mechanisms, involve distinct degradation, pose reconstruction challenges, reconstruction challenges due
备注：
点击查看摘要
Abstract:Although low-dose, sparse-view, and limited-angle CT involve distinct degradation mechanisms, all three pose reconstruction challenges due to limited measurement information and can therefore be addressed within a unified framework. In iterative image restoration with flow matching, prediction errors may cause inference states to deviate from the prescribed training path, while a fixed initialization may be unsuitable for inputs with varying degradation severity. We propose a unified conditional flow matching method that conditions on the initial reconstruction and combines feedback-aware training with degradation-adaptive initialization. During training, a teacher network simulates inference updates to generate shifted states, from which a student network learns to recover high-quality reference images, improving its ability to handle prediction errors. During inference, the starting time is selected based on the noise level or extent of missing projection data, and the noise added to the initial state is adjusted accordingly to adapt the starting point to the input. Experiments on public benchmark datasets show that the proposed method achieves better quantitative results than the compared methods across all three reconstruction tasks, effectively suppressing noise and artifacts while preserving anatomical structures and details.
105. 【2610.39516】Robust LIDAR Sensor-to-Sensor Domain Generalization via Accumulated-Point Scene Completion
链接：https://arxiv.org/abs/2610.39516
作者：Christoph B Rist,Larissa T Triess,Markus Enzweiler,Dariu M Gavrila
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：LIDAR, segmentation, real-world LIDAR exhibits, LIDAR exhibits sensor-specific, performance
备注： 2026 IEEE Intelligent Vehicles Symposium (IV)
点击查看摘要
Abstract:We study domain generalization for LIDAR semantic segmentation. Deep neural networks achieve strong performance on 3D point clouds, yet real-world LIDAR exhibits sensor-specific sampling patterns that hinder transfer across datasets. In many applications, performance on completely unseen sensors is crucial. A model should preferably work on a new setup without prior access to its data or labels. To this end, we introduce a source-only training scheme that adds an occupancy prediction auxiliary objective. We aggregate source-domain scans into a simple world model and use it as the supervision signal, replacing single-scan standard segmentation training. A key property of our approach is that it does not alter the input scan. During training and testing the network consumes the same single-frame LIDAR input, only the supervision signal is enriched via the occupancy objective. At deployment, there is no runtime overhead as the auxiliary head can be dropped for pure segmentation, or kept to obtain scene completion alongside semantic segmentation. In an experimental evaluation across distinct LIDAR datasets, we demonstrate that a model's generalization to new domains can be significantly affected by how it is supervised. In certain setups, performance on unseen domains improves without any explicit domain adaptation (no target scans, no target labels, no sensor priors) and without prediction refinement. At the same time, our method maintains competitive source-domain segmentation performance and incurs no additional inference cost.
Comments:
2026 IEEE Intelligent Vehicles Symposium (IV)
Subjects:
Computer Vision and Pattern Recognition (cs.CV)
Cite as:
arXiv:2610.39516 [cs.CV]
(or
arXiv:2610.39516v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.39516
Focus to learn more
arXiv-issued DOI via DataCite
Related DOI:
https://doi.org/10.1109/IV66570.2026.11623951
Focus to learn more
DOI(s) linking to related resources</p>
106. 【2610.39505】Agent-Wise Test-Time Normalization for Camera-Based Cooperative Perception with Corrupted Collaborators
链接：https://arxiv.org/abs/2610.39505
作者：Yu Wang,Shaohua Wang,Yicheng Li,Kecheng Sun
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Intermediate-fusion cooperative perception, vehicle detect objects, Intermediate-fusion cooperative, fusing features shared, cooperative perception
备注： 11 pages, 8 figures
点击查看摘要
Abstract:Intermediate-fusion cooperative perception lets a vehicle detect objects beyond its own field of view by fusing features shared by nearby agents. When only a collaborator's cameras are degraded by fog, motion blur or exposure errors, the features it shares drift away from the training distribution while the ego input stays clean, and cooperation can become worse than not cooperating at all. Test-time batch-normalization adaptation, the standard training-free remedy, estimates statistics over the pooled batch of all agents and therefore normalizes both the clean ego and the corrupted collaborator with a mixture of two distributions. We show that this mixing harms even the clean ego: when the ego shares BN statistics with a corrupted collaborator, its detection from its own features drops by 3.58 to 6.68 AP@0.5 points although its cameras are untouched. We propose AgentNorm, an agent-wise test-time normalization that i) estimates the normalization statistics of every collaborator from its own current observation and ii) anchors the ego to its source statistics, so that the ego path remains exactly the original model. AgentNorm has no learnable parameters, needs no gradients or labels, is independent of the order of test frames, and runs locally on each agent without changing the transmitted features. On camera-based OPV2V with four fusion checkpoints and collaborator-only corruptions, AgentNorm raises AP@0.5 under severity-5 fog from 29.42-37.25 to 43.32-58.86 and outperforms pooled normalization by 5.53-12.94 points under fog and motion blur, at an extra latency of 2.3 ms per frame. On the real-world DAIR-V2X-C dataset, it outperforms pooled normalization in all ten infrastructure-corruption settings. We also report the cases it does not fix: occlusion, which removes image content rather than shifting feature statistics, and the case in which all agents share the same shift.
107. 【2610.39483】Does Federated Training Degrade Explainability? A Comparative Analysis of Grad-CAM Faithfulness in Centralized vs. Federated Deepfake Detection
链接：https://arxiv.org/abs/2610.39483
作者：Muhammad Uzair Shahid,Muhammad Zabil Mehboob
类目：Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：realistic synthetic media, scale requires training, Automated detection, synthetic media, multiple organizations
备注： 10 Pages, 3 figures, Muhammad Uzair Shahid and Muhammad Zabil Mehboob contributed equally to this work
点击查看摘要
Abstract:Automated detection has become the essential safeguard against realistic synthetic media, but deploying it at scale requires training on sensitive data from multiple organizations, putting privacy and the transparency demanded by forensic review in direct conflict. Federated learning resolves the privacy side by sharing only model updates rather than raw data, but whether this comes at the cost of trustworthy gradient-based explanations, which analysts rely on to decide whether or not to trust a flagged clip, has not been studied. We present the systematic study of Grad-CAM explanation faithfulness under federated deepfake detection, evaluating an EfficientNet-B4 detector on FaceForensics++ across five federated clients under varying degrees of data heterogeneity. We find that the relationship between federation and explanation quality is non-monotonic: at near-IID (Independent and Identically Distributed Data) and moderate heterogeneity, federated models match or exceed the faithfulness of a centralized baseline, but at extreme heterogeneity, explanations break down entirely, becoming spatially incoherent and, on average, causally misleading. This degradation is not a general cost of federated training but a threshold effect tied to client heterogeneity, and we offer practical guidance for identifying when privacy-preserving deepfake detection can be deployed safely and when its explanations can no longer be trusted. Code and data partitions are available for reproducibility.
108. 【2610.39478】Regional Anchor Routing for Multi-Object Motion Transfer
链接：https://arxiv.org/abs/2610.39478
作者：Yuze Li,Fuchen Long,Xinlong Sun,Cheng Yan,Xinyu Zhang,Xi Chen
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：multiple reference videos, motion, requires a model, video Diffusion Transformer, motion transfer requires
备注： Project Page: [this https URL](https://ethan-li123.github.io/RART_page/)
点击查看摘要
Abstract:Compositional motion transfer requires a model to extract motions from multiple reference videos, and apply each motion to a user-specified object in one target scene. Existing methods normally struggle to reliably assign different reference motions to the correct target objects, thus a motion intended for one object may affect another. Meanwhile, some motion-specific tuning approaches are still facing another issue: they must run a new optimization whenever the reference video changes, making repeated transfers inefficient. To address these limitations, we propose RART, a framework that unifies object-wise motion extraction, target-region anchoring, and frame-aligned injection into a frozen video Diffusion Transformer for multi-object motion transfer. For each reference motion, RART extracts temporally ordered features from the masked reference video, and then conditions them on the target first-frame mask to form mask-anchored motion tokens. Frame-aligned motion cross-attention allows each generated frame to read only the motion tokens at the corresponding time slots, composing all object motions in one denoising process. During training, a motion-attention routing loss encourages each target region to attend to its corresponding motion tokens and reduces interference between objects. Therefore, RART can handle novel reference-target combinations without per-video optimization. Experiments show that RART achieves accurate motion preservation, efficient inference, and strong compositional generalization.
109. 【2610.39467】Calibrating Attention-Based Token Pruning for Efficient Multi-Image Synthesis in UMMs
链接：https://arxiv.org/abs/2610.39467
作者：Zimeng Wu,Qijian Tian,Penghui Du,Changqian Yu,Jiaxin Chen,Yunhong Wang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Unified Multimodal Models, Unified Multimodal, Multimodal Models, enable powerful multi-image, long visual contexts
备注：
点击查看摘要
Abstract:Unified Multimodal Models (UMMs) enable powerful multi-image synthesis but incur substantial computation from long visual contexts. While token pruning offers a direct route to accelerating inference, existing methods are primarily adapted from understanding tasks, suffering from insufficient reference image exploration, confused conditional guidance, and pooling-induced ambiguity that degrades generation quality. To address these issues, we propose CATP, the first training-free context-compression method tailored to UMM generation, which \textbf{C}alibrates the discrepancies of general \textbf{A}ttention-based \textbf{T}oken \textbf{P}runing for better quality-efficiency trade-off. Concretely, building on independent layer-wise pruning during denoising, we introduce Guidance-Consistent Pruning (GCP) to synchronize pruning decisions across conditioning branches, and a Taylor-Calibrated Attention Metric (TCAM) that derives a theoretically grounded second-order correction for pooling-induced bias. In addition, a Fast Scoring Operator (FSO) fuses the calibration into tiled online Softmax to reduce pruning cost. Extensive experiments across representative UMMs and benchmarks demonstrate that CATP consistently outperforms state-of-the-art token-reduction methods, achieving up to 2.09$\times$ attention-core speedup.
110. 【2610.39441】PathCowork: Learning from real-world pathology practice to build AI copilots
链接：https://arxiv.org/abs/2610.39441
作者：Sheng Wang,Jialu Yao,Maria V. Rossi,Shunsuke Koga,Esther Baranov,MacLean P. Nasrallah,Songhao Li,Jonathan Xu,Wei Du,Cindy McGrath,Timothy Chao,Anne Wang,Zhaohai Yang,Daniel Akuma,Selemon M. Walle,Olga Pozdnyakova,Edward B. Lee,Zahra Alipour,Annapurna Saksena,Rashmi Tondon,Nikolina Dioufa,David E. Elder,George J. Netto,Xiaowei Xu,Zhi Huang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：answers doctors record, doctors record, answers doctors, interpret the evidence, evidence to reach
备注： 56 pages, including supplementary information. Code, models and demonstration videos: [this https URL](https://github.com/zhihuanglab/PathCowork)
点击查看摘要
Abstract:Medical AI learns from the answers doctors record, but rarely from how they interpret the evidence to reach them. Here we introduce PathCowork, a framework that turns pathologists' narrated examinations into supervision for visual grounding and procedures that guide diagnostic AI agents. Synchronized gaze, narration and navigation link visual findings to the process of interpreting them. Across 48 sessions from 16 pathologists, the platform captured 1,028 minutes of slide review. Using these data, we trained PathCowork-1T for server deployment and PathCowork-27B for local deployment, with improved visual grounding. Across five organ-specific evaluations, agents combining this grounding with recorded diagnostic procedures achieved primary task scores averaging 17.9 and 24.1 percentage points higher than Claude Code and Codex, respectively, with shorter processing times and lower inference costs. Compared to the state-of-the-art multiple-instance learning AIs, the PathCowork agent-proposed regions achieved higher prostate-grading scores with limited slide labels. PathCowork can also help construct pathology AI benchmarks where questions derived from the recordings identified difficulties in recognizing viewing context and rejecting inapplicable premises. In summary, PathCowork proposes a new paradigm to scale up the data collection and train pathology AI through transparent reasoning from diagnostic sessions, opening a path towards systems that learn directly from routine clinical practice.
111. 【2610.39423】FDSML: Formula-Driven Supervised Metric Learning with Parameter-Aware Triplet Loss
链接：https://arxiv.org/abs/2610.39423
作者：Hina Otake,Shinei Arakawa,Keitaro Tanaka,Yoshihiro Fukuhara,Hirokatsu Kataoka,Shigeo Morishima
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
关键词：Supervised Metric Learning, Formula-Driven Supervised Learning, existing methods cast, Formula-Driven Supervised Metric, Supervised Learning
备注： Accepted to WACV 2027. 24 pages (11 main + 13 supplementary). Project page: [this https URL](https://shineiarakawa.github.io/fdsml/)
点击查看摘要
Abstract:Formula-Driven Supervised Learning (FDSL) enables vision pre-training without real images, but existing methods cast pre-training as classification on discretized labels, discarding continuity in the generative process. We propose Formula-Driven Supervised Metric Learning (FDSML), which derives relational supervision from image-generation parameters beyond discrete class labels. FDSML combines FineFractalDB, our refined fractal dataset with higher intra-class variation and more stable structures, with a Parameter-Aware Triplet (PA-Triplet) Loss that assigns positives and negatives according to distances in the underlying Iterated Function System parameter space. This objective trains embeddings to respect parameter-derived relative similarity orderings. Experiments show that FDSML improves transferability to image retrieval while remaining competitive on downstream classification and improving robustness under several corruption settings. On ResNet-50, Recall@1 improves from 65.6% to 74.0% on Stanford Online Products and from 56.2% to 78.4% on Stanford Cars over FDSL pre-training. These results suggest that parameter-derived relational supervision provides a stronger synthetic pre-training signal than label-based classification for transfer beyond recognition.
112. 【2610.39419】Diagnosing the Reference Gap: What Visual Scaffolds Can and Cannot Fix in Multimodal Reasoning
链接：https://arxiv.org/abs/2610.39419
作者：Guizhen Chen,Weiwen Xu,Wotao Yin,Anh Tuan Luu,Hou Pong Chan
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：recognize visual content, large language models, multiple steps, content yet struggle, struggle to reason
备注： 20 pages, 8 figures
点击查看摘要
Abstract:Multimodal large language models can recognize visual content yet struggle to reason reliably over it across multiple steps. In counting, a model may see all objects but lose track of which ones it has counted. We argue that such failures, commonly attributed to perception, reveal a reference gap: the failure to bind language to the relevant visual entities, regions, or locations. Motivated by the theory of visual routines, we test whether external references such as numbering can narrow this gap. We introduce ReGap-Bench, a paired benchmark of five visual reasoning tasks in which the scene, question, and answer are fixed and only the reference varies. Results show that task-matched references improve accuracy by up to 25\%, while mismatched references provide little benefit or even hurt performance. These findings identify the reference gap as a bottleneck distinct from visual perception and show that constructing task-appropriate visual references can serve as a training-free visual skill for improving multimodal reasoning.
113. 【2610.39416】Decoding Geometry, Not Tokens: Efficient VLMs for Robot Affordance and Trajectory Prediction
链接：https://arxiv.org/abs/2610.39416
作者：Akila Ayanthi,Darshana Priyasad,Tharindu Fernando,Dimity Miller,Sridha Sridharan,Clinton Fookes,Peyman Moghadam
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
关键词：geometry problem, move is fundamentally, fundamentally a geometry, Predicting, prediction
备注： IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2027)
点击查看摘要
Abstract:Predicting where to grasp and how to move is fundamentally a geometry problem. Vision-language models treat it as language generation instead. They tokenize spatial coordinates into text and decode them autoregressively. This mechanism ignores spatial structure and truncates continuous precision. Errors compound across coordinate tokens during decoding. In this paper, we show that compact off-the-shelf VLMs with 0.5-2B parameters already encode rich spatial information in their intermediate hidden states. Autoregressive decoding is the dominant bottleneck for trajectory prediction, while affordance prediction depends additionally on feature depth and decoder architecture. We expose the latent spatial information with a lightweight, geometry-compatible prediction head that attaches to a frozen backbone. It reads features from an intermediate hidden layer, conditions a learned query on the task instruction and refines it via cross-attention over the visual tokens to regress affordance boxes and trajectory waypoints directly as continuous values, without modifying the backbone. We evaluate on ShareRobot-Bench and AGD20K. Our 2B variant reduces trajectory error by over 60% and improves ShareRobot-Bench affordance mAP by 85% relative to the autoregressive baseline. It approaches RoboBrain on trajectory prediction (0.251 vs 0.191 DFD) at 3.8x higher throughput. These results demonstrate a favorable accuracy-efficiency trade-off for instruction-conditioned 2D geometric prediction.
114. 【2610.39414】Anatomical Prior-Guided Routing for 3D Lung Tumor Segmentation
链接：https://arxiv.org/abs/2610.39414
作者：Uihyeon Maeng,Jonghun Kim,Hyunjin Park
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Accurate lung tumor, quantitative imaging analysis, lung tumor segmentation, Accurate lung, lung tumor
备注： Accepted to ACCV 2026 (Oral), 18 pages, 6 figures
点击查看摘要
Abstract:Accurate lung tumor segmentation in computed tomography (CT) is essential for treatment planning, response assessment, and quantitative imaging analysis. However, 3D lung tumor segmentation remains challenging because tumors occupy only a small portion of the whole volume and exhibit heterogeneous appearances across anatomical locations. In patch-based training and sliding-window inference for 3D medical image segmentation, location-dependent heterogeneity is further amplified because cropped image patches provide limited global anatomical context, making it difficult for a model to adapt its feature processing according to the anatomical region from which each patch is extracted. To address this limitation, we propose an Anatomical Prior-guided Routing Module (APRM) for lung tumor segmentation. For each image patch, we obtain anatomical prior information of regional occupancy and deterministically select an anatomical routing gate. The selected gate then performs sparse feature routing over a shared expert adapter pool, enabling anatomy-specific feature adaptation while sharing expert capacity across regions. We evaluate APRM on two datasets with various backbone networks and demonstrate consistent improvements over baseline models. Furthermore, feature visualizations show that APRM produces different feature responses depending on patch location, indicating that the proposed routing mechanism effectively adapts feature processing according to anatomical context.
115. 【2610.39408】YOLOF-MLD: A Multi-Level Distillation Framework for Efficient Object Detection
链接：https://arxiv.org/abs/2610.39408
作者：Xiaolong Wu,Liming Zheng,Zitong Liu,Yue Zheng
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Efficient object detection, Efficient object, Feature Pyramid Network, resource-constrained edge devices, object detection
备注： 4 pages
点击查看摘要
Abstract:Efficient object detection on resource-constrained edge devices is essential for safety-critical applications such as fire-smoke monitoring. However, current lightweight frameworks, such as YOLO26, still rely on the parameter-dense multi-branch Feature Pyramid Network (FPN), preventing full exploitation of their lightweight advantage. Replacing FPN with YOLOF's single-input-single-output (SiSo) neck is a promising solution but causes substantial accuracy loss. To alleviate this, we propose YOLOF-MLD, a CPU-efficient multi-level distillation (MLD) framework that combines a YOLO26n backbone, a SiSo neck, and MLD. It aligns features at the backbone via sliced-semantics matching, neck with receptive-field matching, and response levels. On D-Fire and Annotated Fire-Smoke (AFS) datasets, YOLOF-MLD reduces the parameter count of YOLO26n by 35.7\% and achieves a 62.3\% speedup on Raspberry Pi 5, attaining 71.6% and 59.9% mAP@50, respectively, compared with 75.5% and 60.0% for YOLO26n. These results demonstrate a competitive accuracy--latency trade-off for edge deployment. Our code is available at this https URL.
116. 【2610.39403】SymKL: A Unified Representation-Space Measure of Signal Degradation
链接：https://arxiv.org/abs/2610.39403
作者：Dhananjaya Jayasundara,Sudarshan Rajagopalan,Vishal M. Patel
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：existing full-reference measures, Quantifying degradation, inherently modality-specific, full-reference degradation measure, Fisher information
备注：
点击查看摘要
Abstract:Quantifying degradation between a pristine signal and its degraded observation is fundamental to signal processing, yet existing full-reference measures are inherently modality-specific. We introduce SymKL, a unified full-reference degradation measure based on implicit neural representations (INRs). An INR is fitted once to the pristine reference and characterized by its diagonal Fisher information, which captures the local information geometry of the learned representation. Evaluating the Fisher information with the pristine reference and a degraded observation yields two parameter-sensitivity profiles whose symmetric KL divergence defines the degradation score. The proposed formulation requires only the reference and degraded observation, without knowledge of the corruption process or degradation severity, making it applicable across images, audio, and 3D data. We further derive an exact closed-form expression for the diagonal Fisher of SIREN, enabling efficient computation. Experiments on image, speech, and volumetric 3D benchmarks demonstrate that SymKL accurately tracks degradation severity, remains monotonic where conventional metrics saturate or fail, and provides a unified modality-agnostic measure of signal degradation. Code will be made publicly available after the review process.
117. 【2610.39399】ACD-GS: Asymmetric Curvature-aware Densification for 3D Gaussian Splatting
链接：https://arxiv.org/abs/2610.39399
作者：Houqiang Zhong,Tianchi Zhu,Qiang Hu,Xiaoyun Zhang,Li Song,Wenjun Zhang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Gaussian Splatting, view synthesis, powerful paradigm, paradigm for real-time, real-time photorealistic
备注：
点击查看摘要
Abstract:3D Gaussian Splatting (3DGS) has emerged as a powerful paradigm for real-time photorealistic novel view synthesis. However, its training-time densification remains largely heuristic, often resulting in redundant Gaussian proliferation and inaccurate geometric reconstruction. Existing methods predominantly rely on first-order screen-space gradients to trigger cloning and splitting, which fail to distinguish true geometric discontinuities from high-frequency textures or noisy residuals. To address this limitation, we propose ACD-GS, an Asymmetric Curvature-aware Densification framework for compact and geometry-aware 3D Gaussian optimization. Our key insight is that second-order depth responses provide a more reliable structural prior for density control, enabling asymmetric regulation of cloning and splitting to better capture geometric discontinuities. Building on this, we introduce a multi-view photometric consistency constraint to suppress transient single-view errors, together with a capacity-aware adaptive regulation strategy that balances densification and pruning during training. Unlike post-hoc compression or quantization-based approaches, ACD-GS directly prevents redundant Gaussian generation during training, yielding inherently compact representations. Extensive experiments demonstrate that our method reduces nearly 60% of Gaussian count and storage while maintaining competitive rendering fidelity, consistently outperforming existing methods across multiple datasets. Our code is available in this https URL.
118. 【2610.39396】CASE: Concept-Aware Segmentation from Positive and Negative Visual Exemplars
链接：https://arxiv.org/abs/2610.39396
作者：Yi-Jen Tsai,Yen-Yu Lin,Chien-Yao Wang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：single semantic class, semantic segmentation typically, segmentation typically identifies, typically identifies target, negative visual prompts
备注： Accepted to ACCV 2026
点击查看摘要
Abstract:Few-shot semantic segmentation typically identifies target objects based on positive support examples of a single semantic class. In practice, however, positive cues alone are often underspecified: for instance, "red cars" may be confused with cars of all colors, "vehicles" may include cars that should be excluded, and "material-defined objects" may span multiple categories. Fulfilling precise requests requires positive exemplars to provide inclusion evidence and negative exemplars to act as intent-defining constraints for exclusion and disambiguation. We formulate this setting as Specified Visual Prompt Segmentation (SVPS), where the target concept is specified by both positive and negative visual prompts. In this work, visual exemplars refer to support-image regions specified by bounding-box prompts. To study SVPS, we introduce CASE (Concept-Aware Segmentation from Positive and Negative Visual Exemplars), a benchmark built on COCO and PACO. CASE encompasses six category- and attribute-level task families: exclusion, isolation, union, attribute exclusion, attribute isolation, and attribute union. We evaluate SAM 3 under this setting and find that, despite its powerful positive-only few-shot segmentation capability, frozen SAM 3 does not consistently treat negative visual prompts as exclusion evidence. To address this, we present Concept Prompt Adaptation (CPA), a conservative LoRA-based adaptation baseline that improves positive-negative prompt interpretation on CASE while remaining competitive under positive-only few-shot segmentation. Code is available at this https URL.
119. 【2610.39356】Semantic Mismapping in Generalized Category Discovery: Evaluation and Mitigation
链接：https://arxiv.org/abs/2610.39356
作者：Zitai Wang,Qianqian Xu,Cong Hua,Xi Chen,Zhiyong Yang,Xiaochun Cao,Qingming Huang
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：Generalized Category Discovery, Generalized Category, Category Discovery, novel-category discovery, perform old-category recognition
备注：
点击查看摘要
Abstract:Generalized Category Discovery (GCD) aims to perform old-category recognition and novel-category discovery in the same unlabeled set. Early methods address both objectives through clustering and evaluate the exchangeable predicted indices with global Hungarian (GH) matching. Recent methods adopt trainable parametric heads, using supervision for old-category recognition and discovery-oriented objectives for new categories. In this way, the output space becomes partially exchangeable: old indices are semantically anchored, while new indices remain exchangeable. However, these methods still apply global matching, which can reassign anchored old indices to other categories and hide some prediction errors. Empirically, we are surprised to observe that this reassignment can induce an overestimation of new-category accuracy by up to 25.63 points. We term this phenomenon semantic mismapping and further derive a permutation-cycle condition that characterizes when semantic mismapping can increase the global matching score. To fix this issue, we propose Semantics-Preserving Evaluation (SPE), which preserves the supervised semantics of old indices and restricts Hungarian matching to the new region. Furthermore, we propose Mismapping-Mitigating Optimization (MMO) to alleviate the prediction errors that can give rise to semantic mismapping, combining a region-balance loss (RBL) for old/new routing with a novel-diversity loss (NDL) that encourages orthogonality among new prototypes. Theoretically, we show that minimizing the two losses can tighten an upper bound on spurious accuracy gains from semantic mismapping, with controllable residual terms. Empirical results under both GH and SPE demonstrate the effectiveness of our proposed method.
120. 【2610.39344】PicturePerfect: Diffusion-based Same-identity Facial Replacement
链接：https://arxiv.org/abs/2610.39344
作者：Manik Dhingra,Yashesh Savani,Mahesh Reddy,Antoine Mercier,Jiahe Liu,Guillaume Berger
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：computer vision, growing success, success of diffusion, explore their potential, potential for same-identity
备注： Camera-ready version (ECCV 2026 workshop - Third Workshop on Foundation and Generative Models in Biometrics)
点击查看摘要
Abstract:With the growing success of diffusion models in computer vision, we explore their potential for same-identity facial replacement in photographs. Specifically, we propose a diffusion-based method, built on top of a pre-trained text-to-image model, that takes as input a portrait image of a person and a second reference image of the same individual, potentially captured under different conditions. The goal is to seamlessly replace the input face with the reference face, while keeping the background intact. Surprisingly, despite the clear real-world utility of this task, no recently published work has directly addressed face replacement in this specific setting. To support this goal, we construct a large dataset of image pairs depicting the same person under varying facial expressions and poses. Experimental results demonstrate that our approach produces more realistic and identity-consistent results than existing face reenactment models.
121. 【2610.39331】ng Is Not Seeing: Textual Personas Rarely Make Vision-Language Models Answer Like the People They Describe
链接：https://arxiv.org/abs/2610.39331
作者：Jiawei Chen
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Vision-language models, user studies, increasingly prompted, low vision, Vision-language
备注：
点击查看摘要
Abstract:Vision-language models (VLMs) are increasingly prompted with personas, such as "You have low vision", to stand in for people in user studies. We ask whether a persona makes a model see like the person it describes. We compare persona answers with measured people with low vision, color blindness or hearing loss, and with simple rules that use only the facts the persona states. The persona never beats these rules, and it erases the real differences between groups and individuals. Describing blur leaves recognition unchanged in all 14 models we test, while blurring the photo lowers it as it does for people. Inside open models, the description changes neither the image information nor the answer probabilities. Yet the models know the effects they fail to show: they can estimate how blur and illusions affect people. Telling is not seeing: a persona makes a model say what it knows about people, not see what they see.
122. 【2610.39330】Can Image Agents Complete Visual Commissions? A Benchmark and Modular Framework
链接：https://arxiv.org/abs/2610.39330
作者：Tong Li,Lishuai Gao,Qi Zeng,Yifan Li,Cong Wei,Zhuoliang Kang,Xiaoming Wei
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：heterogeneous source materials, produce coordinated deliverables, integrate heterogeneous source, Visual commissions require, Visual Commission Bench
备注：
点击查看摘要
Abstract:Visual commissions require image agents to integrate heterogeneous source materials, preserve requirements across dependent stages, and produce coordinated deliverables. Existing evaluations examine specific capabilities and execution settings, leaving end-to-end commission completion less understood. We introduce Visual Commission Bench (VCB), comprising 600 tasks across 12 categories and three complementary tiers, from capability-focused tasks to complete commissions. VCB uses 5,039 requirement checks and separate visual quality ratings to assess content correctness, visual realization, constraint preservation, and delivery completeness without prescribing an internal workflow. We present V-Protocol, a reference agent that dynamically composes reusable task, cognitive, and expression modules at inference time. Task modules organize delivery, cognitive modules derive content, and expression modules guide visual realization. The modules exchange structured intermediate artifacts, while a shared execution harness tracks artifact revisions and dependencies, manages tool execution, and checks completion. With the same controller and image generator as the agent baselines, V-Protocol achieves 69.0\% strict task success on VCB, compared with 42.0\% for SCOPE. On Mind-Bench, strict accuracy increases from 39.2\% to 45.4\%.
123. 【2610.39324】MRC-MMAgent: Reinforcement Learning for Multi-Role Collaboration in Multimodal Tool-Using Agents
链接：https://arxiv.org/abs/2610.39324
作者：Shengqin Wang,Jie Jin,Xiaofan Li,Wentao Yan,Yu Cheng,Qingsong HuYihang Chen,Weilin Luo,Jingyu Gong,Xin Tan,Yuan Xie,Zhizhong Zhang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Multimodal tool-using agents, Multimodal tool-using, tool-using agents augment, agents augment vision-language, visually complex tasks
备注：
点击查看摘要
Abstract:Multimodal tool-using agents augment vision-language models with access to external information and capabilities, enabling them to solve knowledge-intensive and visually complex tasks that cannot be addressed using visual context alone. However, existing reinforcement learning pipelines largely train single-role agents to search and answer using terminal rewards, overlooking how humans inspect retrieved evidence, reformulate queries, and revise conclusions during search. Consequently, they fail to optimize this self-corrective process over complete interaction trajectories. We propose MRC-MMAgent, a multi-role collaborative reinforcement learning framework that incorporates trajectory assessment and continuation directly into the executable agent loop. First, the framework establishes a multi-role collaborative loop that integrates solving, inspection, and continuation, enabling the agent to acquire evidence through tools, evaluate the preceding trajectory, and resume problem solving when the current answer is insufficiently supported. Second, we introduce a role-coupled learning objective that combines task success, interaction validity, and successful correction with full-trajectory policy optimization across the decisions generated by all roles. Extensive experiments show that, using only the original baseline training set without constructing additional datasets, our method consistently outperforms the corresponding base models and achieves leading or highly competitive performance across benchmarks.
124. 【2610.39322】MemoryFusion: Cross-Temporal Memory Learning for Multimodal Video Fusion
链接：https://arxiv.org/abs/2610.39322
作者：Meiqi Gong,Hao Zhang,Jiayi Ma
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Multimodal videos, facilitate faithful fusion, temporal, facilitate faithful, Memory Bank
备注：
点击查看摘要
Abstract:Multimodal videos contain rich temporal cues that can facilitate faithful fusion. However, existing video fusion methods mainly rely on short-range temporal interactions, leaving valuable long-term information underexplored. Moreover, inconsistent fusion representations can cause flickering artifacts, undermining temporal coherence. To address these coupled challenges, we propose MemoryFusion, a two-stage framework integrating cross-temporal memory learning with progressive temporal refinement. Considering the complementary roles of adjacent and distant frames, the first stage combines a Short-Term Memory Bank for local temporal guidance with a Long-Term Memory Bank that preserves reliable historical information through a fusion-oriented update strategy. To mitigate temporal fluctuations, a Temporal Smoothing Module leverages soft motion cues for coarse background stabilization. Building upon the resulting fusion prior, The second stage further employs a lightweight 3D residual Temporal Refinement Module to correct remaining temporal inconsistencies while enhancing structural details. Extensive experiments demonstrate that MemoryFusion achieves superior fusion quality and temporal consistency over existing frame-based and video-based fusion methods. The source code is publicly available at this https URL.
125. 【2610.39308】Unlimited OCR Works 2
链接：https://arxiv.org/abs/2610.39308
作者：Shu Wei,Jingjing Wu,Qunyi Xie,Lingshu Zhang,Xiangcheng Du,Yannian Fu,Mengjie Wang,Zuyi Wang,Zhichao Liu,Yangliu Xu, YY,Yi Yang
类目：Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：Unlimited OCR, Unlimited OCR ushers, document parsing, OCR, Unlimited
备注：
点击查看摘要
Abstract:Unlimited OCR ushers the era of long-horizon parsing, substantially mitigating the growth of working memory as document lengthens. However, as document length scales to hundreds of pages, the visual context grows substantially, sharply increasing the computational cost of each decoding step and limiting the scalability of ultra-long document parsing. Inspired by human book transcription via localized focus without re-scanning preceding pages, we introduce Unlimited OCR 2, a document parsing model that looks only at what it transcribes. Building on Unlimited OCR, we upgrade all the decoder attention modules to Sparse Reference Sliding Window Attention (SR-SWA), bounding active reference tokens to a fixed capacity and thus sustaining constant computational costs across arbitrarily long documents. By combining SR-SWA with Unlimited-Horizon Decoding Paradigm, Unlimited OCR 2 enables document parsing over virtually unlimited lengths without prohibitive memory overhead. At the hundred-page scale, Unlimited OCR 2 achieves 7.0$\times$ and 5.6$\times$ per-step speedups over Full Attention and R-SWA, respectively. Furthermore, it delivers state-of-the-art end-to-end performance on both MPDocBench and MarathonDocBench, our newly proposed benchmark targeting hundred-page document parsing. Ultimately, the capacity for uninterrupted, one-shot document parsing signifies a fundamental paradigm shift into the OCR 3.0 era.
126. 【2610.39306】CAD-WAM: Anticipating Geometry for Text-to-CAD Generation
链接：https://arxiv.org/abs/2610.39306
作者：Zhengyang Li,Weijie Zheng,Yuxin Cui,Sijie Wang,Wenjie Feng,Qiyu Kang
类目：Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
关键词：Generating editable computer-aided, editable computer-aided design, text requires selecting, Generating editable, requires selecting operations
备注：
点击查看摘要
Abstract:Generating editable computer-aided design (CAD) models from text requires selecting operations that produce the intended changes to the geometry constructed so far. An incorrect operation can compromise subsequent steps, making reliable multi-step generation challenging. Existing methods based on large language models (LLMs) primarily learn to predict commands, without explicit supervision of the geometric states those commands should produce. Inspired by world-action models, we introduce CAD-WAM, which couples geometry-conditioned action generation with next-state prediction. We define complete modeling operations as actions in a CAD domain-specific language (DSL) and encode surface point clouds of intermediate solids into compact latent states using a pretrained variational autoencoder. An LLM backbone generates the next action from the target description and current geometric state, while an auxiliary head on the same backbone predicts the next state to provide geometric supervision during training. We construct step-wise training trajectories from parametric CAD models in the DeepCAD dataset, pairing each trajectory with an expert-level textual description. At inference, each generated action is executed by a CAD kernel, and the resulting solid is re-encoded to inform the next decision. On the DeepCAD test set, CAD-WAM outperforms the task-specific baselines in geometric accuracy and output validity, with larger gains on models requiring more operations, and produces more accurate geometry than general-purpose LLMs prompted with the same DSL. Ablations further show that next-state supervision reduces mean Chamfer distance by 13.2\% relative to an otherwise identical model trained only to predict actions. Code is available at this https URL.
127. 【2610.39302】Proving and Improving the Platonic Representation Hypothesis with JEPAs
链接：https://arxiv.org/abs/2610.39302
作者：Randall Balestriero,Yann LeCun
类目：Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML)
关键词：Platonic Representation Hypothesis, Platonic Representation Theorem, world is central, Platonic Representation, platonic representations
备注：
点击查看摘要
Abstract:Whether models trained on different modalities converge to the same representation of the world is central to AI. The Platonic Representation Hypothesis (PRH) conjectures that they do as they scale, but without a proof it offers neither a guarantee nor a recipe. We turn the hypothesis into a theorem. First, we prove a \emph{Platonic Representation Theorem}: Joint-Embedding Predictive Architectures (JEPAs) trained separately on any number of modalities learn the same representation up to one rotation whenever their positive pairs are two observations of one scene--\emph{good views}--and their misalignment is bounded otherwise. Second, we show that next-token prediction is a JEPA with \emph{bad views}, parts of a single document, which explains why language models align with vision only through scale. Third, we validate the theory from 145k captions to the texts of 940M web images, and uncover a scaling law of alignment that predicts unseen runs within seed noise. Good views need 42 times fewer training samples than next-token prediction on the same corpus to reach the same alignment, and a 109M-parameter text JEPA aligns with DINOv2 and CLIP's image encoders better than every language model we evaluated, up to 72B parameters, and than CLIP's text encoder: next-token prediction is a remarkably inefficient route to platonic representations. As a showcase, one rotation turns separately trained encoders into an image captioner (CIDEr .838 on COCO). Even that rotation needs no pair: on the latest pair-free benchmark, our text JEPA beats every zero-shot accuracy reported with language models up to 8B parameters, e.g., 64\% against at most 56\% top-5 on ImageNet-100. Positive views, and not scale alone, are thus the lever to build platonic representations.
128. 【2610.39301】Youtu-Parsing-Omni: One Encoder, One Schema, All Modalities
链接：https://arxiv.org/abs/2610.39301
作者：Yunhang Shen,Zhangquan Chen,Xiaotian Li,Zhongpeng Cai,Qing Zhou,Zekai Zhang,Kun Zhang,Zezhong Wu,Jian Hu,Haihan Gao,Heting Gao,Xin Li,Yinsong Liu,Deqian Jiang,Di Yin,Haoyu Cao,Xing Sun,Bing Liu,Yanqiu Qu,Yunfei Wu,Kun Yin,Zhixiang Wei,Yongxiang Hua,Qianyu Li,Xiaoxing Hu,Peiyuan Zhang,Lijiang Li,Dong Xu,Zhenchuan Yang,Yiliang Jiang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：structured records grounded, converts documents, space and time, agents that read, structured records
备注：
点击查看摘要
Abstract:We present Youtu-Parsing-Omni, an omni-modal parser that converts documents, images, audio, and video into structured records grounded in space and time, providing a perception layer for agents that read, locate, and cite evidence across modalities. Its design rests on one encoder for every modality and one schema for every task. Youtu-Omni-Encoder, initialized from a pre-trained language model to strengthen text reading, encodes all modalities with shared weights in a unified space-time coordinate system, and a few fusion layers let video frames attend to concurrent audio, bringing cross-modal fusion and temporal alignment into perception itself. OmniSchema unifies the outputs of all ten parsing tasks, allowing a single self-evolving post-training procedure, OmniSchema-Aware On-Policy Distillation (OSAD), to provide a shared token-role-based objective across tasks. In each round, the model is trained on its own rollouts against token-level targets from a reference-conditioned copy of itself, which is then refreshed with the trained weights, so that the parser and its supervision co-evolve. With a single 5 B-parameter checkpoint, Youtu-Parsing-Omni achieves state-of-the-art performance on OmniDocBench v1.6 (96.96 Overall) and leads all open-weight models on OmniParsingBench (75.08 Avg.), with its advantage concentrated on audio and video. It also achieves the best overall chart-parsing score (95.02) and the highest exact chemical-fingerprint agreement among compared systems, and matches a dedicated optical-music model in character-level transcription of typeset sheet music while staying far ahead of general-purpose systems. Code and model weights are publicly available.
129. 【2610.39293】Unlocking Any-Order Generation in Pretrained Autoregressive Image Models
链接：https://arxiv.org/abs/2610.39293
作者：Rishav Pramanik,Marco Pedersoli,Zhaozheng Yin
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：fixed raster-scan order, raster-scan order hardcodes, inflexible in practice, unable to inpaint, powerful in principle
备注： Accepted in NeurIPS 2026, Sydney as a poster
点击查看摘要
Abstract:Autoregressive (AR) image generators are powerful in principle but inflexible in practice: training with a fixed raster-scan order hardcodes a single factorization of the image distribution into the model, leaving it unable to inpaint, outpaint, or edit at inference time. Existing solutions, including the modern discrete-diffusion family, train from scratch on a mixture of token orderings, paying roughly 3x the compute of standard raster-scan training; discrete diffusion further sacrifices compatibility with the AR inference ecosystem (KV caching, vLLM) by construction. We argue that this cost is unnecessary. A pretrained raster-scan AR model has already learned the joint distribution over image tokens; what it lacks is not knowledge but the ability to express that knowledge through non-sequential conditional pathways. We propose a recipe for adapting pretrained raster-scan models into any-order generators using two components: future-aware positional embeddings that resolve a target-position identifiability problem at negligible overhead, and an ordering curriculum that traces a stable optimization path from the pretrained solution to a good any-order basin. Applied to four base models across two architectures and two scales (LlamaGen-L/XL, RAR-L/XL), the recipe substantially outperforms MaskGIT on ImageNet inpainting (FID 6.21 vs. 10.44) and outpainting, is competitive with MAR-L, trades a 1.2-1.9 increase in class-conditional FID for this flexibility, and costs 13-33% additional compute beyond pretraining, while leaving the standard AR inference stack untouched. Inpainting, outpainting, and editing, our results suggest, need not be added to autoregressive image models. They can be unlocked from the ones that exist. Code is available at this https URL.
130. 【2610.39283】PriorMaskMap: Robust Online Vectorized Map Construction with Biased Priors
链接：https://arxiv.org/abs/2610.39283
作者：Haoming Xu,Wei Li,Yu Hu
类目：Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
关键词：geometric scene understanding, Online vectorized map, Online vectorized, vehicles with essential, real-time semantic
备注： Accept by ECCV2026
点击查看摘要
Abstract:Online vectorized map construction provides autonomous vehicles with essential, real-time semantic and geometric scene understanding. While leveraging prior maps presents an opportunity to improve accuracy, they are often biased due to outdated information or temporary road changes. Existing methods, which typically treat priors as fully reliable or only consider limited noise patterns, suffer severe performance degradation when these priors are imperfect. To address this, we propose PriorMaskMap, a robust framework for prior-guided vectorized map construction. Its first innovation is a Prior-Aware Confidence Estimator (PACE) that evaluates the input rasterized map against current visual observations. It assigns confidence scores to regions classified as reliable, erroneous, or blank, creating a foundational mask that guides two subsequent branches. In one branch, the Dual-Prior Query Generator (DPQG) selectively filters and embeds both rasterized and vectorized priors to initialize informative queries from trustworthy map structures. In the other, the Confidence-Guided Prior Fusion (CGPF) module spatially propagates and refines the mask, using the refined version to govern the fusion of BEV features with the prior map, thereby providing rich contextual representations for the map decoder. Evaluation on the nuScenes dataset shows that PriorMaskMap achieves superior mapping precision over existing approaches and exhibits marked robustness when provided with priors containing apparent errors. Our code is released at this https URL.
131. 【2610.39278】WorldFlow: Joint Video and 3D Motion Prediction via Large-Scale Pretraining
链接：https://arxiv.org/abs/2610.39278
作者：Yuhao Lu,Jiongxu Chen,Hexiang Wei,Jiajin Wu,Juelin Li,Mingjian Liang,Fan Yang,Zhuowei Liang,Zihao Liu,Zhennan Wang,Sheng Wang,Lin Shao
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
关键词：Predicting future physical, physical states stands, Predicting future, embodied intelligence, future physical states
备注：
点击查看摘要
Abstract:Predicting future physical states stands as a central challenge in embodied intelligence. Existing video world models leave 3D dynamics implicit, action-based formulations remain tied to specific robot configurations, and 3D motion prediction models use flow-only supervision, providing a limited learning signal for physical dynamics. To circumvent these limitations, we introduce \method{}, a generative world model that jointly predicts future video and query-based 3D trajectories. By parameterizing 3D motion as continuous displacement tokens aligned with the video grid, a unified Mixture-of-Transformers (MoT) backbone jointly denoises both modalities under a rectified-flow objective. This formulation couples explicit geometric dynamics with visual evolution, effectively unlocking the physical commonsense embedded within pretrained video models. To empirically scale and validate this unified architecture, we train \method{} on a heterogeneous mixture of 1.44 million object-level samples spanning simulated and real-world domains. We further introduce WorldFlowBench, comprising 1,678 evaluation samples from 11 data sources across simulated and real-world interactions, including spatial reasoning tasks. Across all three evaluation sets, WorldFlow achieves the lowest average displacement error (ADE) and final displacement error (FDE) among the evaluated learning-based methods. Notably, on the challenging SpatialCog set, WorldFlow achieves an ADE of 0.060 m, a 54.2\% reduction compared with the strongest external baseline. It also reduces ADE and FDE by 55.9\% and 68.5\%, respectively, compared with its flow-only variant. Project page: this https URL.
132. 【2610.39277】Efficient Covariance-Aware Vicinal Latent Diffusion for Continuous Conditional Image Generation
链接：https://arxiv.org/abs/2610.39277
作者：Xin Ding,Shuo Huang,Zhenglong Ding,Sen Zhang,Nenglun Chen,Peibei Cao,Yongwei Wang,Fei Wu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Conditional Generative Modeling, Generative Modeling, continuously valued scalar, valued scalar labels, Continuous Conditional Generative
备注： 14 pages
点击查看摘要
Abstract:Continuous Conditional Generative Modeling (CCGM) aims to synthesize high-dimensional data conditioned on continuously valued scalar labels. However, existing continuous diffusion models face severe data sparsity under imbalanced and long-tailed label distributions, which can lead to visual artifacts and condition inconsistency. Moreover, inference-time classifier-free guidance and dense pixel-space integration introduce substantial sampling overhead, limiting practical deployment. To address these challenges, we propose CAViLD, an efficient latent diffusion framework for continuous conditional generation. CAViLD first constructs a purified condition-aware latent space through density-adaptive vicinal learning, auxiliary semantic regularization, and NIQE-guided negative data augmentation. It then internalizes the conditional guidance direction through an EMA teacher, enabling single-branch conditional inference, and introduces covariance-aware progressive distillation for label-dependent probability-flow ODEs, supporting sampling down to four NFEs. Experiments on RC-49, UTKFace, and Steering Angle demonstrate a favorable trade-off among generation fidelity, continuous condition consistency, and sampling efficiency.
133. 【2610.39265】DeCoRL: Decomposed Consistency Reinforcement Learning for Multi-Image Composition
链接：https://arxiv.org/abs/2610.39265
作者：Zhiqiang Wu,Shuang Sun,Jiale Zhang,Jing Li,Yibo Chen,Xu Tang,Yao Hu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：large-scale image editing, consistency degrades rapidly, Multi-image composition, reference images increases, critical yet challenging
备注： NeurIPS
点击查看摘要
Abstract:Multi-image composition (MICo) is a critical yet challenging capability for large-scale image editing models, where consistency degrades rapidly as the number of reference images increases. While supervised fine-tuning (SFT) can equip models with basic MICo ability, it provides only implicit supervision on the composed image and often fails to learn reliable object-level correspondence across multiple references. We propose DeCoRL, a reinforcement-learning (RL) framework that improves MICo consistency by decomposing holistic MICo consistency into object-level consistency. DeCoRL first establishes instruction-relevant object correspondences across reference images and localizes the corresponding regions in the generated image. It then uses a VLM-based reward model trained on single-object consistency annotations to produce fine-grained, object-wise feedback, which is aggregated into a multi-object consistency reward to guide RL post-training. We further introduce DeCo-Bench, a benchmark covering compositions from 2 to 6 reference images. Experiments on DeCo-Bench and the public MICo-Bench demonstrate consistent improvements in MICo consistency.
134. 【2610.39232】Seeing and Imagining as One: Unified Video Generation and 4D Perception for Embodied World Modeling
链接：https://arxiv.org/abs/2610.39232
作者：Haitang Feng,Zhuoman Liu,Zhennan Wang,Sheng Wang,Juelin Li,Chengzhi Zhao,Hexiang Wei,Lin Shao
类目：Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
关键词：imagine plausible futures, Embodied world, Embodied world models, embodied world model, unified embodied world
备注： Project page: [this https URL](https://gensight-rbs.github.io/)
点击查看摘要
Abstract:Embodied world models should not only imagine plausible futures, but also perceive the geometric, semantic, and motion structure of the evolving world. We present GenSight, a unified embodied world model that brings future video generation and dense perception into a single pretrained video diffusion transformer. GenSight uses a unified spatiotemporal representation for video, depth, embodiment segmentation, and first-frame-relative 2D flow. Built on a shared DiT backbone, GenSight uses lightweight task modulation to predict all modalities within the same architecture. All tasks are jointly optimized within a unified velocity prediction space. At inference, GenSight supports two complementary modes: multi-step future video generation with aligned dense predictions and single-step dense perception from an observed video. By unifying modality representation, model architecture, and prediction space, GenSight avoids separate modality-specific pipelines and couples appearance, geometry, semantics, and motion in a shared representation. It promotes cross-modal consistency and allows dense perception to benefit from the spatiotemporal priors learned through video generation. Trained on embodied manipulation data spanning multiple robot embodiments and human-hand interactions, GenSight demonstrates strong performance in both video generation and dense 4D perception, highlighting the effectiveness of unifying seeing and imagining for embodied world modeling.
135. 【2610.39214】Fast rigid rotational alignment of voxelized 3-D density maps in sliced Wasserstein distance
链接：https://arxiv.org/abs/2610.39214
作者：Oscar Mickelin,Yunpeng Shi,Amit Singer,Eric J. Verbeke
类目：Numerical Analysis (math.NA); Computer Vision and Pattern Recognition (cs.CV)
关键词：sliced optimal transport, optimal transport, voxelized three-dimensional density, three-dimensional density functions, sliced optimal
备注：
点击查看摘要
Abstract:We develop a fast method for the rigid alignment of voxelized three-dimensional density functions over $\mathsf{SO}(3)$ using sliced optimal transport. Our approach efficiently transforms each density map into a tensor of one-dimensional quantile functions, taken from central slices over the unit sphere, such that the Euclidean distance of the transformed volumes approximates the sliced 2-Wasserstein distance. Our formulation is therefore compatible with existing, fast, voxel-based Euclidean alignment algorithms. We provide theoretical analysis and numerical experiments demonstrating that the sliced 2-Wasserstein distance is a robust metric for comparing three-dimensional shapes and for computing their rigid alignment.
136. 【2610.39201】COGS: A Diagnostic Benchmark for Attribute-Aware Object-Level 3D Gaussian Splatting Perception
链接：https://arxiv.org/abs/2610.39201
作者：Zheng Xiao,Yifan Liu,Weijian Deng,Jiashu Li,Ziqin Liu,Zhenjun Han,Jianbin Jiao
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Gaussian Splatting, increasingly explored, input representation, Gaussian, Splatting
备注： Accepted at WACV 2027, Datasets and Benchmarks Track. 20 pages, including references and supplementary material
点击查看摘要
Abstract:3D Gaussian Splatting (3DGS) is increasingly explored as an input representation for 3D perception. Yet it remains unclear how current 3D architectures exploit heterogeneous Gaussian attributes. This ambiguity makes accuracy hard to interpret and limits principled architecture design for 3DGS perception. However, existing object-level 3DGS datasets lack the appearance variation or artifact control needed for systematic attribute-level analysis. We introduce COGS, a large-scale 3DGS dataset with 15,675 artifact-filtered and appearance-rich objects across 40 categories. Built on COGS, we establish COGS-Bench for object classification and attribute-usage diagnosis, complemented by external validation. Our benchmark shows that Gaussian attributes are not uniformly beneficial: selected subsets can outperform the full attribute set, and the effects of appearance, opacity, and covariance-related attribute combinations vary substantially across architectures. We further study this finding with HALO, a reference model that preserves attribute-specific evidence before hierarchical aggregation. Experiments on COGS-Bench and external 3DGS datasets demonstrate that COGS supports systematic diagnosis of 3DGS attribute usage, while HALO highlights the value of role-preserving attribute organization. Our dataset is available at this https URL.
137. 【2610.39183】Multi-level Constrained Learning for Underwater Image Enhancement
链接：https://arxiv.org/abs/2610.39183
作者：Wangtao Bao,Liping Xie,Kanjian Zhang,Huimin Lu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Underwater image enhancement, image enhancement aims, recover detail degraded, wavelength-dependent attenuation, Underwater image
备注：
点击查看摘要
Abstract:Underwater image enhancement aims to correct color and recover detail degraded by wavelength-dependent attenuation, scattering, and noise. Paired-reference supervision constrains output similarity, but does not by itself specify the color mapping, the strength of local detail corrections, or consistency with the degraded observation. We propose MCL-UIE, a paired-supervised framework that addresses these requirements by constraining restoration operations and the training objective. Monotonic Chromatic Transform (MCT) restricts global color mapping in log-LMS space through positive diagonal entries and bounded cross-channel coefficients. Uncertainty-Gated Wavelet Reconstruction (UGWR) modulates high-frequency residuals with a learned confidence gate calibrated against normalized wavelet reconstruction errors. During training, Reference-Anchored Degradation Replay (RADR) renders the enhanced image and its paired reference using shared latent degradation proxies, encouraging both to reproduce the input. The three components share a restoration network that combines structured color correction, locally controlled detail recovery, and reference-anchored replay. MCL-UIE ranks first in thirteen of sixteen full-reference dataset-metric comparisons and produces competitive visual results on no-reference benchmarks. Objective ablations indicate that confidence calibration is most effective when combined with replay supervision. RADR is used only during training and adds no inference-time computation. Our code is available at this https URL.
138. 【2610.39180】EmbodiExp: Compositional Experience Transfer for Embodied Agents
链接：https://arxiv.org/abs/2610.39180
作者：Zhifeng Gu,Ying Chen,Rui Jiang,Lei Yi,Bing Wang
类目：Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：previously effective strategies, agents turn high-level, turn high-level plans, tasks remains challenging, embodied agents turn
备注：
点击查看摘要
Abstract:Experience can help embodied agents turn high-level plans into grounded actions, but transferring it across tasks remains challenging: previously effective strategies may be inapplicable to the current state or fail to produce their intended outcomes. We present EmbodiExp, a framework for constructing transferable experiences and composing them online during embodied task execution. Our key observation is that tasks often share subgoals whose completion can be assessed independently of overall task success. We therefore introduce an outcome-grounded subgoal experience that specifies a subgoal, applicability conditions, an adaptable policy, and a verifiable outcome. During execution, applicable experiences guide action generation, while verified outcomes can establish the conditions required by subsequent experiences, enabling online composition. Subgoal-level outcome attribution further enables failure utilization: failed trajectories can still contribute verified local successes and condition-specific guidance on actions to avoid. Experiments on multiple embodied benchmarks under a fold-based protocol with frozen experience banks demonstrate the effectiveness of EmbodiExp.
139. 【2610.39179】Edited Endpoints as Partial Goals for Image-to-Video Generation
链接：https://arxiv.org/abs/2610.39179
作者：Sihyun Kim,Sanghyun Jo,Kyungsu Kim
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：source image layout, endpoint images make, image layout, improve transition completion, Edited endpoint images
备注： 34 pages, 8 figures, 28 tables. Project page: [this https URL](https://wannagetacat.github.io/Edited-Endpoints-I2V/)
点击查看摘要
Abstract:Edited endpoint images make the target of an image-to-video transition explicit and can improve transition completion, but also preserve context in the source image's layout. Full-frame endpoint conditioning turns this preserved context into requirements on the video's final state, potentially inducing temporal artifacts. We measure discontinuity with Hard-Cut (HC), late background reversal with Background Motion Reversal (BMR), and motion suppression with Dynamic Degree (DD). These artifacts appear across five video models spanning training-free and learned endpoint conditioning, in model-specific forms. We study Partial, a selective conditioning scheme that constrains only an edit-relevant region estimated from image-editor attention and leaves the remaining terminal content to the video model. Across three backbones on TC-Bench-I2V, Partial reduces temporal artifacts relative to full-frame conditioning, with no significant difference in Transition Completion Ratio (TCR); on Wan2.2-Turbo, uniformly weakening conditioning does not reproduce these reductions. A donor-context intervention keeps full-frame conditioning and replaces only the context outside the edit-relevant region with the final frame of a Partial video. It lowers HC and BMR on HunyuanVideo-1.5 and lowers BMR while restoring DD on Wan2.2-Turbo, with no significant difference in TCR, so constraint extent alone does not account for these artifacts. Together, these findings suggest that edited endpoints should serve as partial future goals, although edit relevance does not guarantee future validity. Code and data will be released at this https URL.
140. 【2610.39167】EgoContact: Evaluating Physical Intelligence with Egocentric Video
链接：https://arxiv.org/abs/2610.39167
作者：Adam Deryło,Alexey Gavryushin,Daniel Cremers,Marc Pollefeys,Xi Wang
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
关键词：learning generalizable manipulation, generalizable manipulation priors, video has emerged, scalable sources, sources of data
备注： Project page: [this https URL](https://egocontact.github.io)
点击查看摘要
Abstract:Egocentric video has emerged as one of the most scalable sources of data for learning generalizable manipulation priors. Yet, despite its growing use in pretraining robotic policies, it has been underutilized as a validation tool for physical intelligence, a key component of which is understanding where to make contact with an object to achieve an intended interaction. We present EgoContact, a large-scale egocentric dataset for evaluating the generalizability and precision of learned manipulation priors through future contact-point prediction. EgoContact contains 22.5K manually verified interaction events, each with a pre-contact observation, semantic intent, and pixel-precise contact locations. We find that current methods often struggle to predict precise interaction contacts and sometimes fail to localize the target object in diverse, in-the-wild scenes, even when allowed multiple predictions. Furthermore, we demonstrate the usefulness of EgoContact's annotated interactions for robotic manipulation: simply using EgoContact as the memory bank of an existing affordance-transfer policy improves grasp success from 14.2% to 31.2% in simulation and from 16% to 58% on a real robot.
141. 【2610.39162】A Novel Approach to Evaluating Evaluation Metrics for Multi-Output Structured Prediction
链接：https://arxiv.org/abs/2610.39162
作者：Akshay Vyas,Angelo Pimienta,Nicholas Ruozzi
类目：Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
关键词：multiple equally valid, multi-output structured prediction, ground truth label, equally valid outputs, making reliable evaluation
备注： Extended version of the AAAI 2026 paper, including the technical appendix. 18 pages, 8 figures. Code: [this https URL](https://github.com/akshay25vyas/invariants-metric-eval)
点击查看摘要
Abstract:In multi-output structured prediction tasks, while only one ground truth label may be provided in the training data, multiple equally valid outputs may be possible, making reliable evaluation a persistent challenge. We postulate that human evaluators implicitly use task-specific invariants, e.g., object boundaries in colorized images or named entities in translations, to judge if an output is acceptable. Under this assumption, we introduce a notion of approximate task-specific invariants and use them as diagnostic tools to evaluate a variety of existing metrics for vision and language tasks. We use these task invariants as part of a framework to systematically test metric reliability by encouraging domain-relevant invariants in model outputs via an augmented loss function. In our experiments, we observe that enforcing invariants with an augmented loss yields substantial improvements in popular distributional metrics while more traditional metrics change only marginally. Through this invariants-driven evaluation, we expose where standard metrics fail to detect meaningful differences, and we highlight the conditions under which distributional metrics succeed or still fall short.
142. 【2610.39161】Joint Rigid Alignment and Morphable Reconstruction from Partial 3D Face Scans
链接：https://arxiv.org/abs/2610.39161
作者：Eli A. Underwood,Nicolas Rondan,Angel V. Peterchev,J. Matias Di Martino
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：canonical representation requires, representation requires solving, Aligning a partial, common frame, canonical representation
备注： 9 pages, 4 figures, 4 tables
点击查看摘要
Abstract:Aligning a partial, real-world 3D face scan to a canonical representation requires solving two coupled problems at once: estimating the rigid pose that brings the scan into a common frame, and completing the facial shape in regions the scan never observed. Existing pipelines typically decouple these steps, running a generic point-cloud registration front end followed by a separate non-rigid fitting or deformation stage. Building on the GeoTransformer superpoint-matching architecture, we propose a single network that predicts both a rigid pose and a shape-complete point cloud in one forward pass: a cross-attention regressor, conditioned on transformer point features and predicted coarse correspondences, maps a partial scan to per-patch coefficients of a 3D morphable face model learned offline from a corpus of 8,000 identities. The resulting morphed template is then used, in the same forward pass, as the target for fine-grained Sinkhorn matching and rigid pose estimation, supervised purely by geometric alignment to ground-truth morphed geometry, so no hand-labeled coefficients are required. Against six classical and learned non-rigid registration baselines on held-out identities, our method leads the strongest ones by 3-4 times on per-vertex error and baselines that collapse the head onto the visible region by more than an order of magnitude, and is simultaneously the most accurate and fastest method on ground-truth-free scans from two independent real-scan datasets, NoW and NPHM. Code will be made publicly available.
143. 【2610.39160】CLaW: Codec-Guided Adaptive Latent Watermarking for Traceable Diffusion Image Generation
链接：https://arxiv.org/abs/2610.39160
作者：Huangchao Song,Lingyun Yu,Cong Li,Peiqi Jiang,Mingqi Fang,Hongtao Xie
类目：Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR)
关键词：increasingly realistic AI-generated, realistic AI-generated content, increasingly realistic, copyright infringement, rapid advancement
备注： Accepted at NeurIPS 2026
点击查看摘要
Abstract:With the rapid advancement of text-to-image diffusion models, increasingly realistic AI-generated content has raised serious concerns about misuse and copyright infringement. Digital watermarking offers a promising solution by enabling user-level traceability, yet existing methods remain difficult to deploy at scale, hindered by costly model retraining or detection, poor effectiveness-fidelity trade-offs, and limited robustness to image transformations. To address these challenges, we propose CLaW (Codec-guided Latent Watermarking), a robust and efficient watermarking framework for traceable diffusion image generation with frozen diffusion backbones and inversion-free detection. Specifically, CLaW first uses a pretrained watermark codec to map each watermark message to a latent residual, enabling low-cost sampling-time injection. Then, the residual is injected within a late denoising window, where image semantics are better preserved and the watermark signal is less perturbed by subsequent denoising updates, yielding a better balance between watermark effectiveness and image fidelity. Furthermore, we introduce a decoder-guided adaptive injection mechanism that uses decoding confidence as feedback to dynamically calibrate watermark strength, reinforcing weak watermark signals and improving robustness under image transformations. Extensive experiments show that CLaW preserves high visual fidelity and achieves state-of-the-art robustness under image transformations, enabling reliable user-level traceability.
144. 【2610.39158】VEvo-Harness: Causally-Guarded Self-Evolving Harness for Agentic Image Generation
链接：https://arxiv.org/abs/2610.39158
作者：Yuan Wang,Yuhang Yu,Mengting Chen,Tianrun Xu,Zihao Pan,Xuetao Feng,Wei Li
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：achieving pixel-level fidelity, satisfying densely constrained, densely constrained user, constrained user specifications, increasingly agentic workflows
备注：
点击查看摘要
Abstract:The central bottleneck in visual generation and editing has migrated from achieving pixel-level fidelity to satisfying densely constrained user specifications through increasingly agentic workflows. Prevailing agentic generation pipelines remain shallow generate-verify-rewrite loops treating VLM judgments as coarse visual feedback, either discarding repair experience or admitting ungated updates that poison skill library. With memorize and govern periods left open, they degrade into ungoverned iterators of repeated trial-and-error rather than continual self-evolving harness. Towards this issue, we present \textbf{VEvo-Harness}, the first causally-guarded continual self-evolving harness system for agentic visual generation. VEvo-Harness introduces four coupled innovations: (1) \textit{a visual traceback} that densifies sparse judgments into root-cause attributions with a hierarchical verification tree; (2) \textit{a visual continual governance} that admits only causally valid skill patches through counterfactual dual-arm image generation replay; (3) \textit{a lifecycle-managed harness state} that hierarchically distills root-cause-anchored episodes across working, experience, and pattern tiers into a persistent skill library of routable and causally verified capabilities; (4) \textit{a heterogeneous sandbox policy} that orchestrates a diverse action space of repair primitives, dynamically routing cost-aware interventions based on root causes to select the most effective strategy. VEvo-Harness surpasses state-of-the-art agentic methods and rivals proprietary generators (\textit{e.g.}, GPT-Image, Nano-Pro) while maintaining the continual harness evolution and generalization.
145. 【2610.39151】STRATA: Reinforcement-driven Active Sampling with Tri-paradigm Adaptation for Efficient 3D Spatial Transcriptomics
链接：https://arxiv.org/abs/2610.39151
作者：Junchao Zhu,Ruining Deng,Chongyu Qu,Zhengyi Lu,Yanfan Zhu,Yuechen Yang,Yu Wang,Shilin Zhao,Haichun Yang,Yuankai Huo
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：enables joint characterization, enables joint, gene expression, Spatial transcriptomics, joint characterization
备注：
点击查看摘要
Abstract:Spatial transcriptomics (ST) enables joint characterization of tissue morphology and gene expression. However, 3D volumetric ST acquisition remains expensive. AI-driven methods offer a low-cost alternative but remain limited in accuracy, while recent 2D section-wise ST acquisition still rely on dense grid sampling, resulting in redundant measurements and undersampling of biologically informative regions. To bridge the gap, we propose STRATA, a reinforcement-driven active sampling framework with tri-paradigm adaptation for reconstructing dense 3D transcriptomic maps from sparse voxel-wise ST acquisition. STRATA constructs a 3D graph integrating HE morphology with single-cell priors and employs a reinforcement learning agent to estimate morphological sparsity and transcriptomic informativeness, enabling budget-aware selection of informative spots. Given these sparse measurements, an adaptive tri-paradigm predictor combines retrieval, regression, and cross-layer imputation through lightweight fusion, dynamically weighting their contributions. Experiments across datasets demonstrate improved sampling efficiency and accuracy, particularly under low-budget settings. Code is available at this https URL.
146. 【2610.39147】Kinematic Causal Attribution: Black-Box Explanations of Motion versus Geometry for 4D Point-Cloud Video Models
链接：https://arxiv.org/abs/2610.39147
作者：Zerui Wang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：body part moves, Kinematic Causal Attribution, assistive monitoring, part moves, practitioner cannot check
备注： 27 pages, 6 figures, 14 tables
点击查看摘要
Abstract:4D perception models, which classify 3D point-cloud videos, are being built into robots, assistive monitoring and AR, yet before deploying one a practitioner cannot check which cue it relies on: how a body part moves, or what shape it has. 3D attribution is single-frame, video attribution is confined to RGB pixels, gradient maps need white-box access, attention is absent from state-space backbones, and point-level perturbations assume a point correspondence across frames that real depth and LiDAR data lack. We present Kinematic Causal Attribution (KCA), a black-box, interventional attribution method that queries only the model's outputs. Its freeze operator removes a region's motion while keeping its shape, and a shape-free proxy removes the shape while keeping the motion. Together they span a motion$\times$geometry cube for every tracked region, whose order-symmetric Shapley decomposition is complete and whose interaction is measured rather than assumed away. The operators rewrite per-frame point sets, so they need no correspondences, and a context-averaged estimator handles structures that span several regions. We validate KCA on the official PSTNet, P4Transformer and Mamba4D trained on MSR-Action3D. The decisive test is an audit: two binary tasks are built from real clips whose classes differ only in the motion or only in the shape of the same arm, and the motion share of the arm's importance follows the defining cue on all three backbones (0.83 vs. 0.60 on P4Transformer, 0.86 vs. 0.30 on PSTNet, 0.92 vs. 0.61 on Mamba4D), a distinction no single attribution map can make. The split costs no faithfulness: the necessity maps have the lowest deletion AUC on P4Transformer and PSTNet, tie with the best baseline on Mamba4D, and lose their correlation with the trained model's once the weights are randomised. Skeleton ground truth, re-sampling stability and cost complete the validation.
147. 【2610.39038】LightGenBench: A Benchmark for 3D Emission Generation
链接：https://arxiv.org/abs/2610.39038
作者：Dongchen Yang,Xingguang Yan,Manolis Savva
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Recent generation methods, rich textures, Recent generation, Recent, generation
备注： Project page: [this https URL](https://3dlg-hcvc.github.io/lightgenbench/)
点击查看摘要
Abstract:Recent generation methods produce 3D assets with detailed geometry and rich textures. However, these assets are non-emissive: they reflect light but emit none. We study a new task, emission generation: predicting spatially varying RGB emission over a 3D asset's surface from its geometry, non-emissive materials, and a rendered view. To support this task, we introduce LightGenBench. Our benchmark makes three contributions. First, we curate a dataset of 37k artist-authored emissive assets in UV, voxel, and multi-view form. Second, we establish baselines across four method families: UV-domain, sparse-voxel, multi-view, and segmentation-based generation. Third, we design an evaluation protocol that measures where an asset emits and how, complemented by user-study comparisons on renderings with varying emission strength. Across the four families, the sparse-voxel baseline TRELLIS.2-Emission performs best because generating emission natively in 3D avoids the failure modes of the other representations, such as the fracturing artifacts of UV space and the occluded emitters that multi-view generation cannot see. However, TRELLIS.2-Emission only reaches an IoU of 0.271, leaving the task far from solved. We hope LightGenBench establishes emission generation as a benchmarkable problem for 3D generation.
Comments:
Project page: this https URL
Subjects:
Computer Vision and Pattern Recognition (cs.CV)
Cite as:
arXiv:2610.39038 [cs.CV]
(or
arXiv:2610.39038v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.39038
Focus to learn more
arXiv-issued DOI via DataCite</p>
148. 【2610.39014】he Search for Statistical Indistinguishability Between Real and Fake Images
链接：https://arxiv.org/abs/2610.39014
作者：Mohaimin Rafi,Otavio Moraes Rosa,Anshuman Chhabra,Utkarsh Ojha
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：strikingly similar, real, generative models, Images, Images produced
备注：
点击查看摘要
Abstract:Images produced by generative models appear strikingly similar to real ones. In this work, we study whether perfection has been reached by generated content in terms of matching the properties of real data. We train deep neural networks to separate real data from fake and observe their test accuracies. Only a chance performance has any scope of proving statistical indistinguishability. However, since generative models are known to sometimes introduce small errors, e.g., low level checkerboard artifacts, we create progressive abstractions of the original real/fake images to potentially bring them to a space where those artifacts get wiped away. These include operations such as blurring, downsizing, cropping, converting RGB images into edge/semantic maps, etc. To our surprise, almost none of the abstractions are able to make fake data indistinguishable from real. The only cases when neural network based classifiers fail to separate the two categories is when the data is abstracted to such an extent that the original content seems mostly lost to the human eye, e.g., downsizing a $256 \times 256$ RGB image of human face to $4 \times 4$ resolution. We observe this trend across four different datasets of varying complexity (MNIST, CelebA, FFHQ and GenImage) and across various types of generative models (GANs, VAEs, diffusion models). This shows that the nature of fake artifacts extends all the way from low level to very high level in images. The simple implication is that fake images are much more different from real than we are used to think.
149. 【2610.39008】Fed-IGTD: A federated learning-based framework for tabular-to-image encoding
链接：https://arxiv.org/abs/2610.39008
作者：David Chushig-Muzo,Eva Milara,Francisco J. Lara Abelenda,Luis Carlos Estrada-Petrocelli,Diego H. Peluffo-Ordóñez
类目：Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
关键词：shown promising results, transform tabular data, image-like representations, transform tabular, shown promising
备注：
点击查看摘要
Abstract:Tabular-to-image encoding methods, which transform tabular data into image-like representations, have shown promising results in predictive tasks. Despite their benefits, these methods follow a centralized paradigm, in which the data are collected and models are trained on a central device. This raises privacy and security concerns, particularly the risk of data leakage, unauthorized data collection and access to users' private information. Federated Learning (FL) provides a decentralized framework in which multiple clients jointly optimize a shared model without directly sharing training data. We introduce Fed-IGTD, a horizontal FL-based approach in which clients share a common feature space and upload only local statistics, from which the server reconstructs the global correlation matrix and obtains a feature-to-pixel mapping to train a convolutional neural network via Federated Averaging (FedAvg). We studied two differential privacy (DP) mechanisms: (i) a one-shot Gaussian mechanism protecting the correlation-aggregation step; and (ii) a client-level DP-FedAvg variant protecting the training with privacy loss composed via a Renyi-DP accountant. We evaluated Fed-IGTD on five datasets under a systematic sweep of non-IID severity and privacy budgets. Privacy protection applied to the mapping-construction step has little impact on predictive performance. By contrast, DP-FedAvg substantially degrades utility at moderate privacy budgets with a small client cohort (K=5). Ablation study indicates that this degradation is driven mainly by the injected noise rather than by uniform averaging or update clipping. Macro-F1 exhibits sensitivity to performance degradation than accuracy under non-IID heterogeneity and class imbalance.
150. 【2610.39001】AlbumVSR: Identity-Preserving Face Video Super-Resolution from a Personal Photo Album
链接：https://arxiv.org/abs/2610.39001
作者：Nianyao Du,Xijun Wang,Stanley H. Chan
类目：Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
关键词：Severe blur, identity cues needed, compression can erase, cues needed, needed to faithfully
备注： 5 pages, 2 figures, 2 tables. Nianyao Du and Xijun Wang contributed equally
点击查看摘要
Abstract:Severe blur, downsampling, noise, and compression can erase the identity cues needed to faithfully restore a person's face. We therefore use photographs of the same person as an external source of identity. AlbumVSR conditions a frozen, one-step streaming diffusion transformer on a personal album. We use four MagFace/CLIP set tokens to encode the album, and 64 texture tokens which come from a fused UV atlas rendered at the current window's estimated face pose. These tokens are appended to the transformer's cached cross-attention context. On an identity-disjoint VoxCeleb2 benchmark with heavy synthetic degradation, AlbumVSR raises ArcFace similarity to the ground truth from 0.586 to 0.670 and face PSNR from 24.98 to 26.71 dB. On the 91 test identities, its mean identity similarity is higher than that of all 14 baselines.
151. 【2610.38957】What do MLLMs know about 'Waldo'? Evaluating Visual Reasoning under Multi-Instance Multi-Object Challenges
链接：https://arxiv.org/abs/2610.38957
作者：Saumya Mundra,Ajoy Mondal,Vineeth N Balasubramanian,C. V. Jawahar
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Multimodal Large Language, Large Language Models, Multimodal Large, Large Language, Recent advances
备注：
点击查看摘要
Abstract:Recent advances in Multimodal Large Language Models (MLLMs) have led to significant progress in visual reasoning capabilities. However, existing benchmarks curate image-question pairs pertaining to object categories with few instances, enabling trivial object referencing and limited visual ambiguity. In this work, we introduce MIMO-Bench, a benchmark for evaluating MLLMs in a Multi-Instance per Multi-Object categories (MIMO) setting. Each image-question pair in our benchmark pertains to instances belonging to object categories with multiple instances. The presence of such similar feature distractors causes hindrance in basic visual reasoning tasks, necessitating the referring expression to be composed of multiple visual subtasks to uniquely and unambiguously identify the target object. MIMO-Bench comprises 3,802 image-question pairs across attribute binding and numerical counting tasks. Despite impressive performances on existing benchmarks, both SOTA open-source (InternVL, QwenVL) and proprietary (Gemini, GPT-5) models exhibit substantially lower performance on MIMO-Bench, highlighting their limitations in the MIMO setting. The benchmark will be made publicly available.
152. 【2610.38951】GradSplat: Gradients Inside the Forward Pass for Feed-Forward 3D Gaussian Splatting
链接：https://arxiv.org/abs/2610.38951
作者：Jingong Chen,Linlian Jiang,Guangyuan Li,Xinxin Zuo,Yang Wang,Chulwoo Pack,Kyle Gao,Kwanghee Won
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Gaussian Splatting predicts, Gaussian Splatting, Splatting predicts, Gaussians optimized, Gaussians
备注：
点击查看摘要
Abstract:Feed-forward 3D Gaussian Splatting predicts a scene from a few images, but our predictor is less accurate than its own Gaussians optimized per scene on those images. This gap lies mostly in the shape of the Gaussians, their scale and rotation, since optimizing the shape alone recovers almost all of it and opacity and color recover less. Gradient steps on the input views correct this shape, and the model holds these views at inference. Log-anisotropy correction changes the axis ratios of each Gaussian in log space at constant volume, and in-forward gradient refinement renders the Gaussians into the input views and applies the gradient of the photometric loss with a learned recurrent update, first to opacity and color and then to the shape. GradSplat is a feed-forward model that refines its own Gaussians by learned gradient steps inside its forward pass, with no target view. On both complete six-view RealEstate10K protocols, it has the best PSNR and SSIM among nine methods and keeps its accuracy from $6$ to $42$ input views. Without fine-tuning, it is the most accurate on DL3DV and has the best PSNR and SSIM on ScanNet++. The code will be released at this https URL.
153. 【2610.38926】OneScene-Bench: Do Vision-Language Models Integrate Evidence Across Views?
链接：https://arxiv.org/abs/2610.38926
作者：Swathi Suhas,Anurag Das,Bernt Schiele,Jonas Fischer
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：increasingly process multiple, coherent scene-level representation, increasingly process, Vision-language models, integrate these observations
备注： Accepted at LM4RWD Workshop, Conference on Neural Information Processing Systems (NeurIPS), 2026
点击查看摘要
Abstract:Vision-language models (VLMs) increasingly process multiple views of the same scene, but it remains unclear whether they integrate these observations into a coherent scene-level representation. Existing multi-image evaluations do not isolate this capability, as tasks may admit correct solutions through independent per-view processing without requiring cross-view correspondence. We introduce OneScene-Bench, a controlled benchmark designed to explicitly distinguish when cross-view integration is necessary. Our benchmark evaluates three complementary capabilities on real-world multi-view scenes: unique object counting, cross-view consistency, and spatial reasoning. For counting, we control object overlap across views to separate cases solvable by simple per-view aggregation from those requiring instance-level correspondence. For consistency, we introduce localized edits in shared and view-specific regions, testing whether models distinguish genuine scene contradictions from differences caused by viewpoint and visibility. For spatial reasoning, we construct object relations that require combining spatial evidence across viewpoints to compare physical proximity. We evaluate twelve state-of-the-art open-weight and proprietary VLMs and find that counting accuracy drops substantially on complementary views, ranging from 4.5-50.0%, compared with 45.5-81.5% on complete-overlap examples. Moreover, cross-view consistency remains around 50% across models, including recent proprietary VLMs, despite substantially stronger performance on spatial reasoning. Our results reveal a substantial gap between processing multiple images and integrating evidence across them, highlighting the need for evaluations that explicitly test scene-level multi-view understanding. Code and data are available at this https URL.
154. 【2610.38830】Beyond Region Memory: Local Point Tracking for Vocal Cord Localization in Video Laryngoscopy
链接：https://arxiv.org/abs/2610.38830
作者：Huayu Wang,Gyungmin Ko,Jung Heon Kim,Jia-Xian Jian,Dae Kon Kim,Dong Keon Lee,Hsiang-Wei Huang,Pau-Choo Chung,Linda Shapiro,Richard B. Utarnachitt,Jenq-Neng Hwang,Nathan J. White
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Reliable vocal-cord localization, rapid tissue deformation, emergency video laryngoscopy, video laryngoscopy remains, laryngoscopy remains visually
备注： Accepted by WACV 2027
点击查看摘要
Abstract:Reliable vocal-cord localization in emergency video laryngoscopy remains visually challenging due to rapid tissue deformation, specular highlights, and instrument occlusions. We propose Local Point Tracking (LPT), a robust framework that combines a single-frame detector, local point propagation, and stateless mask refinement. LPT follows a localize first, snap second pipeline. First, a single-frame semantic detector generates sparse anchors, within which LPT initializes local query points (SAM-supported by default, with a regular grid as fallback). A point-propagation module tracks these points through detector-missed frames and recovers temporal bounding boxes from their geometry. Second, because point-derived boxes can be loose around the visible boundary, a box-snap step refines the selected box to the target boundary. This decomposition keeps temporal state in local point coordinates rather than a stateful mask or whole-region representation, while the spatial refinement module handles single-frame boundary precision without feeding mask errors back into temporal memory. Extensive evaluations on two distinct datasets demonstrate that our proposed LPT method achieves strong overall performance against existing baselines, with the best mAP_50, success AUC, and missing rate on both datasets. Furthermore, ablation studies confirm that point tracking primarily enhances temporal tracking robustness, while box-snap refinement effectively improves boundary tightness.
155. 【2610.38825】Evaluating Visual Realism of AI-Generated Videos with Eye Tracking
链接：https://arxiv.org/abs/2610.38825
作者：Danqing Shi,Lan Jiang,Jake Lyons,Miri Zilka
类目：Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
关键词：judgments or ratings, explicit human feedback, human feedback, Visual Turing Test, Human evaluation
备注：
点击查看摘要
Abstract:Human evaluation of AI-generated videos has so far relied on explicit human feedback, such as judgments or ratings. While these signals capture the conclusions viewers reach, they fail to offer insights into the perception that leads to them. This paper investigates eye tracking as a complementary human evaluation signal for AI-generated videos. We conducted two Visual Turing Test (VTT) experiments with a total of 90 participants on videos generated by latest generative models, in which participants watched each video twice and then gave their feedback. Across the experiments, we collected 4,286 valid eye-tracking trials on 438 videos. Our analysis indicates that gaze patterns offer informative evaluation signals that align with explicit human feedback. Building on the findings, we propose the first computational method GazeEval to predict the perceived visual realism of AI-generated videos from eye-tracking data. It achieves 71.3% accuracy in prediction, providing evidence that eye tracking data carries informative signals about how people evaluate a video. We release the study data to support further research on eye tracking for evaluating AI-generated video.
156. 【2610.38776】A 211K Parameter Foundation Model: Genetic Prompting in Neural Cellular Automata
链接：https://arxiv.org/abs/2610.38776
作者：John Kalkhof,Niklas Ihm,Daniel Kirste,Anirban Mukhopadhyay
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Neural Cellular Automata, modern foundation models, models enables broad, modern foundation, broad task generalization
备注：
点击查看摘要
Abstract:The scale of modern foundation models enables broad task generalization. In medical imaging, where segmentation demands vary widely across clinical contexts, such flexibility is invaluable. But current models, such as SAM trade adaptability for scale, locking their utility behind high-end infrastructure and leaving behind the low-resource environments most in need of diagnostic support. To bridge this gap, we introduce Foundational-NCA, a lightweight foundation model capable of adapting to a wide range of medical segmentation tasks without retraining. Built on Neural Cellular Automata (NCA), Foundational-NCA incorporates gradient-based learning with genetic optimization over a latent task space, using a compact task vector to guide predictions. This enables dynamic adaptation via a genetic prompt mechanism, allowing the model to generalize across tasks while keeping the parameter count magnitudes lower at just 211 thousand, compared to SAM-based models with 38.9 to 636 million parameters. We evaluate Foundational-NCA on 15 trained and 6 entirely unseen segmentation tasks from the TotalSegmentator dataset, achieving an average Dice score of 84.5% on known tasks and 60.4% on unseen ones. These results demonstrate that task generalization is possible in ultra-compact models, laying the groundwork for resource-efficient, deployable foundation models in clinical environments where infrastructure is limited. By reducing model size by over 180 times, Foundational-NCA takes a step toward a more equitable future for medical AI, one where powerful tools are within reach of every healthcare setting.
157. 【2610.38737】RuleForge: Inferring and Instantiating Latent Visual Rules as Executable Programs
链接：https://arxiv.org/abs/2610.38737
作者：Weifeng Xu,Yaxiong Wang,Zhedong Zheng
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：visual rule understanding, Visual rule reasoning, Visual rule, abstract concept, rule
备注： 13 pages, 5 figures
点击查看摘要
Abstract:Visual rule reasoning asks whether multimodal models can recover an abstract concept from a few examples. Existing studies either infer rules from visual examples but evaluate only classification or verbalization, or supervise executable generation with rules provided as input. Consequently, the relationship between visual rule understanding and reliable rule instantiation remains insufficiently explored. We present Bongard-Forge100k, a controlled benchmark, together with RuleForge, a backbone-agnostic framework that jointly learns visual rule understanding and rule-conditioned executable generation. The benchmark contains 108,797 problems: 86,000 for training, 19,297 for testing familiar rules, and 3,500 for testing held-out rule compositions. The framework progresses through rule-understanding training, joint understanding and scene-program supervision, and refinement via execution-grounded reinforcement learning with Understanding Replay. Executing generated programs under multiple random seeds provides verifiable feedback on executability, rule satisfaction, diversity, and robustness. With UniLIP as the primary backbone, RuleForge improves multiple-choice accuracy over the base model by 20.93 percentage points on familiar rules and 28.42 percentage points on held-out rule compositions, while increasing scene-level rule satisfaction on familiar rules from 2.91% to 70.99%. Additional experiments on InternVL3.5, Qwen3-VL, and Show-o2 demonstrate the effectiveness of the proposed training data and framework across diverse multimodal architectures.
158. 【2610.38720】nkGround: Benchmarking Fine-grained Visual Grounding Capabilities of Vision-Language Models for Cursive Handwriting in the Wild
链接：https://arxiv.org/abs/2610.38720
作者：Sidra Hanif
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：final ink, ordered stroke points, stroke trajectory recovery, trajectory recovery, ordered sequence
备注：
点击查看摘要
Abstract:Human handwriting is produced as an ordered sequence of pen movements, but an offline image keeps only the final ink. Handwriting trajectory recovery aims to restore the ordered stroke points and their pen-up/pen-down transitions. Existing image-trajectory pair datasets are recorded with digitizing tablets under controlled conditions, which limits writer coverage and stylistic diversity. Moreover, there is no benchmark that measures the visual grounding capabilities of general-purpose Vision-Language Models (VLMs) for the stroke trajectory recovery task. In this work, we introduce inkGround, a benchmark for stroke trajectory recovery from full-page, camera-captured handwriting. inkGround pairs 537 page images (~18.7k words, ~163k strokes) with reference trajectories produced by a pretrained offline-to-online model, filtered by OCR agreement, and verified by human raters. Because these references are pseudo-labels that represent one plausible pen trajectory, we also evaluate VLMs on the IAM-OnDB dataset, which provides recorded pen trajectories. We cast the task as text-conditioned point grounding: given a page image and a target word, the model must locate the word and output its ordered stroke points. Outputs are scored by trajectory fidelity (DTW, Chamfer distance), readability of the rendered strokes (Accurate Rate, Correct Rate), and style similarity. We benchmark four open-source and three frontier closed-source VLMs with a shared in-context prompt. Closed-source models, notably Gemini-3.1-Pro, produce legible trajectories close to the reference, and this ranking also holds on IAM-OnDB.
159. 【2610.38694】Geometry Matters in Packed 3D Attention
链接：https://arxiv.org/abs/2610.38694
作者：Amil Khan,Anantajit Subrahmanya,B.S. Manjunath
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
关键词：long-sequence modeling, memory layout, sequence, sequence position, attention
备注： Accepted to NeurIPS 2026 (Poster)
点击查看摘要
Abstract:Packing a 3D volume into a token sequence is not merely long-sequence modeling. Once the volume is packed, sequence order no longer encodes the geometry of the sample---so a positional mechanism indexed by sequence position can learn the memory layout rather than the object. This is a general failure mode for attention over packed geometric data. We introduce PACK3D, a local-global attention operator that separates memory layout from attention geometry. PACK3D keeps local neighborhood attention exact, compresses only distant context into contiguous 3D block summaries, and indexes every query-key interaction with axial RoPE on true token-grid coordinates rather than packed positions. Because the rotation is applied before a block is summarized, the query-to-summary logit remains a weighted sum of query-relative 3D displacements; compressed global context therefore stays spatially interpretable after packing, instead of collapsing to anonymous pooled content. We measure the failure mode directly. On a controlled synthetic task---train on one packing, change only the memory order of the tokens, hold the physical labeling rule fixed---learned positional embeddings and packed-sequence RoPE solve the training layout perfectly yet collapse under Morton, block-contiguous, and random repacking, while axial grid-coordinate RoPE stays exact throughout. The same coordinate system carries to real data: on Allen WTC-11 hiPSC microscopy, PACK3D improves over a parameter-matched 3D U-Net on the 15-structure joint task and over a matched learned-position local-global baseline, with controls isolating the coordinate system, not encoder capacity, as the source of the gain. A fused Triton implementation matches windowed-attention latency at 65k tokens while supplying the long-range context windowing cannot---so the operator is not only correct but trainable at scale.
160. 【2610.38678】AskStream: When and What to Ask Proactively under Uncertainty in Streaming Video Understanding
链接：https://arxiv.org/abs/2610.38678
作者：Lu Wang,Zhuoran Jin,Longxiang Wang,Junhong Chen,Yupu Hao,Yao Xu,Daojian Zeng,Yubo Chen,Kang Liu,Guang Liu,Jun Zhao
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：multimodal large language, Recent advances, large language models, advances in multimodal, multimodal large
备注： 42 pages, 9 figures. Project: [this https URL](https://www.askstream.top) ; Code: [this https URL](https://github.com/wanglu-cs/AskStream) ; Dataset: [this https URL](https://huggingface.co/datasets/wanglucs/AskStream)
点击查看摘要
Abstract:Recent advances in multimodal large language models have increased interest in interaction paradigms for video understanding. The evolution from offline to streaming and proactive interaction has gradually shifted models from passive responders toward more proactive agents that can decide when to answer. However, existing proactive paradigms typically assume that the information required to answer a query is complete, leaving models to only wait or answer. In the real world, however, information is often incomplete due to uncertainty from either the video or the user. In proactive paradigms, such uncertainty can cause the model to wait forever or generate highly hallucinatory responses. We therefore argue that proactive interaction should be redefined with a third action: ask, allowing models to actively ask for missing information. To study when and what models should ask before answering, we introduce AskStream, the first multi-turn benchmark spanning video and user uncertainty for streaming video understanding, with four task categories and eight subtasks. It contains 438 videos with 2,714 manually annotated frame-level time windows across three types: wait, ask, and answer. Across both multiple-choice and open-ended settings, even the strongest model attains only 52.18% on our proposed Ask-F1 metric, highlighting substantial limitations in proactive information seeking under uncertainty.
161. 【2610.38623】Cross-Primitive Consistency in Jev
链接：https://arxiv.org/abs/2610.38623
作者：Xu Liu,Guikun Chen,Kanzhi Wu,Wenguan Wang
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：large language models, enables efficient decision-making, Jev enables efficient, typed querying primitives, language models
备注：
点击查看摘要
Abstract:Jev enables efficient decision-making with large language models (LLMs) through typed querying primitives, allowing the same semantic state to be expressed in different formats, such as binary and multiple-choice queries. However, the consistency of decisions across primitives has not been systematically evaluated. This work introduces PrimaCons, a four-axis benchmark for assessing cross-primitive consistency while preserving the semantics of the query. Evaluations across three Jev sizes reveal consistency violations across all four axes. Our analysis shows that these violations follow systematic patterns and that cross-primitive discrepancies predict decision failure with a ROC-AUC of 0.91. To mitigate these inconsistencies, we propose the Cross-Primitive Residual Calibrator (CPRC), which attaches a lightweight residual network to a frozen backbone and jointly optimizes predictive accuracy and cross-primitive consistency. Under a leave-one-domain-out protocol, CPRC reduces cross-primitive inconsistency by 58.57% and improves accuracy by +2.97%., demonstrating its effectiveness in both in-domain and out-of-domain settings. Our findings establish cross-primitive consistency as a key dimension for evaluating Jev and provide a practical approach to improving the reliability of its decisions. Code and benchmark is available at this https URL.
162. 【2610.38613】RIM: Temporal Run-Time Monitoring of 3D Object Detection via Activation Statistics
链接：https://arxiv.org/abs/2610.38613
作者：Hakan Yekta Yatbaz,Konstantinos Koufos,Mehrdad Dianati
类目：Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
关键词：intelligent transportation systems, apparently plausible outputs, silently miss safety-critical, produce apparently plausible, Reliable perception
备注： 6 pages. Accepted for publication at the 2026 IEEE 23rd International Conference on Smart Communities: Improving Quality of Life using AI, Robotics and IoT (HONET)
点击查看摘要
Abstract:Reliable perception is essential to intelligent transportation systems, yet deep neural networks can silently miss safety-critical objects while continuing to produce apparently plausible outputs. Existing run-time monitors for 3D object detection generally assess each frame independently and therefore cannot exploit the temporal evolution of occlusion, point sparsity, and scene complexity. We present TRIM, a lightweight temporal monitor that detects missed-object failures without modifying or retraining the underlying detector. TRIM summarises intermediate activations using compact channel-wise statistics and processes short sequences of these summaries with a temporal convolutional network. Experiments on NuScenes show that temporal monitoring improves AUROC from 0.686 to 0.704 and reduces the false alarm rate at 95% recall from 0.818 to 0.759 compared with single-frame monitoring. Early and intermediate backbone features provide the strongest monitoring signal, while the temporal inference head requires only 0.159 ms on a CPU.
163. 【2610.38611】V-CAPE: Visual Consistency Assessment for Product Image Editing
链接：https://arxiv.org/abs/2610.38611
作者：Iaroslav Melekhov,Roberto Alcover Couso,Francesco Taioli,Daniel Coelho,Artur Bekasov,Virginia Fernandez Arguedas,Jose Miguel Grande Saiz
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：natural language instructions, Recent advances, editing enable complex, enable complex transformations, background replacement
备注：
点击查看摘要
Abstract:Recent advances in generative image editing enable complex transformations such as background replacement, object manipulation, and pose adjustment from natural language instructions. However, verifying that these transformations preserve visual consistency, defined as the preservation of an object's intrinsic visual attributes, including geometry, texture, color, and fine-grained details, remains a largely unsolved problem. This challenge is particularly pronounced at scale, where manual inspection is infeasible and automated quality assessment (AutoQA) could become essential. In this work, we introduce Visual Consistency Assessment for Product Image Editing (V-CAPE), a large-scale dataset for evaluating visual consistency in generative image editing. The dataset consists of 40K unique source-target product image pairs, each annotated with binary accept/reject labels and structured failure modes capturing fine-grained inconsistencies such as geometric distortions, pose errors, and identity mismatches. V-CAPE combines real-world edited images with a synthetic subset constructed from 3D assets, enabling both realistic evaluation and controlled analysis. Vision-language models (VLMs) across all evaluated size categories on AutoQA achieve accuracy below 65%, indicating that the dataset poses a substantial challenge regardless of model scale. Our results highlight fundamental limitations of current models in verifying visual consistency under transformation, positioning V-CAPE as a benchmark for advancing fine-grained image editing evaluation.
164. 【2610.38600】Data Scaling Laws in Stereo Matching: From In-domain Fitting to Cross-domain Generalization
链接：https://arxiv.org/abs/2610.38600
作者：Tong Zhao,Di Liang,Yiming Zeng,Bo Xu,Yingzhi Tang,Chen Bai
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：million-scale synthetic images, generalizable foundation models, foundation models trained, era of large, developing towards generalizable
备注： WACV 2027
点击查看摘要
Abstract:In the era of large models, stereo matching is developing towards generalizable foundation models trained on million-scale synthetic images. However, how model performance scales up with synthetic data is still underexplored. In this work, we preliminarily investigate the in-domain and cross-domain data scaling laws in stereo matching. A realistic synthetic dataset, named XStereo, containing 200 thousand samples is first built to serve as a high-quality benchmark for data scaling research. In-domain training on five large-scale datasets demonstrate a two-regime scaling behavior with data, transitioning from the standard power law to saturation, where the critical transition point is approximately invariant under a fixed model. Cross-domain performance exhibits a substantially weaker and less predictable dependence on the source-domain data scale. The distribution and coverage diversity are more effective for diminishing domain gaps. More instructive suggestions are provided for high-efficiency data construction and training. The insightful findings are significant verification and extension of the existing scaling laws in both language and vision, paving the way to efficient and stronger stereo matching. The XStereo dataset is public at this https URL.
165. 【2610.38593】SpectralTF-MAE: A Pre-trained Spectral Foundation Model for Tokamak Fusion Plasma Diagnostics
链接：https://arxiv.org/abs/2610.38593
作者：Xiao Wang,Zhijie Lu,Duoqing Yang,Jiangtao Yu,Jin Tang,Zichao Lin,Hongming Zhang,Guosheng Xu,Bo Lyu,Qiang Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：provide rich information, existing learning-based approaches, characterizing plasma states, diagnostics provide rich, provide rich
备注：
点击查看摘要
Abstract:Spectroscopic diagnostics provide rich information for characterizing plasma states in tokamak fusion experiments, while existing learning-based approaches are often designed for individual tasks and rely heavily on labeled data. In this work, we propose SpectralTF-MAE, a self-supervised spectral foundation model for tokamak plasma diagnostics. The model is pretrained on paired charge-exchange recombination spectroscopy (CXRS) and X-ray crystal spectroscopy (XCS) measurements through masked spectral reconstruction. To better capture diagnostically important spectral structures, we introduce a peak-aware masking strategy together with a peak-enhanced reconstruction objective, encouraging the model to focus on characteristic spectral regions. A shared Transformer encoder is employed to learn transferable representations across different diagnostic modalities. The pretrained model is further adapted to several downstream tasks, including plasma-parameter prediction, cross-diagnostic spectral generation, and spectral peak analysis. Experiments on EAST diagnostic data demonstrate that the proposed framework achieves competitive performance across different tasks and consistently benefits from self-supervised spectral pretraining. These results indicate the potential of SpectralTF-MAE as a general and transferable representation learning framework for plasma spectroscopic diagnostics.
166. 【2610.38587】RoSAMDepth: Robust Self-supervised Depth Estimation Leveraging Segment Anything Model
链接：https://arxiv.org/abs/2610.38587
作者：Xuanang Gao,Zhiwei Ning,Gengming Zhang,Jiaxi Cao,Runze Yang,Zhonglong Zheng,Jie Yang,Rong Xiao,Wei Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：maintain high-quality depths, aims to maintain, maintain high-quality, depth estimation aims, depth
备注： 13 pages, 8 figures, including supplementary material. Published in CVPR 2026
点击查看摘要
Abstract:Robust depth estimation aims to maintain high-quality depths across diverse conditions. However, most existing methods estimate depth without taking into account the object-level information. As a result, the predicted depth may easily deviate within objects and become blurred under adverse conditions. To overcome this weakness, we propose RoSAMDepth, a novel framework that can assist robust self-supervised depth estimation in leveraging rich and diverse object-level priors from the Segment Anything Model (SAM). We focus on incorporating object-level information across three key aspects: a segment-guided representation contrasting method that injects object-level awareness into the feature representation space; an adaptive regional outlier masking strategy combined with a regional Gaussian likelihood loss that enforces regional depth smoothness; and an object-level reliability estimation strategy that mitigates the influence of unreliable supervision. Extensive experiments across multiple datasets and diverse weather conditions demonstrate that our method produces sharper, more accurate depth predictions, consistently outperforming state-of-the-art methods. Our code is available at this https URL.
167. 【2610.38547】Detecting Fine-Grained Differences between Image Datasets
链接：https://arxiv.org/abs/2610.38547
作者：Leander Girrbach,Alexander von Recum,Kirill Bykov,Zeynep Akata
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：data distributions differ, concepts, distributions differ, data distributions, fundamental statistical problem
备注：
点击查看摘要
Abstract:Detecting whether and in particular how two data distributions differ is a fundamental statistical problem. For example, we might be interested in which concepts (expressed in natural language) differ between two classes, the outputs of different generative models, or the same model with different prompt specifications. When working with large and complex image datasets, we intuitively think that first selecting concepts from a broad list using fine-grained patch embeddings, and then optimizing the concepts to better fit the data is an effective workflow. However, prior work instead uses MLLMs to propose concepts and image-level embeddings to select them. Therefore, we systematically compare these techniques and find that selecting concepts from broad lists can replace MLLM-proposals at obviously lower cost. MLLM inference can then be spent on optimizing selected concepts instead to achieve substantial improvements. Additionally, by designing a new benchmark that focuses on low-prevalence and multi-concept distinctions, we find that selecting nonredundant concepts instead of only the most distinguishing concepts matters, and we introduce techniques for how to effectively do this. Finally, we show how our methods address important problems like detecting spurious correlations and biases. Overall, our work presents a fundamental perspective shift on how we analyze interpretable distributional differences between datasets.
168. 【2610.38518】Self-Consistency Geometry Injection for Multi-view Consistent Image Generation
链接：https://arxiv.org/abs/2610.38518
作者：Yupeng Xiao,Yankai Wu,Jianjie Luo,Jiajie Lin,Yi Yu,Zhenguo Yang
类目：Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
关键词：Recent diffusion models, achieved impressive progress, Recent diffusion, geometry, views efficiently
备注： Accepted by ACCV 2026
点击查看摘要
Abstract:Recent diffusion models have achieved impressive progress in single image to multi-view generation by synthesizing consistent novel views efficiently. Existing methods usually explicitly or implicitly couple geometry inference and texture synthesis, making generation vulnerable to inaccurate geometry prediction. Although several geometry-guided methods have explored decoupling, they usually treat predicted geometry as a fixed condition and do not explicitly model whether the geometry is reliable. We therefore propose Reliability-Guided Multi-View Generation (RGMV), which comprises a parameter-free Self-Consistency Geometry Gate (SCG) and a Reliability-aware Geometry Injection (RGI) module. Specifically, a geometry proposal generator produces multiple geometry predictions for the same reference image and target viewpoints under different stochastic sampling configurations. SCG measures their spatial agreement to derive a reliability map, which RGI uses to modulate the injection strength of geometry-conditioned residuals in the RGB synthesis model. Moreover, RGMV can adapt to different geometry proposal generators. As a result, RGMV preserves stronger geometric guidance in stable regions while attenuating unreliable geometry cues before they propagate into multi-view RGB generation. Extensive experiments show that reliability-aware geometry injection improves the geometric quality of multi-view generation. Code: this https URL.
169. 【2610.38516】he Power Watershed as a Differentiable Segmentation Layer
链接：https://arxiv.org/abs/2610.38516
作者：Galia Akl(AUB),Laurent Najman(LIGM, KUSTAR),Hugues Talbot(OPIS, CVN)
类目：Computer Vision and Pattern Recognition (cs.CV); Neural and Evolutionary Computing (cs.NE); Image and Video Processing (eess.IV); Optimization and Control (math.OC)
关键词：seeded Random Walker, classical tool, tool of morphological, Random Walker, layer
备注：
点击查看摘要
Abstract:Can a Watershed transform, the classical tool of morphological segmentation, be placed at the end of a deep network and trained through? The Power Watershed (PW) is the limit of the seeded Random Walker (RW) as p grows to infinity with edge weights raised to the power p. At every finite p the RW solves a sparse positive-definite linear system and is exactly differentiable with respect to the edge weights and the unaries through one adjoint solve; its multi-label form costs one right-hand side per label, and its limit is the PW. We study this family as the last layer of a segmentation network, the watershed layer (WS layer). A conditioning study measures how the solve, the gradient and the distance to the limit evolve with p, and the implicit gradient is checked against unrolling. On synthetic data with correct edge costs the WS layer transforms a soft prediction into an exact boundary. On nuclei segmentation (PanNuke), a binary WS layer is level with a thresholded U-Net. With the instances as labels, the WS layer turns the U-Net into an instance segmenter whose number of labels is set per image by the markers rather than by the output layout, and the network is trained through that partition: instance F1 score rises from 0.656 to 0.795 over a U-Net with connected components, level with the same network followed by a tuned marker watershed. At the training temperature the partition is a watershed of the learned costs, and the finite temperature lets the solve hedge, discounting weak markers and sharing contested pixels. The hedging puts the WS layer 0.015--0.020 F1 ahead of the baseline when trained on a tenth of the data, at the price of having to deploy the finite-p solve rather than its exact limit at inference. Code and the full experimental record are released.
170. 【2610.38511】Similar Scores, Opposite Failures: Human Correction as an Instrument for Video Caption Evaluation
链接：https://arxiv.org/abs/2610.38511
作者：Ankit Khedia,Astitva Jaiswal,Chaitanya Prasad,Anisha Raju,Aangeeras
类目：Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
关键词：Video captioning models, single-sentence image captions, Video captioning, increasingly evaluated, designed for single-sentence
备注： 29 pages, 10 figures, 17 tables. Analysis code included as ancillary files. Data available on request at [this https URL](https://data.deccan.ai)
点击查看摘要
Abstract:Video captioning models are increasingly evaluated with metrics designed for single-sentence image captions. We test whether those metrics recover human judgement on dense, multi-window video description, and find that they do not. Six frontier models captioned 196 short clips. Trained annotators post-edited every caption against a failure taxonomy fixed before annotation, once as a whole-clip overview and once window by window, and a quality-control reviewer checked each annotation, yielding 1,144 overview verdicts and 4,672 window verdicts. Reference-overlap metrics do not separate the captions reviewers accepted from those they rejected: BLEU-4 reaches an AUC of 0.510 and a CIDEr-style tf-idf cosine 0.528, and restricting references to reviewer corrections leaves BLEU-4 at 0.479 (95% CI 0.431-0.529), indistinguishable from chance. Review granularity changes what is detected. The same annotators record camera and shot errors on 2.7% of captions when reading the overview and on 12.8% when reading window by window, a 4.7x gap that holds across every robustness check we ran, on the mode most likely to need a near-total rewrite. Correct-overview rates span 54.1-72.7%, but only 4 of 15 pairwise gaps survive Bonferroni correction (5 after adjusting for reviewer and clip effects), and accuracy and derived severity order the models differently. Models with statistically indistinguishable accuracy have sharply different failure profiles, which no single scalar shows. We argue that caption evaluation should record what was wrong rather than only how often, and we report the taxonomy, the per-model profiles, the corpus properties that predict difficulty, and the code that regenerates every number in the paper.
171. 【2610.38507】UnMixMamba: Unmixing Guided State Space Subpixel Mapping of Crops from MODIS Time Series
链接：https://arxiv.org/abs/2610.38507
作者：Motasem Alkayid,Yimin Zhu,Naser El-Sheimy,Lincoln Linlin Xu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Resolution Imaging Spectroradiometer, Moderate Resolution Imaging, labels supports acreage, Moderate Resolution, Imaging Spectroradiometer
备注：
点击查看摘要
Abstract:Subpixel mapping (SPM) of crop types from \SI{250}{m} Moderate Resolution Imaging Spectroradiometer (MODIS) time series to field-scale Agriculture and Agri-Food Canada (AAFC) labels supports acreage estimation and agricultural monitoring. Three difficulties remain: (i)~spectral-temporal coupling and mixed pixels that break direct coarse-to-fine classifiers; (ii)~intra-class endmember variability that a single spectrum per crop cannot capture; and (iii)~uniform refinement on the high-resolution (HR) grid, which treats every subpixel the same, so most of the effort goes to easy pixels while errors remain in the minority class. UnMixMamba addresses them as follows. For~(i), a temporal-spectral-spatial decoupled (TSSD) factorization estimates simplex fractions and drives AAFC classification from the same mixture field, instead of mapping low-resolution (LR) observations directly to HR logits. For~(ii), \(\times 8\) abundance super-resolution and a learnable per-crop variant endmember bank reconstruct the MODIS cube at LR and HR under a linear mixing model. For~(iii), the predicted crop fractions decide where a fixed Mamba token budget is spent, so HR refinement focuses on the minority class instead of the full grid. On Alberta 2015 under a fixed spatial-block protocol, UnMixMamba reaches OA 0.8397, mIoU 0.6640, AA 0.7837, and Kappa 0.7715, ahead of ST-DRes, ESPCN, FineCrop, AgriFM, and classical SPM baselines. The task is closed-set four-crop mapping inside a known agricultural mask.
172. 【2610.38499】Residual Motion Splatting for Continuous Space-Time Video Super-Resolution
链接：https://arxiv.org/abs/2610.38499
作者：Vincent Herfeld(Thoth),Eva Coupet{é},Julien Mairal(Thoth),Jean Ponce(ENS-PSL)
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：arbitrary spatial scales, space-time video super-resolution, video super-resolution produces, Continuous space-time video, spatial multi-frame super-resolution
备注：
点击查看摘要
Abstract:Continuous space-time video super-resolution produces a high-resolution, highframe-rate video from a low-resolution, low-frame-rate one at arbitrary spatial scales and temporal instants. Classical spatial multi-frame super-resolution has shown that detail beyond a single frame's sampling limit can be recovered by aligning complementary observations with sub-pixel precision and accumulating them on a finer grid. Existing learning-based methods largely depart from this principle: some align and fuse frames only at low resolution before regressing a continuous high-resolution representation, while others interpolate each frame to the output resolution before splatting. In both cases, the per-frame samples are resampled before fusion, so their sub-pixel positions are lost before frames are combined. Restoring this principle in a learned continuous model raises two challenges: estimating sub-pixel motion at unobserved times, and handling holes and occlusions when splatting a coarse grid onto a finer one. We address the first by predicting per-pixel continuous trajectories that refine linear optical-flow paths with smooth residual motion, using features integrated over the full video sequence. We address the second by directly splatting consecutive low-resolution features in a learned latent space onto a high-resolution grid at their predicted sub-pixel positions with predicted visibility weights, before a decoder resolves remaining ambiguities. This makes temporal interpolation and multi-frame superresolution a single operation. Our method improves both reconstruction fidelity and perceptual quality on multiple standard benchmarks over the state of the art, while running up to an order of magnitude faster.
173. 【2610.38497】Vision-Based Density Estimation for Adaptive Traffic Control
链接：https://arxiv.org/abs/2610.38497
作者：Mohammad Usaid,Deepak Silaych,Hrithik Mhatre,Archak Mittal
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：growing urban centers, inadequate transport infrastructure, transport infrastructure place, rapidly growing urban, infrastructure place increasing
备注： 14 pages, 5 figures, 2 tables
点击查看摘要
Abstract:Traffic congestion is becoming a critical challenge in rapidly growing urban centers, particularly in developing nations where rapid urbanization and inadequate transport infrastructure place increasing strain on transport networks. In highly heterogeneous and chaotic traffic flows, camera-based traffic-state estimation is limited by severe vehicle occlusion and overlap. In this paper, we propose a framework to optimize traffic flow directly from stochastic visual observations. The architecture is structurally decoupled into three primary components: (a) vehicle density estimation, (b) predictive dynamics modeling, and (c) policy optimization. The proposed methodology introduces a convolutional neural network (CNN) that learns the mapping from camera feeds to spatial density maps, utilizing an adaptive k-nearest-neighbor (k-NN) Gaussian kernel to handle varying congestion levels. This feeds into an ST3DNet predictive module for spatiotemporal forecasting, followed by a hierarchical Graph-Transformer Multi-Agent Reinforcement Learning (MARL) algorithm to optimize signal phases. The density estimation stage was evaluated on 40 annotated intersection images from Rambagh 4, Jaipur, India, with a reported five-fold validation mean absolute error (MAE) of $3.32 \pm 0.95$ vehicles and a mean absolute percentage error (MAPE) of 14.5%. These results pertain to the perception component; empirical training and evaluation of the forecasting and control stages remain future work.
174. 【2610.38494】Calibrating Track-Query Updates at Test Time for Multi-Object Tracking
链接：https://arxiv.org/abs/2610.38494
作者：Riku Inoue,Shogo Sato,Yu Mitsuzumi,Kazuhiko Murasaki,Tomoyasu Shimada,Takashi Sano
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：localizing multiple objects, Multi-Object Tracking, requires localizing multiple, identities over time, localizing multiple
备注：
点击查看摘要
Abstract:Multi-Object Tracking (MOT) requires localizing multiple objects and maintaining their identities over time. Recent query-propagation trackers carry target-specific representations across frames and update them with current-frame image features, yet identity association remains difficult when visually similar targets overlap or cross. During such interactions, current-frame query updates may contain components directed toward nearby active tracks, which may affect identity association in later frames. To investigate this issue, we analyze track-query updates preceding identity switches (IDSWs), where the same target is assigned a different track ID. Our analysis shows that (i) propagated track-query representations change more abruptly before IDSWs, (ii) current-frame updates have larger positive projections along directions toward other active tracks, and (iii) bounding-box intersection over union (IoU) retrieves the counterpart track involved in a subsequent IDSW with high recall when that track remains active before the switch. Motivated by these findings, we propose Test-time Track-Query Calibration (TQC), a training-free and gradient-free method for query-propagation trackers. TQC calibrates the tracker output before propagation using a bounded correction estimated from update components directed toward nearby active tracks. It requires only track-query representations and bounding boxes available during inference, without retraining or architectural modification. Across nine DanceTrack and SportsMOT test settings, TQC improves HOTA and AssA in all settings and IDF1 in eight, with an average HOTA gain of +0.38 and limited inference overhead. Controlled variants on six validation settings further show that TQC's correction direction yields more consistent HOTA improvements than alternative directions with matched correction norms.
175. 【2610.38482】Geometry-Aware Trajectory State Modeling for Few-Shot Action Recognition
链接：https://arxiv.org/abs/2610.38482
作者：Zhiqi Miao,Marko Haralović,Stanislav Levendeev,Sarvesh Angadi,Jannatul Ferdous,Dan-Cristian Ploesteanu,Jiapan Guo,Estefanía Talavera
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Few-shot action recognition, expose discriminative motion, discriminative motion dynamics, Few-shot action, requires representations
备注： This work is accepted by The IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2027
点击查看摘要
Abstract:Few-shot action recognition (FSAR) requires representations that expose discriminative motion dynamics from only a few labeled videos. Recent trajectory-based methods leverage long-term point tracks for explicit temporal correspondence, but often treat trajectories as carriers of sampled visual features or summarize their motion with compact 2D displacement-based descriptors. We instead view an action as the ordered evolution of 2.5D trajectory states and their local geometric interactions. Based on this view, we introduce GeoTRAM, a geometry-aware trajectory state modeling framework for FSAR. GeoTRAM lifts sparse tracks into 2.5D trajectories, where StateMotion models ordered intra-trajectory state evolution and RelMotion captures state-conditioned local interactions through time-varying relative geometry and motion. Quality-filtered spectral sampling provides more reliable trajectory initialization; complementary semantic, motion, and depth cues are separately refined and preserved for joint spatio-temporal modeling; action-guided reweighting and top-$\rho$ trajectory selection support complementary frame- and trajectory-level matching. Experiments on five FSAR benchmarks demonstrate strong gains on the motion-centric Something-Something benchmarks while maintaining competitive performance on more appearance-driven benchmarks and a favorable accuracy--efficiency trade-off.
176. 【2610.38462】Automatic Iconographic Annotation of Medieval Illuminations: Comparing Generative Vision-Language Models and Supervised Discriminative Classifiers
链接：https://arxiv.org/abs/2610.38462
作者：Vincent Jolivet(ENC, CJM),Lucas Terriel(ENC),Chahan Vidal-Gor{è}ne(CJM, ENC)
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：medieval manuscripts creates, automated iconographic indexing, large-scale digitisation, digitisation of medieval, medieval manuscripts
备注：
点击查看摘要
Abstract:The large-scale digitisation of medieval manuscripts creates a growing need for automated iconographic indexing. We evaluate two complementary paradigms for this task on 9,735 full-folio images annotated with 2,333 TIMEL concepts: generative Vision-Language Models (VLM) and supervised discriminative classifiers (SDC). The VLM experiments evaluate fine-tuning, structured JSON decoding, taxonomybased candidates generation, and deterministic reranking. The discriminative experiments evaluate flat, hierarchy-aware, and patch-attention heads over a frozen DINOv3 encoder. Our results highlight key functional differences: ML-Decoder (SDC) achieves the strongest overall retrieval performance (precision@1 of 0.696, MRR of 0.806), with loss calibration (Asymmetric Loss) serving as the primary performance lever. While SDCs provide superior precision and reliability within a closed vocabulary, VLMs offer strong semantic generalization and broad taxonomic coverage, particularly through structured JSON targets and verify-calibrate reranking (hierarchical F1 of 0.559). Furthermore, model behavior strongly depends on concept abstraction: SDCs excel at perceptually grounded descriptive concepts, whereas VLMs provide useful hypotheses for complex interpretative scenes. Finally, we find that enforcing explicit hierarchy within model architectures limits coverage compared to spatial feature aggregation. Overall, rather than competing approaches, SDCs and VLMs emerge as complementary components for future hybrid indexing systems combining structured retrieval with exploratory annotation assistance. As a methodological contribution, we define the soft F1 metric to provide a taxonomy-aware evaluation that credits semantically close predictions based on lowest common ancestors.
177. 【2610.38425】IntReal: A Real-World Dataset and Benchmark for Multimodal Interleaved Generation
链接：https://arxiv.org/abs/2610.38425
作者：Shaohan Yu,Jie Qin,Wende Tan,Haiwen Diao,Limeng Qiao,Yueming Lyu,Chenyang Si
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：consume multimodal content, routinely consume multimodal, Humans routinely consume, Unified Multimodal Models, interleaved generation
备注： [this https URL](https://yushaohan.github.io/IntReal/)
点击查看摘要
Abstract:Humans routinely consume multimodal content, where textual and visual information is naturally interleaved. Although Unified Multimodal Models (UMMs) can jointly generate text and images, multimodal interleaved generation remains largely underexplored. Progress in this area is hindered by two key limitations: (1) the lack of well-defined interleaved generation tasks and corresponding training data, and (2) the absence of fine-grained evaluation protocols that can reliably diagnose model performance across different capability dimensions. To this end, we present IntReal, comprising two complementary components: IntReal-100K, a real-world dataset of multimodal interleaved samples spanning four major categories, and the IntReal Evaluation Suite, which evaluates model performance across seven complementary dimensions using specialized expert models, enabling coarse-to-fine diagnosis. Extensive experiments validate our evaluation suite through low inter-metric redundancy (max pairwise absolute Spearman's $\rho \leq 0.42$ ), high agreement with human judgments (83.3% accuracy), and strong consistency across different expert evaluators (Spearman's $\rho 0.9$ ), supporting the complementarity and reliability of our metrics. By benchmarking four families of UMMs, we show that IntReal Evaluation Suite reveals distinct strengths and weaknesses across models. Post-training experiments demonstrate the utility of IntReal-100K for improving interleaved generation and further show that training unimodal generation can also benefit interleaved generation. These findings suggest that interleaved generation can benefit from modality-specific optimization, rather than requiring joint text-image training at every stage.
178. 【2610.38413】Structure-retained low-rank adapters for weather synthesis
链接：https://arxiv.org/abs/2610.38413
作者：Shunxin Wang,Alexandros Stergiou,Luuk Spreeuwers,Estefania Talavera,Nicola Strisciuglio
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Real-world deployment, environmental interferences, limited by environmental, Real-world, Improving model robustness
备注：
点击查看摘要
Abstract:Real-world deployment of computer vision models, particularly in outdoor scenarios, is often limited by environmental interferences, such as adverse weather. Improving model robustness across weather conditions is thus critical for reliability. However, sourcing high-quality data relevant to isolated weather effects is a significant challenge. Most datasets include changes in dynamic elements across observations of the same scene: subjects such as pedestrians and vehicles change between shots, while camera viewpoints vary. These aspects make it difficult to decouple solely weather-induced distribution shifts from the scene. In this work, we address this with a novel weather synthesis framework that utilizes structure-retaining Low-Rank Adapters (LoRA). We analyze the attention blocks in Stable Diffusion XL (SDXL) and identify where relevant structural information is mostly processed during image-to-image translation. Through an architectural analysis of SDXL, we learn structural LoRA matrices for the corresponding blocks and then leverage them for weather synthesis, eliminating the need for external conditioning such as ControlNet. Experimental results demonstrate that our method improves upon existing synthesis approaches, achieving greater realism while preserving structure. This provides a robust pipeline for generating high-fidelity training and evaluation data for outdoor computer vision applications.
179. 【2610.38389】Visually Irrelevant Hallucinations in Medical Vision-Language Models
链接：https://arxiv.org/abs/2610.38389
作者：Ke Zou,Nan Zhou,Tao Yang,Meng Liu,Linchao He,Zehui Liao,Ling Huang,Guangren Huang,Yidi Chen,Yi Zhang,Hu Chen,Huazhu Fu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：show strong performance, visual question answering, Medical vision-language models, producing free-form clinical, radiology report generation
备注：
点击查看摘要
Abstract:Medical vision-language models (VLMs) show strong performance in radiology report generation (RRG) and medical visual question answering (Med-VQA), producing free-form clinical descriptions and answers from medical images. However, they may generate clinically meaningful statements that lack visual evidence or contradict image content. We refer to these errors as visually irrelevant hallucinations and investigate them at the phrase level. We propose CLEAR (Counterfactual Localized Evaluation of Alignment and Response), a model-agnostic post-hoc framework for phrase-level hallucination risk estimation. CLEAR decomposes generated responses into visually assessable clinical phrases. For each phrase, phrase-conditioned grounding identifies a candidate visual region, which is selectively perturbed by a localized counterfactual intervention. CLEAR characterizes the intervention using two complementary auditing cues: response cues, capturing semantic and clinically meaningful changes in the generated response, and alignment cues, measuring changes in image-phrase compatibility. These cues are normalized and jointly modeled to estimate phrase-level hallucination risk. Across four datasets spanning RRG and Med-VQA, CLEAR outperforms state-of-the-art hallucination detection baselines across multiple medical and general-purpose VLMs. On MIMIC-CXR, CLEAR improves AUPRC from 85.10% to 90.23%. Its effectiveness is further validated on a GT-derived controlled VIH benchmark. These results establish visually irrelevant hallucination as an important fine-grained reliability failure mode in medical VLMs and demonstrate the value of model-agnostic visual-support auditing for more trustworthy multimodal clinical generation.
180. 【2610.38360】SparseTokenGS: Gaussian-based Sparse-View CBCT Reconstruction via Reusable Token Pool
链接：https://arxiv.org/abs/2610.38360
作者：Zihan Deng,Zhenzhen Gong,Yujie Liu,Jiaxin Gong,Dingkun Liu,Kexian Xie,Yixing Huang,Peng Jin,Tianye Niu,Zhisheng Wang,Fenglin Liu,Hongming Shan,Lequan Yu
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：low-dose volumetric imaging, introduces severe artifacts, Sparse-view cone-beam computed, low-dose volumetric, image-guided interventions
备注： 10 pages, 12 figures. Submitted to IEEE Transactions on Medical Imaging
点击查看摘要
Abstract:Sparse-view cone-beam computed tomography is widely used for fast, low-dose volumetric imaging and lesion localization in dentistry and image-guided interventions, but it introduces severe artifacts. Feed-forward Gaussian methods transfer the reconstruction cost to training and decode each coordinate from projection-aligned features and neighboring anchors. However, information from the whole scan is written only when the field is constructed and cannot be retrieved during coordinate decoding, leaving the volume inconsistent within the scan. This is a common limitation of such methods: once the field is fixed, decoding uses only local features and reconstruction quality drops substantially. We propose SparseTokenGS, a Gaussian-based method for sparse-view cone-beam computed tomography reconstruction via a reusable token pool. Anchor Token Fusion and Point-Token Attention query this shared pool at the anchors and at the query coordinates, respectively. The local Gaussian prior is retained, and the volume is reconstructed in one forward pass, without per-scan optimization or a projection-domain loss. On three public chest and dental datasets, SparseTokenGS achieves the highest reconstruction performance among the compared methods. On clinical dental scans, reconstruction from only six projections reaches 29.51 decibels in peak signal-to-noise ratio and a structural similarity of 0.842 against the full-view reference, and off-the-shelf tooth-and-pulp segmentation keeps these structures separable for root-canal localization. Experiments on volcanic rock and chicken show performance after same-object adaptation. A reusable token pool gives feed-forward reconstruction a mechanism that keeps scan-wide context available through decoding and can benefit future rapid clinical lesion localization and analysis. The code is available at: this https URL.
181. 【2610.38358】Volley3D: Unified 3D Full-Sequence Reconstruction from Monocular Broadcasts
链接：https://arxiv.org/abs/2610.38358
作者：Noâm Boussouf,Adrien Maglo,Mohamed Chaouch,Hideo Saito
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：notoriously ill-posed due, single-view sports broadcast, reconstruction from single-view, moving cameras, depth ambiguity
备注： 8 pages + 10 pages of supplementary. Videos available on the project webpage: [this http URL](http://noambsf.github.io/Volley3D)
点击查看摘要
Abstract:3D reconstruction from single-view sports broadcast is a pivotal capability for automated analytics and immersive AR/VR replays, yet notoriously ill-posed due to difficult-to-calibrate moving cameras and depth ambiguity. Unlike widely studied racket sports, team sports such as volleyball feature unpredictable multi-contact sequences and high altitude jumps. These dynamics categorically invalidate the standard planar projection assumptions relied upon by existing methods. To resolve these limitations without 3D training supervision, we propose a complete, physics-grounded pipeline tailored to monocular broadcast volleyball. First, we include a correction module to the perspective drift induced by ground-plane projection by coupling temporal jump spotting with Newtonian interpolation for 3D player placement. Second, for 3D ball trajectory estimation, we design a dedicated sub-trajectory segmentation architecture alongside a collision-matching algorithm. By anchoring physical equations to reconstructed 3D hand vertices, we lift noisy ball 2D detections into a spatiotemporally coherent 3D scene. Evaluations on multi-camera benchmarks demonstrate our approach significantly reduces 3D spatial drift compared to 2D-reprojection baselines.
182. 【2610.38352】WeVisCut: The Whole Editing Room in One Agent From Narration to Long-Form Video
链接：https://arxiv.org/abs/2610.38352
作者：Chenyu Mu,Wenming Weng,Qinrui Wang,Xingsong Ye,Hao Yu,Jun He,Chong Sun,Chen Li,Jing Lyu,Erkun Yang,Cheng Deng
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：narration requires interdependent, Producing long-form videos, Producing long-form, requires interdependent decisions, coherent across shots
备注：
点击查看摘要
Abstract:Producing long-form videos from narration requires interdependent decisions that remain coherent across shots. Existing agentic editors often use multi-stage workflows driven by general models, leaving the quality-cost trade-off of compact, task-adapted policies unclear. We address this gap with WeVisCut, a long-form script-to-video system that connects tool execution with policy learning. Its agent harness maintains shared production state so that tool feedback informs subsequent decisions and repairs. Because finished videos do not expose these decisions, our four-stage data pipeline combines editing inversion with verified replay to turn recovered editing intent into executable supervision. Applied to 3,750 finished videos, this pipeline yields over 130K training examples. We use this supervision to train a single 9B policy that coordinates production and recovers from failures within the same harness. Moreover, we introduce a 100-task benchmark and five metrics assessing editing quality and delivery reliability. Under a shared tool environment, WeVisCut substantially outperforms its base model and achieves comparable performance to Claude Opus 5 while using nearly 33x fewer generated tokens per successful output minute. It also outperforms task-matched native editors on our proposed metrics.
183. 【2610.38349】Aligning Misaligned Benchmarks for Single-View Novel View Synthesis
链接：https://arxiv.org/abs/2610.38349
作者：Yang-Che Tseng,Erqun Dong,Jeong-gi Kwak,Gopal Sharma,Shakiba Kheradmand,Stephan Garbin,Hossam Isack,Kwang Moo Yi
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：dramatically enhanced recently, enhanced recently, dramatically enhanced, quantify their performance, methods
备注：
点击查看摘要
Abstract:While single-view novel view synthesis methods have been dramatically enhanced recently, how to quantify their performance has yet to be standardized. In this work, we demonstrate that an important factor lies in how each method's own reference frame is aligned with the so-called ground-truth pose. Existing methods often rely on test-time optimization of target poses for this alignment, which inherently is ill-posed due to the complexity of the optimization landscape of camera poses vs render accuracy, leading to methods being evaluated at an arbitrary local optimum. Some recognize this limitation and thus rely on 3D reconstructions of entire scenes using training and test views to align poses, which may be infeasible. In this work, instead, we propose to use traditional 2D-to-2D and 2D-to-3D image matching techniques to align 3D reconstructions. We also evaluate only within co-visible regions, between different views, which is the only place where a `correct' ground truth can be defined. Our evaluations show that with proper coordinate alignment, novel-view synthesis methods that use explicit 3D Gaussians achieve the state of the art, despite being previously reported in the literature to be outperformed by pure image-based 2D methods. We further provide a simple baseline that directly utilizes an off-the-shelf video generator and a feed-forward geometry estimator. Our simple baseline is already comparable to state-of-the-art methods specifically trained for novel-view synthesis, without any task-specific training.
184. 【2610.38346】Knowledge-structured and evidence-aware semantic self-distillation for glomerular pathology foundation model
链接：https://arxiv.org/abs/2610.38346
作者：Qiming He,Jing Li,Zihan Wang,Tian Guan,Yifei Ma,Zimo Zhao,Yanxia Wang,Hongjing Chen,Yexing Zhang,Yizhi Wang,Xinrui Chen,Xitong Ling,Minxi Ouyang,Yujie Xie,Lianghui Zhu,Yiqing Liu,Qingxia Hou,Shuyan Zhao,Xiaoqin Wang,Lili Ma,Qiang Huang,Zhiyuan Shen,Junru Cheng,Siqi Zeng,Chao He,Yonghong He,Zhe Wang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：learn rich pathological, rich pathological morphology, lack explicit concept-level, explicit concept-level supervision, models learn rich
备注：
点击查看摘要
Abstract:Visual foundation models learn rich pathological morphology but lack explicit concept-level supervision. Vision--language pretraining can provide this supervision but requires costly image--text pairs. Here, we introduce GloSAGE, a semantic self-distillation framework that uses lesion profiles predicted by a pretrained visual teacher to construct structured pathology descriptions. Reliability-guided selection identifies image--text anchors, while Knowledge-Structured Pathology Text Synthesis and Evidence-Guided Negative-pair Construction account for partially observed lesions. Image--text alignment trains projection layers while retaining the frozen visual encoder; an intermediate Semantic Transfer Layer supplies features for downstream tasks. Developed on GloMillion and evaluated across seven glomerulus- and case-level cohorts, GloSAGE supported zero-shot lesion recognition, image--text retrieval, hard-negative discrimination and concept-intervention analyses. It outperformed its source visual model in most lesion-recognition tasks and improved lesion grading, while supporting competitive case-level disease classification, injury grading and baseline renal function estimation. Ablations supported the combined contributions of knowledge-structured text and evidence-guided negatives, and the intermediate transfer layer improved mean lesion-recognition ROC-AUC relative to the complete projection output. A frozen language model conditioned on GloSAGE features generated glomerular descriptions. These findings suggest that teacher-derived lesion profiles can provide annotation-efficient semantic supervision for extending a glomerular visual foundation model.
185. 【2610.38320】SSQA: Sibling-Selective Quadtree Attention for Hierarchical Modeling in Perception Tasks
链接：https://arxiv.org/abs/2610.38320
作者：Yufan Chen,Arnav Bali,Angela Liu,Laura Zheng,Ming C. Lin
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：including aerial platforms, autonomous driving systems, navigation in robotics, including aerial, Perception tasks
备注： 8 pages, 6 figures, 4 tables. Published at ICRA 2026 (oral)
点击查看摘要
Abstract:Perception tasks for navigation in robotics, including aerial platforms such as drones and autonomous driving systems, are inherently structured. Drone-mounted cameras typically capture sky above, terrain below, and obstacles or man-made structures in between, while driving data often contains organized road layouts, lane markings, and surrounding agents. Motivated by these axis-aligned structural priors, we note that such information is typically more structured than in generic image tasks. We hypothesize that processing information in a quadtree-esque manner can not only model features effectively in a hierarchical manner, but also offers an efficient linear-time alternative to vanilla attention mechanisms, which run in quadratic time. In this paper, we propose Sibling-Selective Quadtree Attention (SSQA), which models image tokens hierarchically as a structured, full quadtree. We show analytical complexity analysis that guarantees linear-time feature modeling, in addition to empirical experiments comparing inference speeds with other popular modeling approaches, such as Mamba 2 and Quadtree Attention. Our results, benchmarked across several tasks, show that we achieve results at least as good, if not notably better, as others at a fraction of the computational costs.
186. 【2610.38319】Extremely Sparse-View Computed Tomography from 2D Projections via Pose-Aware Diffusion Priors
链接：https://arxiv.org/abs/2610.38319
作者：Linrui Dai,Jinxiu Liang,Chu Zhou,Mae Yamaguchi,Ryuya Takahashi,Imari Sato
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Extremely sparse-view computed, sparse-view computed tomography, radiation tolerance limits, Extremely sparse-view, computed tomography
备注： NeurIPS 2026 poster
点击查看摘要
Abstract:Extremely sparse-view computed tomography (xSV-CT) is critical when scanning time, motion, or radiation tolerance limits each scan to only a few projections. In many such scenarios, the same acquisition limits also constrain dataset construction, making pre-reconstructed 3D volumes from dense scans unavailable. This exposes a gap in current sparse-view approaches: Optimization-based methods, which use the sparse measurements directly but lack data-driven priors to suppress artifacts, and learning-based methods, which use strong priors but usually require densely scanned training volumes acquired outside the xSV-CT regime. We study this train-sparse / infer-sparse setting by learning a pose-aware diffusion prior from one 2D projection per training sample, then guiding diffusion sampling with measured-view fidelity, shared-volume reprojection, and projection-physics consistency for tomographic reconstruction. On two public CT datasets across 1-, 4-, and 8-view settings, the proposed method achieves the best novel-view synthesis scores and best or competitive reconstruction quality compared with baselines that use dense-volume supervision.
187. 【2610.38308】Does 3D Pose Improve Action Valuation in Football? An Empirical Study of Pose Representations and Learning Objectives
链接：https://arxiv.org/abs/2610.38308
作者：Rikuhei Umemoto,Kenjiro Ide,Soham Majumder,Ulf Brefeld,Keisuke Fujii
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：Football action valuation, valuation estimates players', estimates players' contributions, action valuation estimates, Football action
备注： Submitted to IEEE Access
点击查看摘要
Abstract:Football action valuation estimates players' contributions to future scoring. Existing multi-agent deep reinforcement learning (MADRL) models use 2D positions and velocities, omitting posture information relevant to imminent actions. We investigate whether adding 3D pose to a SARSA-based model improves observed-action values relative to attack returns: goals, conceded goals or final expected possession value. Using UEFA Champions League 2022/23 event, tracking and 3D pose data, we compare six representations (no pose, raw joints, torso and head orientation, and spatial-temporal graph convolutional network (ST-GCN) embeddings) across five learning settings and three seeds. The settings vary temporal-difference targets and action-prediction loss, with Monte Carlo return regression as a reference. Across 50 on-ball and off-ball comparisons, pose robustly improved return prediction in only three, all using torso and head orientation, while worsening it in twelve, mostly with ST-GCN. Training choices had greater effects: removing the action-prediction loss raised on-ball skill of position-only values from 0.003 to 0.040 and their correlation with expected goals from 0.23 to 0.60; the improvement in return prediction persisted at matched training length. Separately, pose improved prediction of the on-ball action 0.2 s ahead (macro AUC 0.914 to 0.934 with raw joints), while the passer's pose during execution improved pass-success prediction (AUC 0.844 to 0.856). Within this training protocol, pose informed imminent actions but rarely improved observed-action values.
188. 【2610.38303】Don't Ask the LLM: Visual Token Pruning with Spectral Structure and Semantic Relevance
链接：https://arxiv.org/abs/2610.38303
作者：Intaek Shin,Geuntaek Lim,Sungjune Park,Yukyung Choi
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：substantially increasing inference, long visual-token sequences, increasing inference costs, Large vision-language models, represent images
备注： Accepted to WACV 2027. 19 pages (10 pages main paper, 9 pages supplementary material)
点击查看摘要
Abstract:Large vision-language models (LVLMs) represent images and videos as long visual-token sequences, substantially increasing inference costs. These costs motivate visual-token pruning, whose effectiveness depends on both what signal is used for selection and when that signal becomes available. Our analysis reveals complementary limitations: visual-side [CLS] attention produces scattered, background-biased selections under aggressive compression, while stronger query-conditioned signals become available after substantial computation within the large language model (LLM). This raises a natural question: can structurally coherent, query-aware token selection be performed before consulting the LLM? We therefore propose ST-Pruner (Spectral-Textual Pruner), a training-free framework for pre-LLM visual-token selection. It combines attention-independent spectral saliency with inverted patch-text similarity through parameter-free multiplicative fusion. Across diverse LVLMs, ST-Pruner retains over 95% of full-token performance using at most 11.1% of visual tokens and achieves a 3.2x end-to-end speedup on LLaVA-NeXT. Code is available at this https URL.
189. 【2610.38296】Hidden Commitment: When Language Models Silently Pick a Side and How Steering Can Surface It
链接：https://arxiv.org/abs/2610.38296
作者：Samuel Dawit Assefa,Jae Won Cho
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Language-model answers, visible record, internal decision, decision that produced, Language-model
备注： Presented at AI4Good @ ICML 2026
点击查看摘要
Abstract:Language-model answers are often treated as a visible record of the internal decision that produced them. We test this assumption in conflicting-context question answering, where two supplied documents support incompatible answers. A model can process the disagreement and still return a one-sided answer, even under prompts that ask it to disclose conflict. In Qwen 0.8B, hidden source commitment is captured by a residual source-choice direction whose sign tracks which document the answer uses. Sparse middle-layer MLP writers feed the same coordinate, and steering the coordinate changes the chosen source. A transfer run on Gemma 3 1B preserves the source-choice readout and sparse-write bridge under LLM-audited labels. Conflict disclosure follows a different pattern, appearing weaker, later-layer, and only partially steerable. These results support a practical monitoring strategy for retrieval systems that detects silent source commitment with residual readouts and pairs conflict-surfacing prompts with disclosure steering when the model has selected a side under disagreement.
190. 【2610.38289】PulmoWorld: An Asynchronous Multimodal Medical World Model for Longitudinal Pneumonia Dynamics
链接：https://arxiv.org/abs/2610.38289
作者：Cheng Tang,Xinyu Zhang,Jiawei Chen,Minghui Zhang,Shaocong Wei,Wei Li,Yuxing Chen,Kongcheng Zhang,Chun-Mei Feng,Nanqing Dong,Xinyuan Zhu,Ziyan Huang,Junjun He,Yirong Chen,Lingtong Huang,Min Cen,Ming Hu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：organ function, describe it arrive, asynchronous state estimation, introduce PULMOWORLD, longitudinal pneumonia dynamics
备注： 49 pages, 17 figures
点击查看摘要
Abstract:Pneumonia changes across respiration, inflammation, and organ function, while the measurements that describe it arrive at different times and are often incomplete. We introduce PULMOWORLD, a world model for longitudinal pneumonia dynamics that connects asynchronous state estimation, recursive simulation, and evidence-constrained planning. History-aware filtering separates state propagation from observation-driven correction, so a measurement updates the shared belief only when it is observed. That belief supports recursive forecasts of clinical values and modality representations, and summaries of future prognosis. Cap-MBRL searches imagined continuations inside an authorized action set and replaces a frozen fallback only when a candidate meets the risk budget and a predicted-cost margin. On the pneumonia cohort MASS, recursive forecasts are competitive with task-trained sequence models, and simulated future states retain prognostic information. Trained separately on public MIMIC-IV-CXR chest radiographs and clinical trajectories, this model improves prognostic discrimination over latest-value fusion in development and is competitive on long-horizon error on the locked pneumonia panel. In five known-mechanism MASS-Syn environments, Cap-MBRL lowers selection regret relative to direct fallback execution.
191. 【2610.38284】ConvRoute-ViT: Convolutional Token Routing for Efficient Vision Transformers
链接：https://arxiv.org/abs/2610.38284
作者：Manish Das,Shiv Ram Dubey,Satish Kumar Singh
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：achieve strong image, image recognition performance, strong image recognition, self-attention increases quadratically, achieve strong
备注： 5 pages, Conference
点击查看摘要
Abstract:Vision Transformers (ViTs) achieve strong image recognition performance, but their self-attention increases quadratically with the number of input tokens and makes them computationally expensive. The standard implementation of ViT models treats all patch tokens with equal importance across every layer of the model. This uniform treatment introduces redundant computation and may reduce the model's focus on the most discriminative object regions. In this paper, we propose ConvRoute-ViT, a convolutional token routing mechanism for efficient vision transformers that performs token reduction through convolutional routing. Our model preserves the most relevant patch tokens according to class-token attention at selected layers, while the remaining tokens are spatially reorganized and compressed into a compact set of convolutional summary tokens. This design reduces the sequence length without completely removing the contextual information of the less informative tokens. By applying the routing process hierarchically, the proposed framework constructs compact and discriminative representations for deeper layers. Experiments on CIFAR-100 and TinyImageNet show that ConvRoute-ViT improves computational efficiency while maintaining classification accuracy compared with the base ViT.
192. 【2610.38278】BudSplat: Feed-forward 3D Gaussian Splatting under a Rendering Budget
链接：https://arxiv.org/abs/2610.38278
作者：Muyu Xu,Fangneng Zhan,Yu Wei,Ling Shao,Shijian Lu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Splatting has recently, Gaussian Splatting, Gaussian primitives directly, predicting Gaussian primitives, per-scene optimization
备注：
点击查看摘要
Abstract:Feed-forward 3D Gaussian Splatting has recently emerged as a promising alternative to per-scene optimization, enabling fast novel view synthesis by predicting Gaussian primitives directly from input images. However, it quickly faces a bottleneck of rendering efficiency while scaling from sparse to dense multi-view inputs, which offer more geometric constraints but produce a rapidly growing set of candidate Gaussians. Existing solutions cannot reliably distinguish redundant primitives, leading to many redundant primitives from repeatedly observed regions and increased rendering cost without much gain in visual quality. We propose BudSplat, a compact feed-forward 3DGS framework that learns to keep the most valuable Gaussians under a limited primitive count. BudSplat has two novel designs. The first is a detail-aware importance score that is extracted from multi-level visual tokens, providing an explicit cue for regions where pruning is likely to harm perceptual quality. The score can be injected into voxel fusion to preserve high-value local contributors during multi-view aggregation. The second is quality-aware Gaussian pruning that selects a compact set of primitives by jointly considering rendering contribution and detail importance. The quality-aware pruning enables local quota redistribution that helps avoid over-allocating primitives to large redundant regions. Experiments on RE10K, DL3DV, and ACID show that BudSplat achieves a favorable trade-off among rendering quality and efficiency.
193. 【2610.38265】reemapMix: Dirichlet-Controlled Multi-Image Augmentation for Probability and Ordinal Supervision
链接：https://arxiv.org/abs/2610.38265
作者：Ejafa Bassam,Konstantin Garbers,Yingsheng Geng,Dalton Jens,Jiaxu Qian,Guyue Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：improves visual recognition, Data augmentation improves, observed data, augmentation improves visual, improves visual
备注： Accepted at NeurIPS 2026. 40 pages including references, appendices, and checklist. Ejafa Bassam and Konstantin Garbers contributed equally
点击查看摘要
Abstract:Data augmentation improves visual recognition by exposing models to synthetic training examples that encourage generalization beyond the observed data. Mixed-sample methods extend this idea by combining multiple images. However, many common approaches either mix only pairs of images or rely on fixed small layouts, limiting control over source count, visible area proportions, and induced supervision. We propose TreemapMix, a distribution-controlled multi-image augmentation that samples source weights from a Dirichlet distribution and assigns them to area-proportional regions using a treemap partition. The resulting known region areas support two complementary objectives. TreemapMix with soft cross-entropy (SCE) treats areas as soft class-probability targets, while TreemapMix with a Plackett-Luce (PL) objective converts areas into ordinal supervision. On ImageNet-1K under matched training recipes, TreemapMix-SCE achieves substantially lower calibration error than evaluated mixed-sample baselines, while TreemapMix-PL achieves the strongest Top-1 accuracy. Transfer experiments on object detection and instance segmentation suggest that TreemapMix-pretrained features remain useful beyond classification. These results show that controlled multi-image composition can expose the same region-area information as either probability-matching or rank-matching supervision. Code is available at this https URL.
194. 【2610.38259】ETA-Motion: Efficient Topology-Agnostic Motion Diffusion via Sparse Joint Learning
链接：https://arxiv.org/abs/2610.38259
作者：Seoyoung Hwang,Tae-Kyun Kim
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：aims to synthesize, motion generation, Topology-agnostic motion generation, motion generation aims, motion
备注： Accepted to WACV 2027; Project page : [this https URL](https://sy-hwang.github.io/eta-motion-page/)
点击查看摘要
Abstract:Topology-agnostic motion generation aims to synthesize motions for diverse characters with arbitrary skeletons. Unlike conventional motion generation, which assumes a fixed skeletal template, this setting must handle diverse joint configurations and inconsistent joint semantics across characters. Existing methods address this challenge through dense full-joint modeling, which incurs substantial computational cost and hinders learning transferable motion patterns across diverse skeletal topologies. We present ETA-Motion, a topology-agnostic motion generation framework based on sparse joint learning. Instead of learning motion from all joints, our method uses a compact subset of joints to reduce redundancy and trains the motion generator on the resulting sparse skeleton. To facilitate correspondence across diverse topologies, we introduce semantic joint embeddings that combine textual and visual cues, together with topology-aware training objectives for preserving skeletal structure. Although ETA-Motion is trained only on sparse skeletal observations, it supports full-skeleton motion generation during inference. Experimental results demonstrate that ETA-Motion outperforms existing topology-agnostic baselines while significantly improving efficiency and generalization to unseen motions and unseen topologies.
195. 【2610.38241】MEV: A Multi-Event Video Dataset for Long-Take Generation
链接：https://arxiv.org/abs/2610.38241
作者：Peiyuan Zhu,Shaoan Xie,Yifan Shen,Wen Tian,Zijian Li,Guangyi Chen,Kun Zhang
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：short clips depicting, Current datasets, short clips, clips depicting, depicting a single
备注： Accepted to NeurIPS 2026. Project page: [this https URL](https://mev-dataset.github.io/)
点击查看摘要
Abstract:Current datasets for text-to-video generation are largely limited to short clips depicting a single event or coarse-grained captions. As a result, they provide limited supervision for modeling long-take temporal dependencies, often leading to temporally inconsistent generations when describing multiple events. In this paper, we introduce MEV, a multi-event video dataset designed for long-take video generation. MEV provides high-quality real-world videos where subjects can perform a sequence of distinct, non-overlapping events, featuring precise temporal boundaries and fine-grained procedural annotations. This event-centric discretization reduces confounding factors and enables models to support temporally localized conditioning and build long-take videos from atomic event units. Furthermore, we introduce a comprehensive annotation pipeline that segments videos into temporally grounded events, and generates structured captions at both the global level and the event level. In addition, under an event-centric evaluation protocol, we conduct multi-event generation experiments by fine-tuning on MEV with temporally grounded event conditioning, leading to improved long-take event transitions and more coherent content. We believe this work provides a publicly available resource for multi-event fine-tuning and event-centric evaluation, together with practical insights into learning from temporally structured video-text data.
196. 【2610.38231】3DCodeEdit: Benchmarking Instruction-Guided 3D Program Editing
链接：https://arxiv.org/abs/2610.38231
作者：Binjie Wang,Xinran Zhao,Shoubin Yu,Xiang Li,Silong Yong,Chenyang Yang,Tianyu Chen,Ruofei Du,He He,Chen Zhao,Tongshuang Wu
类目：Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
关键词：Language models increasingly, models increasingly create, increasingly create, precise control, editing existing programs
备注：
点击查看摘要
Abstract:Language models increasingly create 3D content through code, with practical workflows often relying on editing existing programs rather than regenerating them, for speed and precise control. Such edits require understanding what a program builds and foreseeing how a change reshapes it, yet prior benchmarks mostly evaluate generation or simple edits against a reference. In this paper, we introduce 3DCodeEdit, a benchmark of 2,592 instruction-guided editing tasks over 108 Blender programs that construct objects entirely in code. Its edits range from parameter to structural changes, style redesign, and scene composition, and are judged from renders against human-reviewed checklists. Across 17 models, the average success rate is only 37%, and even the strongest models complete fewer than 60% of structural and style edits. As models become more capable, their failures shift to geometric foresight, where the code runs but builds the wrong shape. We therefore study whether multi-turn feedback can compensate for this missing foresight. Feedback from a separate reviewer model that describes how the renders differ from the request helps most, more than maximum reasoning effort or an agent harness, and can let a 27B model surpass the single-shot rate of a 2.8T-parameter model.
197. 【2610.38223】Silhouette-Based Gaitomics for Frailty Assessment in Older Adults
链接：https://arxiv.org/abs/2610.38223
作者：Laura McDaniel,Ryan Roemmich,Crystal Szczesny,Peter Abadir,Rama Chellappa,Basudha Pal
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：video-based frailty assessment, Walking offers, Fried Frailty Phenotype, support interpretable video-based, interpretable video-based frailty
备注：
点击查看摘要
Abstract:Walking offers a window into physical function that may support interpretable video-based frailty assessment. We developed silhouette-based "gaitomics" to connect frailty predictions with measurable patterns of body configuration and movement. In 66 older adults assessed using the Fried Frailty Phenotype, standardized silhouettes and gait energy images yielded 90 descriptors of shape, temporal variation, motion, and texture. Minimum redundancy maximum relevance selection using the mutual information difference criterion prioritized informative features while limiting redundancy. Each classifier was evaluated using nested stratified cross-validation with five outer and three inner folds at the participant level, with feature ranking and feature-count selection confined to training partitions. Shapley additive explanations examined contributions to held-out predictions. Bilateral silhouette difference led the aggregate feature-selection rankings for three-class classification and both comparisons involving frail participants. Attribution profiles differed across tasks, with texture and consecutive-frame silhouette differences prominent in three-class classification. Among classifier-specific evaluations, CatBoost achieved the highest mean three-class micro-AUC of 0.647 \pm 0.085. Leading pairwise mean micro-AUCs were 0.833 \pm 0.249 for frail versus nonfrail, 0.765 \pm 0.168 for frail versus prefrail, and 0.554 \pm 0.129 for prefrail versus nonfrail classification. These findings connect model predictions to explicit gait characteristics while revealing task-dependent feature priorities and limited separation of prefrail and nonfrail participants. The framework provides an interpretable basis for further investigation, with larger independent studies needed to establish reliability and clinical relevance.
198. 【2610.38185】Complementing Text Based Moderation with Multimodal Grounding: Detecting Fraudulent Inducement Behaviors on Short Video Platforms
链接：https://arxiv.org/abs/2610.38185
作者：Xiaoyong Yu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：join private chat, private chat groups, gray industry operations, click external links, external links frequently
备注：
点击查看摘要
Abstract:Short video platforms have become a primary channel for black and gray industry operations in internet finance, where inducements such as prompts to like, follow, join private chat groups, or click external links frequently steer users toward fraud. Because such videos are short, dynamic, and often convey intent through fleeting visual cues rather than stable text, a moderation system must guarantee detection recall and localize the offending segment so reviewers inspect only the relevant clip. Existing OCR+LLM pipelines transcribe on-screen text frame by frame and classify the transcript, but exploit only textual information. Many inducements are conveyed through visual cues OCR cannot observe, and keyframe-based localization cannot recover precise temporal boundaries. We explore OCR+MLLM, which augments the text pipeline with a Multimodal Large Language Model, and compare three routes: OCR+LLM, OCR+MLLM, and a large zero-shot MLLM, along detection, latency, token consumption, and deployment complexity. OCR preserves the text signal that guarantees recall, while the MLLM jointly performs OCR-grounded discrimination and temporal grounding, localizing both textual and non-textual inducements. We construct a dataset of 2,419 internet finance short videos (66.49 hours) with 3,194 annotated events across four categories, and study reward design for RL-based grounding under class imbalance. We report two findings: auxiliary reward terms are exploited through reward hacking, and stronger localization rewards degrade class-balanced detection. We therefore design a recall-weighted F2 localization reward with class-balanced resampling. Experiments show OCR+MLLM localizes better than the keyframe OCR+LLM baseline (mIoU 26.52 vs. 14.49) and detects non-textual inducements that text-only pipelines miss, while maintaining 67.0 percent macro recall.
199. 【2610.38172】Data Amount Heterogeneity Improves Performance in Federated Learning for Neuroimaging
链接：https://arxiv.org/abs/2610.38172
作者：Paul Kurian,Tamoghna Chattopadhyay,Abhijith Shaji,Greg Ver Steeg,Paul M. Thompson,Jose Luis Ambite
类目：Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Neurons and Cognition (q-bio.NC)
关键词：Federated Learning, secure analysis, analysis of distributed, Federated, data
备注： SIPAIM 2026
点击查看摘要
Abstract:Federated Learning (FL) allows for secure analysis of distributed data without data sharing. We and others have successfully applied FL to neuroimaging tasks, such as Alzheimer's disease classification from structural magnetic resonance imaging (MRI). FL trains a deep learning model locally at each site, encrypts the local model parameters and sends them to a federation controller that aggregates the local models under homomorphic encryption, the resulting community model is sent back to the sites for further training, and the process repeats for a number of rounds. FL often achieves learning performance (e.g., accuracy) close to centralized training, but no data leaves the sites and model parameters are protected through encryption and other measures. In this paper, we show that federations composed of sites with heterogeneous amounts of data (1 large site plus small sites) perform better than homogeneous (same data per site) federations, particularly in low data regimes, while reaching performance close to centralized training. Having a big site often results in faster convergence and higher final accuracy. Heterogeneous federations better exploit sites with small amounts of data, a common situation in large consortia such as ENIGMA.
Comments:
SIPAIM 2026
Subjects:
Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Neurons and Cognition (q-bio.NC)
Cite as:
arXiv:2610.38172 [cs.LG]
(or
arXiv:2610.38172v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.38172
Focus to learn more
arXiv-issued DOI via DataCite (pending registration)</p>
200. 【2610.38162】Neighbourhood shortcuts in building damage assessment from very-high-resolution satellite imagery: diagnosis and a building-evidence plug-in
链接：https://arxiv.org/abs/2610.38162
作者：JianChong Guo,YuTing Wan,Kefan Li,AiLong Ma,YanFei Zhong
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：imagery supports disaster, supports disaster response, Building damage assessment, buildings, imagery supports
备注： 22 pages, 9 figures, 11 tables. Code: [this https URL](https://github.com/SummerTide/neighbourhood-shortcut-bda)
点击查看摘要
Abstract:Building damage assessment (BDA) from very-high-resolution imagery supports disaster response, and deep models now achieve high overall accuracy on public benchmarks. Because damage is spatially clustered, most buildings share the damage level of their neighbours, so overall scores are dominated by such buildings, whereas damaged buildings among intact ones are rare and easily overlooked. A high overall score therefore cannot tell whether a model judges the building itself or relies on its neighbourhood as a shortcut, and this study tests which of the two current models do. We stratify buildings by their agreement with their neighbours, move buildings into new neighbourhoods with their pixels unchanged, and diagnose nine models covering the main paradigms of BDA. All models miss a third to half of the isolated damaged buildings, and in non-flood events they call 73--87% of intact buildings damaged once these are moved into a damaged neighbourhood. We propose a building-evidence plug-in that adds two experts to an existing model without retraining it. One expert sees only the building, and the other sees its surroundings but is trained with seamlessly swapped neighbourhoods so that it learns not to trust them. With each of the nine models, the plug-in maintains or improves overall accuracy, reduces errors on buildings that disagree with their neighbours by 8--18%, and cuts by 40--69% the intact buildings that a damaged neighbourhood turns into damaged ones. It also raises the detection of synthetic isolated damage from 1--6% to 13--39% and carries over to optical--SAR data and to an unseen earthquake. High overall scores thus conceal a systematic reliance on the neighbourhood, which building-level evidence reduces at low cost.
201. 【2610.38158】REEL: Same Frames, Different Answers and Different Costs
链接：https://arxiv.org/abs/2610.38158
作者：Xin Liang,Jiangpeng He
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Prefix State, longer recordings, understanding keeps moving, Video understanding, Prefix
备注： Accepted at the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2027, Datasets and Evaluation Track
点击查看摘要
Abstract:Video understanding keeps moving toward longer recordings, and the longer a recording is, the more questions can be asked about it. But processing it again for every question does not scale, so many systems watch the video once, build a memory, and answer later questions from it. Yet what to remember is decided before anyone knows what will be asked, and whatever is left out cannot be recovered later. Moreover, whether the savings pay off is judged by token counts, which fall when the input gets shorter but not when the model reuses what it has already computed. We present REEL, an evaluation framework that compares three saved representations built from exactly the same observed frames, so that a change in answers can come only from what the system remembered. Text writes a short record of what happened, Visual keeps the encoder outputs, and Prefix State also keeps the language-model prefix. Each is compared with full reprocessing (Frames) by paired correctness changes, cumulative GPU time, and storage size. Across six vision-language models and two benchmarks, Text and Frames differ in correctness on 20\% to 29\% of questions, even when their mean accuracy is almost the same. By contrast, Visual and Prefix State stay much closer to Frames. With unlimited storage, Prefix State is the cheapest in total GPU time among memories that keep almost all of the answers Frames gets right, for every model on both benchmarks. The trade-off is storage: Prefix State keeps up to about 21 GiB per video, whereas Text needs only kilobytes. Reusing computation and discarding information both look like efficiency, but only the second removes evidence that later questions may need, and neither mean accuracy nor token count tells them apart. REEL makes that difference measurable for any system that answers many questions from one saved memory.
Comments:
Accepted at the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2027, Datasets and Evaluation Track
Subjects:
Computer Vision and Pattern Recognition (cs.CV)
Cite as:
arXiv:2610.38158 [cs.CV]
(or
arXiv:2610.38158v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.38158
Focus to learn more
arXiv-issued DOI via DataCite</p>
202. 【2610.38153】StepControl: Long-Horizon Human Motion Generation from Foot-Contact Schedules
链接：https://arxiv.org/abs/2610.38153
作者：Hojun Jang,Jinseok Bae,Donggeun Lim,Taesun Kim,Seungbin Jeong,Younghwan Lee,Yong Seok Kim,Young Min Kim
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：largely focused, action should occur, Controllable human motion, human motion generation, Controllable human
备注： WACV 2027, 8 main pages
点击查看摘要
Abstract:Controllable human motion generation has largely focused on specifying what action should occur or where the body should move. We study a complementary question: can full-body motion be controlled by specifying only when each foot contacts the ground? We introduce StepControl, an autoregressive diffusion model conditioned on a future binary left/right foot-contact schedule, without requiring target root and body poses, or semantic prompts. StepControl represents contact as explicit temporal events and injects them through dedicated cross-attention with classifier-free guidance. We further introduce a stochastic mixture noise schedule and warmup strategy for stable long-horizon autoregressive generation. Experiments show that StepControl follows prescribed contact timing more accurately than adapting existing spatially guided motion generators to contact constraints. It maintains effective contact control over long continuous rollouts and responds to edited schedules such as cadence changes, stop-and-go pauses, limping gaits, and bilateral jumping. Our results show that minimal foot-contact schedules provide an effective and intuitive temporal interface for controllable motion generation.
203. 【2610.38121】What Does LOL Measure? Hidden Label Shifts in Low-Light Enhancement Benchmarks
链接：https://arxiv.org/abs/2610.38121
作者：Adityan Thirumani Srinivasan
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
关键词：low-light image enhancement, leading methods ship, methods ship hand-set, supervised low-light image, hand-set test-time adjustments
备注： 11 pages, 5 figures, 8 tables
点击查看摘要
Abstract:LOL-v1 and LOLv2-Real are the standard benchmarks for supervised low-light image enhancement, and several leading methods ship hand-set test-time adjustments with their evaluation scripts. We audit both benchmarks with 18 public checkpoints of eight methods under one metric implementation that reproduces every published number to within 0.09 dB, and find that their test labels differ from their training labels. (i) The 15 LOL-v1 test labels are rendered differently: the same captures appear in LOLv2 with labels of identical luminance but 1.48x lower saturation, and LOL-v1's own training labels match the LOLv2 rendering; we confirm this on the official release. (ii) The LOLv2-Real test set is the contiguous block of images 690-789, whose labels are 14% darker than the training labels (Cliff's delta = -0.45, p 1e-12) with no comparable shift in the input statistics we examined; methods trained without any labels score up to 5.5 dB higher on test than on train. (iii) All 15 LOL-v1 test inputs reappear in LOLv2-Real, 12 in its training set. The released test-time settings are consistent with compensating for these shifts. Separating what can be calibrated on training data from what only test-set tuning finds, the test-only residual has median 0.00 dB on LOLv2-Syn, which has no shift, but 0.37 dB on LOL-v1 and 0.92 dB on LOLv2-Real (up to 3.9 dB). Under the shift, test-set checkpoint selection inflates a from-scratch model by 0.9-2.0 dB, and removing the released settings changes which method ranks first. We give concrete reporting recommendations.
204. 【2610.38109】Who Called? V33DA: A Physically Verified Multimodal Benchmark for Vocal Attribution in Zebra Finch Groups
链接：https://arxiv.org/abs/2610.38109
作者：Maris Basha,Yuhang Wang,Xiaoran Chen,Longbiao Cheng,Luca Yapura,Anja T. Zai,Mathieu Salzmann,Richard Hahnloser
类目：ound (cs.SD); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
关键词：animal group produced, Deciding which member, studying social communication, verified ground truth, physically verified ground
备注： Accepted at NeurIPS 2026 (Evaluations and Datasets Track). 43 pages, 3 figures. Dataset: [this https URL](https://huggingface.co/datasets/songbirdini/v33da) ; Code: [this https URL](https://github.com/marisbasha/v33da)
点击查看摘要
Abstract:Deciding which member of an animal group produced a vocalization and where that vocalization was produced are central to studying social communication but are rarely evaluated with physically verified ground truth. Existing resources either localize sounds without identifying the caller, or recognize individuals from isolated vocalizations without reliance on the 3D candidate geometry. We close this gap with V33DA, a dataset-centered multimodal benchmark for spatial vocal attribution in social zebra finches: it comprises 33,625 verified events coupling 5-channel audio with three camera views, calibrated 3D pose for every visible candidate, and per-bird radio telemetry across 10 individuals and three experiments. Caller identity is verified from an on-body accelerometer-derived vibration signal; this channel is withheld from benchmark models and used only by an on-body reference ceiling. Given a detected vocalization and the visible candidate birds, the task is to select the caller. We evaluate it under three regimes: session-disjoint testing on familiar birds from held-out recording dates, held-out-experiment transfer to a new cohort, and leave-one-individual-out evaluation on a bird whose vocalizations are held out from training. Reference methods expose a consistent gap between in-domain accuracy and transfer to unseen individuals: methods that can exploit familiar vocal identity perform well on known callers but degrade sharply on unseen ones, whereas candidate-aware methods preserving explicit spatial reasoning transfer substantially better. The on-body reference stays near-perfect across regimes, confirming the internal consistency of the accelerometer-verified labels. We release V33DA with fixed protocols and code for the evaluated methods, and V33DA++, an unevaluated companion set of verified overlapping-call events and +/-2 s context windows.
205. 【2610.38103】he Prestige: Benchmarking Cognitive Visual Reasoning using Magic Tricks
链接：https://arxiv.org/abs/2610.38103
作者：Shuo Wen,Beixi Du,Edwin Meriaux,Junming Shi,Chloe Si,David Meger,Doina Precup,Gregory Dudek
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：explain unexpected outcomes, probes cognitive visual, deliberate misdirection challenges, professional magic performances, introduce The Prestige
备注： 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Track on Evaluations and Datasets
点击查看摘要
Abstract:We introduce The Prestige, a video understanding benchmark that probes cognitive visual reasoning through professional magic performances, where deliberate misdirection challenges models to track events and explain unexpected outcomes. The benchmark comprises open-ended questions ranging from temporal and causal understanding to robustness against adversarial questions. Evaluation of leading open-source and proprietary VLMs reveals a substantial gap between even the strongest model and human performance. We also find systematic discrepancies between human and automated grading and identify recurring model failure modes. Ablations further show that models can select correct multiple-choice answers even when they fail to generate correct explanations independently, motivating further investigations into question formats. Together, these findings highlight limitations in both VLM reasoning and its evaluation, underscoring the need for more reliable video understanding models and more rigorous benchmarks.
206. 【2610.38086】From Next Day Wildfire to Next Days Wildfire: A New Multi-Continent Dataset of Active-Fire Spread for Sequence-to-Sequence Forecasting
链接：https://arxiv.org/abs/2610.38086
作者：Jose Guillen,Renzo Zanca,Valentin Barriere
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：enabled data-driven forecasting, satellite time series, suffer from limitations, size restricts, burn for weeks
备注： Accepted at WACV 2027. 19 pages including supplementary material, 12 figures, 13 tables
点击查看摘要
Abstract:Benchmarks such as WildfireSpreadTS (WSTS) have enabled data-driven forecasting of wildfire spread from multimodal satellite time series, but they suffer from limitations: size restricts them to large fires that burn for weeks, and they are from similar environments. We introduce AFSeq, a multimodal dataset of 1,194 recent fire events (2023 to 2026) from 8 different countries across three regions, dominated by the short, small fires that existing benchmarks underrepresent. Our dataset is the largest and most diverse to date for time-series-based wildfire spread prediction. We also introduce UTAE-AR, an extension of a classical U-Net with Temporal Auto Encoder, enhanced with recurrent modules before the decoder. This allows the model to not only predict the next day but the whole sequence in an autoregressive way. Moreover, our model can feed its own predicted fire mask back one day at a time, producing the day-by-day forecasts that single-step models cannot. Our results show that UTAE-AR surpasses state-of-the-art on both WSTS, WSTS+ and AFSeq for next-day accuracy while having 21.5% of their parameters, and being able to predict confidently at a longer horizon.
207. 【2610.38080】Efficacy of Monocular Depth Estimation in Lunar Environments
链接：https://arxiv.org/abs/2610.38080
作者：Michael W. Lanighan,Kenji Brameld,Seth Gee
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
关键词：demonstrated highly accurate, in-distribution terrestrial domains, monocular depth estimation, Recent advances, NASA Ames Polar
备注： Accepted at the International Conference on Space Robotics (iSpaRo) 2026
点击查看摘要
Abstract:Recent advances in monocular depth estimation have demonstrated highly accurate 3D reconstructions within in-distribution terrestrial domains. This study investigates the efficacy of several state-of-the-art approaches to reconstruct out-of-distribution lunar environments using the NASA Ames Polar Optical Lunar Analog Reconstruction (POLAR) stereo dataset and Apollo 17 surface imagery. By leveraging these datasets, we establish a robust analog-to-field validation pipeline that evaluates quantitative performance against ground-truth physical analogs and qualitative performance against historical field data. Our results indicate that select state-of-the-art techniques can successfully generalize to produce high-fidelity relative 3D reconstructions of the unstructured lunar surface.
208. 【2610.38065】Vision Tomography: Towards Portable 3D Bidirectional Face-Skull Inference from Visual Data
链接：https://arxiv.org/abs/2610.38065
作者：Zidu Wang,Meng Xu,Miao Xu,Hengyuan Ma,Jiankuo Zhao,Xutao Li,Zhen Lei,Xiangyu Zhu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Traditional computed tomography, Traditional computed, Vision Tomography, skull geometry, fixed in location
备注： Submitted to TIP, under review
点击查看摘要
Abstract:Traditional computed tomography (CT) can capture paired 3D face and skull geometry, but it remains costly, fixed in location, and exposes patients to radiation. For applications such as remote diagnostics, digital healthcare, forensic identification, missing-person investigation, and archaeological analysis, there is an urgent need for a portable, safe, and accessible solution to capture 3D skull geometry. In this work, we introduce Vision Tomography (VT), a concept and framework capable of reconstructing paired 3D face-skull geometry from various visual modalities, and also enabling joint manipulation of 3D facial and skull geometry. To achieve this, we first construct a dataset of over 200 samples, where each case includes a CT-based skull, a CT-based face, and a high-fidelity textured face scan. Secondly, we propose a novel dense ray matching registration method that ensures topological consistency across face, skull, and their tissue correspondences. Building on this registration, we introduce the 3D Bidirectional Face-Skull Morphable Model (BFSM) as the foundation of Vision Tomography, which enables shape inference between the face and skull through a shared coefficient space, while also modeling tissue thickness variation to support one-to-many facial reconstructions from the same skull, reflecting individual facial changes such as age or body fat. Finally, we demonstrate the effectiveness of Vision Tomography across various forensic and medical applications, including paired 3D face-skull reconstruction from images or sketches and predicting surgical outcomes with joint face-skull editing. Extensive experiments confirm the robustness and accuracy of our method. Vision Tomography is available at this https URL
209. 【2610.38045】Classifying Urban Locations by Cyclist Facility Usage and Directional Alignment: A Validated Camera-Based Framework from Multi-Site Observations
链接：https://arxiv.org/abs/2610.38045
作者：Ao He,Mohit Gupta,Deborah Salon,Ricardo Eiris,Vivian W. H. Wong
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Agencies invest, Toggle, Toggle Hugging Face, invest in bicycle, bicycle facilities
备注： 19 pages, 6 figures, 2 tables. Accepted for presentation at the 106th Transportation Research Board (TRB) Annual Meeting, Washington, D.C., January 10-14, 2027
点击查看摘要
Abstract:Objectives: Agencies invest in bicycle facilities but lack tools to assess whether cyclists use them as designed. We examine whether usage patterns follow facility configurations and whether higher bike lane usage is associated with less wrong-way riding. Methods: Motion-triggered cameras covered 19 deployments at 17 locations in Tempe, Arizona, in 2023. Manual review identified 1,963 cycling events; excluding 20 crossings left 1,943 labeled events for analysis. An automated pipeline on the same captures counted 92.9 percent of the full manual total (per-deployment r = 0.99), and every automated wrong-way call was visually verified. Deployments were classified using thresholds on facility usage shares. Findings: Deployments were bike-lane dominant (9), sidewalk dominant (6), or mixed (4). Cyclists in bike lanes rode against the flow in 4.6 percent of events; on sidewalks, riders opposed the near-side traffic flow in 49.6 percent. Across deployments, bike lane share and wrong-way rate were negatively correlated (Spearman rho = -0.87, p 0.001, n = 16). An independent volunteer count in the same city showed a similar sidewalk-direction pattern (rho = 0.79). Novelty: The study combines multi-site still-camera observations with a fully hand-verified benchmark to characterize facility use and travel direction. Practical Applications: The workflow offers planners a low-cost screen for locations with high sidewalk cycling and travel against the near-side traffic flow, supporting more detailed site investigation.
Comments:
19 pages, 6 figures, 2 tables. Accepted for presentation at the 106th Transportation Research Board (TRB) Annual Meeting, Washington, D.C., January 10-14, 2027
Subjects:
Computer Vision and Pattern Recognition (cs.CV)
Cite as:
arXiv:2610.38045 [cs.CV]
(or
arXiv:2610.38045v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.38045
Focus to learn more
arXiv-issued DOI via DataCite
Submission history From: Ao He [view email] [v1]
Tue, 27 Oct 2026 22:59:17 UTC (8,768 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Classifying Urban Locations by Cyclist Facility Usage and Directional Alignment: A Validated Camera-Based Framework from Multi-Site Observations, by Ao He and 4 other authorsView PDFHTML (experimental)TeX Source
view license
Additional Features
Audio Summary
Current browse context:
cs.CV
prev
|
next
new
|
recent
| 2026-10
Change to browse by:
cs
References Citations
SciXGoogle Scholar
Semantic Scholar
export BibTeX citation
Loading…
BibTeX formatted citation
loading…
Data provided by:
Bookmark
checked="checked"class=“labs-tab-input”>
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)
mathjaxToggle();
We gratefully acknowledge support from our major funders,
member institutions, ,
and all contributors.
About
Help
Contact
Subscribe
Copyright
Privacy
Accessibility
Operational Status (opens in new tab)
210. 【2610.38042】Can Event Detectors Leave the Broadcast Camera? Evidence from Tennis
链接：https://arxiv.org/abs/2610.38042
作者：David Soeiro-Vuong,Andre-Louis Rochet
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：statistics are derived, easy to confuse, tennis statistics, monocular video, phone and court-level
备注： 16 pages, 4 figures, 5 tables. Benchmark, evaluator and frozen predictions: [this https URL](https://github.com/David-dsv/tennis-bounce-hit-benchmark)
点击查看摘要
Abstract:Ball bounces and racket hits are the events from which tennis statistics are derived, and on monocular video the two are easy to confuse. We release a benchmark of 345 bounce and hit annotations at frame level on eight clips that mix broadcast, fixed elevated, phone and court-level viewpoints, with a frozen development and test split, a confusion-aware matcher, clip-level bootstrap intervals and a blind second annotation of three clips. On it we score supervised precise spotters with published broadcast weights, two zero-shot vision-language models, and RallyArb, an arbitration layer over a ball track and player poses that fits no parameter on event labels. The supervised spotters score near-perfectly on broadcast footage and almost nothing on phone and court-level footage, where the pooled bounce F1 of E2E-Spot is 0.130. RallyArb keeps a bounce F1 above 0.8 on the three phone and court-level clips, two of which were used to tune it, and finds 8 of 11 bounces on the held-out one, but it places hits too far from the contact frame to match human timing. The vision-language models detect events on every viewpoint and miss the contact frame by about 0.1 s. Fine-tuned leave-one-clip-out from its broadcast weights, the spotter reaches about 0.8 bounce F1 on two phone and court-level clips and stays at 0.2 on a grazing phone angle. The annotations, the evaluator and the frozen predictions of every scored row are released.
211. 【2610.38013】Where Can We Fool the Model's Mind? Response-Guided Semantic Probing and Adaptive Routing for Transferable Attacks on Black-Box MLLMs
链接：https://arxiv.org/abs/2610.38013
作者：Ahmed Abdelnaby,Mohamed Elmahallawy
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Large Language Models, Multimodal Large Language, Language Models, Large Language, ground language
备注：
点击查看摘要
Abstract:Multimodal Large Language Models (MLLMs) ground language in visual evidence, yet their semantic reasoning can be exploited through transferable black-box adversarial attacks. Existing attacks typically allocate perturbations using surrogate-model saliency, which may not reflect how an unseen MLLM semantically interprets or relies on different image regions. We introduce VISTA (Vulnerability-Informed Semantic Transfer Attacks), a black-box framework that identifies semantically vulnerable visual regions from an MLLM's responses without access to gradients, logits, parameters, or internal representations. VISTA supports two attack settings: (i) VISTA-Proxy, a pure-transfer setting that probes an open-source MLLM and transfers the resulting adversarial examples to unseen commercial MLLMs without querying them; and (ii) VISTA-Victim, a query-assisted setting that probes the target MLLM to capture model-specific semantic vulnerabilities. VISTA builds an Uncertainty-Based Multimodal Behavior (UBMB) map from response drift, conflict, misleading, and semantic shifts, then uses entropy-adaptive Sinkhorn routing and a surrogate ensemble to focus perturbations on vulnerable regions. Across GPT-5, Gemini-3.5-Flash, and Claude Sonnet 4.6, VISTA achieves up to 97 % attack success rate (ASR) while maintaining a Peak Signal-to-Noise Ratio (PSNR) above 30 dB. Notably, VISTA-Proxy achieves up to 93 % ASR without querying the final victim, demonstrating that response-derived semantic vulnerability can enable effective transfer across heterogeneous MLLMs.
212. 【2610.38007】Frequency-Adaptive Gaussian Grids for Efficient Image Representation
链接：https://arxiv.org/abs/2610.38007
作者：Anatolii Evdokimov,Stavros Diolatzis,Bingxuan Li,Akshay Jindal,Patsorn Sangkloy,Qi Sun
类目：Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
关键词：increasingly higher-resolution images, powerful tool, tool to mitigate, mitigate the demand, demand for increasingly
备注：
点击查看摘要
Abstract:Neural representation has emerged as a powerful tool to mitigate the demand for increasingly higher-resolution images. However, a fundamental trade-off remains between structured (e.g., feature grids, which offer spatial consistency) and unstructured (e.g., 2D Gaussians, which adaptively capture local variances) representations. This choice significantly affects how well an algorithm represents an image. Here, we present Gaussian-Grid, a hybrid neural representation that adaptively integrates structured and unstructured primitives. By dynamically partitioning computation between 2D Gaussian distributions and their attached multi-resolution feature grids based on the local frequency content, Gaussian-Grid automatically optimizes its architecture to match the underlying complexity of the target signal. We evaluate our approach using high-resolution natural images and demonstrate its applications as GPU-native dataset storage, showing $1.24$-$2.14\times$ decoding speed improvement and PCIe data traffic reduction of $23\times$-$176\times$ over JPEG + ZIP data archiving. The results show that Gaussian-Grid maintains small memory footprints while achieving high reconstruction fidelity across a variety of datasets.
213. 【2610.37983】Hierarchical Diffusion for Camera-controlled Autoregressive Video Generation
链接：https://arxiv.org/abs/2610.37983
作者：Kaihua Chen,Yiming Gong,Deva Ramanan,Tarasha Khurana
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：single image, address the task, Abstract, generation, time
备注： Project page: [this https URL](https://kaihua-chen.github.io/wanar_fixar/)
点击查看摘要
Abstract:We address the task of long-video, camera-controlled generation from a single image. The dominant autoregressive paradigm suffers from two key limitations: (1) error accumulation over time, as distant frames are conditioned on previously generated outputs rather than ground truth, leading to drift; and (2) ambiguity in how to incorporate camera control. Standard solutions to both--such as self-forcing and conditioning on explicit 6-DOF poses--either incur significant computational overhead or produce drifting scene geometries over time. We propose a simple and lightweight framework that combines autoregressive generation (WanAR) with hierarchical long-context denoising (FixAR). To enforce scene geometry remains consistent with input cameras, we draw from feed-forward novel-view synthesis and reformulate video generation as a warping-and-inpainting problem. To mitigate drift, we introduce a hierarchical denoising strategy in which different stages condition on different temporal histories. Specifically, we first generate frames conditioned on the immediate past, and then refine them using a second denoiser conditioned on a temporally subsampled long-range history to enforce global consistency. This fine-to-coarse scheme enables each generated frame to effectively condition on the full temporal context without incurring a prohibitive cost. Experiments on RealEstate10k and DL3DV demonstrate that WanAR+FixAR outperforms prior methods for camera-controlled long-video generation, producing results that are both temporally stable and consistent with input camera trajectories.
214. 【2610.37965】Meta-Initialization of Implicit Neural Representations to Accelerate Deformable Medical Image Registration
链接：https://arxiv.org/abs/2610.37965
作者：Marat O. Zinovev,Dmitry V. Sorokin
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：Implicit neural representation, deformable medical image, medical image registration, image registration require, Implicit neural
备注： 14 pages, 5 figures, 2 tables. English-language author version of a paper presented at GraphiCon 2026, Saint Petersburg, Russia, 22-24 September 2026. Russian-language proceedings forthcoming. Related work updated for this version
点击查看摘要
Abstract:Implicit neural representation (INR) methods for deformable medical image registration require the network parameters to be optimized for each new image pair, making adaptation speed a practical limitation. This work investigates whether early adaptation can be accelerated using an initialization learned from a collection of previously solved registration tasks. We examine two schemes: Reptile meta-initialization and STRAINER transfer of the shared component of a coordinate network. KAN-IDIR is used as the base INR displacement-field model. Experiments are conducted on OASIS-1, with 364 consecutive pairs used for training and 49 pairs for test-time adaptation. Registration quality is evaluated using mean Dice and HD95, together with the fraction of voxels with a negative Jacobian determinant at iteration 1500. In the configurations studied, both schemes yield higher Dice and lower HD95 at early iterations than random initialization. Reptile numerically outperforms STRAINER under a small iteration budget, although their pretraining budgets are not matched. As the number of iterations increases, the differences between the schemes shrink. STRAINER pretraining takes 39-40 minutes, compared with 44.98 hours for Reptile.
215. 【2610.37937】You Don't Need Aligned Representations: Knowledge Distillation via Random Prototype Spaces
链接：https://arxiv.org/abs/2610.37937
作者：Hamza Etcibasi,Ramazan Gokberk Cinbis
类目：Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：widely held assumption, Prototype Knowledge Distillation, explicitly aligned feature, knowledge distillation, Random Prototype Knowledge
备注： To appear at NeurIPS 2026. 10 pages main text plus appendices. Code: [this https URL](https://github.com/HamzaEtcibasi/RPKD)
点击查看摘要
Abstract:A widely held assumption in knowledge distillation is that teacher and student representations must occupy a compatible or explicitly aligned feature space. We propose RPKD (Random Prototype Knowledge Distillation), which discards this assumption by comparing both networks against a shared bank of randomly initialized, frozen prototype vectors. Alignment is therefore carried out in an external reference space rather than in the teacher's native feature coordinates: the teacher side is projected by fixed random maps that carry no trainable parameters, and only the student adapts. Two branches handle the transfer: a logit branch aligning global prototype similarity distributions and a feature branch matching spatially-resolved prototype activation maps. Both operate within the same fixed prototype space, which can be interpreted as a random feature embedding that approximately preserves inter-sample similarity structure, so the prototype bank and the distillation objective stay unchanged across teacher-student pairs while only the input projections are instantiated per architecture. In low-category settings, decoupling the prototype vocabulary from the task label space yields richer supervision than the label space alone, an advantage absent from logit-based methods, whose supervisory signal collapses as class count shrinks. This advantage is especially relevant in real-world scenarios where the label space is typically constrained. Extensive experiments demonstrate the effectiveness of RPKD, achieving a maximum gain of +7.98% over OFA on CIFAR-100, +12.44% on ImageNet-100, and +1.28% on ImageNet-1K. Source code is available at this https URL
216. 【2610.37918】Venus-DefakerOne2: A Vision-Forensics Foundation Model with Broader Coverage
链接：https://arxiv.org/abs/2610.37918
作者：Chenfan Qu,Yachao Liang,Mingqi Fang,Jiangwei Xie,Zhenming Wang,Baoyue Zhang,Changjiang Jiang,Zheng Chang,Fanli Jin,Xuekang Zhu,Kaiwen Feng,Junchi Li,Weihong Zhang,Haoyue Wang,Chenfeng Zhang,Yinan Xia,Fangming Cui,Dong Fan,Jingjing Liu,Tianyi Zhang,Luobin Chen,Ji-Zhe Zhou,Sheng Li,Wei Wang,Bin Li,Ran He,Xinpeng Zhang,Linna Zhou,Jian Liu,Weiqiang Wang
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Real-world forgery, inherently heterogeneous, forgery is inherently, digital manipulation, Real-world
备注： Technical Report
点击查看摘要
Abstract:Real-world forgery is inherently heterogeneous: it manifests not only as digital manipulation but also as physical spoofing, and is carried not only by images but also by videos. Despite substantial progress, a unified anti-forgery foundation model covering such a broad threat spectrum remains absent. In this paper, we present Venus-DefakerOne2, which substantially extends the forensic capability boundary along both the forgery-form and media-type dimensions. Trained on tens of millions of open-source and in-house samples, it incorporates two key designs. Tide-Style-Tuning alternates LoRA and full-parameter fine-tuning to balance computational efficiency, generalization, and hard-sample fitting while mitigating catastrophic forgetting. Entropy-based Test Adaptation determines the optimal prediction modality by analyzing prediction entropy. Venus-DefakerOne2 achieves new state-of-the-art performance across a comprehensive suite of benchmarks, including DeepFake faces (96.7%), AI-generated images (93.3%), tampered documents (87.6%), manipulated natural images (89.6%), physical spoofs (93.1%), and AI-generated videos (96.0%). It significantly outperforms both specialized expert models and its predecessor, DefakerOne, establishing itself as a powerful and versatile foundation model for binary-decision forensic tasks.
217. 【2610.37917】What Actually Matters? A Paired Evaluation of Inference Settings for $π_{0.5}$ on BEHAVIOR-1K
链接：https://arxiv.org/abs/2610.37917
作者：Ruben Hammele,Samira Huber,Mohammad Nazeri,Amirreza Payandeh,Alexandyr T. Card,Aaron Nguyen,Anuj Pokhrel,Daeun Song,Xuesu Xiao,Sören Pirk
类目：Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：general-purpose household robotics, household robotics, offer a promising, promising paradigm, paradigm for general-purpose
备注：
点击查看摘要
Abstract:Vision-Language-Action (VLA) models offer a promising paradigm for general-purpose household robotics. However, inference and data-collection strategies that have been deployed in smaller-scale settings do not necessarily generalize to large-scale, diverse task suites. To address this gap, we conduct a systematic, paired evaluation of previously proposed inference and data-collection strategies across all 100 tasks. Our evaluation uses a fine-tuned $\pi_{0.5}$ model based on the 2025 BEHAVIOR Challenge winner, which achieves a Q-score of 0.39 on the public test set. These results show that proposed strategies do not transfer uniformly to new environments and tasks. Project page, videos, and detailed analysis: this https URL
218. 【2610.37882】FYC: FPR-Disparity Weight Perturbation and Youden Calibration for Fair Deepfake Detection
链接：https://arxiv.org/abs/2610.37882
作者：Uzoamaka Ezeakunne,Taiwo Olabamipe,Deevesh Jain,Xiuwen Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Deepfake detectors routinely, false-positive rates differ, routinely achieve high, achieve high aggregate, substantial demographic disparities
备注： 9 pages, 4 figures
点击查看摘要
Abstract:Deepfake detectors routinely achieve high aggregate accuracy while exhibiting substantial demographic disparities in their errors: false-positive rates differ by double digits across gender and skin tone. Most existing remedies require either full retraining with a fairness-aware objective or demographic-annotated training data, both impractical when only a pretrained checkpoint is available, while label-free post-hoc calibrators leave the detector's weights unchanged. We introduce FYC (FPR-disparity Weight Perturbation and Youden Calibration), a two-stage post-hoc method that perturbs a detector's weights under an FPR-disparity criterion, then calibrates a separate decision threshold for each demographic group - requiring no architectural changes and no full retraining. Evaluated against every detector benchmarked on AI-Face v2 across six intersectional gender x skin-tone groups, FYC outperforms the field on 18 of 21 reported fairness and utility metrics, including purpose-built fairness architectures, at a fraction of the training cost. We further identify why the choice of fairness criterion is critical: the original Demographic Parity objective used in prior weight-perturbation work is architecture-dependent and unpredictable, improving two of four evaluated architectures, leaving a third largely unchanged, and substantially degrading the fourth (PG-FDD), with no way to anticipate the outcome for a given checkpoint in advance. Conditioning the same objective on real images only - our FPR-disparity criterion - eliminates this unpredictability, improving fairness and accuracy jointly across all four architectures with no exceptions.
219. 【2610.37881】Intersectional Invariant Learning for Fair Deepfake Detection
链接：https://arxiv.org/abs/2610.37881
作者：Uzoamaka Ezeakunne,Taiwo Olabamipe,Deevesh Jain,Xiuwen Liu
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Deepfake detection models, experience substantially higher, Deepfake detection, substantially higher false-positive, high overall accuracy
备注： 4 pages including appendices, 7 figures, 8 tables
点击查看摘要
Abstract:Deepfake detection models achieve high overall accuracy but exhibit large performance disparities across demographic groups, causing some groups to experience substantially higher false-positive rates than others. Existing fairness-aware methods address this by constraining model outputs to equalize group predictions, but do not tackle the underlying feature-level bias. As a result, models may still rely on demographic characteristics, such as skin texture, as indirect cues for detecting manipulated content. To address this problem, we propose Intersectional Invariant Learning (IIL), a framework that improves feature learning rather than only adjusting model outputs. IIL alternates between identifying the most challenging demographic partitions and learning features that remain consistent across those partitions. The framework also includes a false-positive-rate fairness objective and group-balanced sampling across six gender x skin-tone intersectional groups. We evaluate IIL on a large-scale fairness benchmark containing over one million images, 10 baseline detectors, and 21 fairness and utility metrics. Results show that IIL achieves Pareto-optimal performance, simultaneously attaining an AUC above 99%, a false-positive rate under 2.5%, and the lowest intersectional fairness disparity (FMEO) among all compared methods. IIL outperforms prior fairness-aware methods on the large majority of the 21 evaluation metrics, while reducing both false-positive rates and intersectional fairness disparities by several-fold compared with the strongest competing method.
220. 【2610.37874】PairCLIP-SWA: Similarity-Guided Balancing and Stochastic Weight Averaging for Generalized Deepfake Image Detection with OpenCLIP
链接：https://arxiv.org/abs/2610.37874
作者：Md. Nakib,Sudipto Sarkar,A. J. A. Jubayer Talukder,Mohammad Ariful Haque
类目：Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
关键词：Deepfake detectors, choice is rarely, single checkpoint, validation accuracy, Deepfake
备注： 15 pages, 8 figures, 12 tables. Code and experiment outputs: [this https URL](https://github.com/DarkJ0Y/PairCLIP-SWA)
点击查看摘要
Abstract:Deepfake detectors are almost always deployed as the single checkpoint that scored best on a validation split, yet the reliability of that choice is rarely measured. We measure it with PairCLIP-SWA, a single-backbone detector that adapts an OpenCLIP ViT-L/14 visual tower by progressive block unfreezing on a class-balanced corpus built by CLIP-similarity-guided one-to-one subsampling, and averages the fine-tuning trajectory in weight space. Trained in-domain on the IEEE SP Cup 2025 (DF-Wild-Cup) benchmark under the same protocol as the strongest published three-backbone ensemble, the weight-averaged detector reaches 97.04% accuracy, 98.64% AU-ROC and a 4.07% equal error rate, against 94.46%, 97.60% and 5.00%. Its training log shows why averaging matters: over the averaged epochs the non-averaged checkpoint swings across 5.34 points of validation accuracy while its validation loss more than doubles, whereas the average is never worse and is up to 5.99 points better. On an OpenForensics-derived benchmark the same instability becomes a misranking: the checkpoint that validation accuracy selects (98.43%) scores 88.85% on a disjoint 10,905-image test partition, rejecting 21.1% of genuine photographs, while the terminal average, ranked last on validation, scores 94.76%. A study of twenty-two in-house configurations on the identical DF-Wild-Cup split, including preprocessing ablations, shows that CLIP pre-training rather than ensembling drives accuracy, that secondary face cropping costs about 34 points, and that frequency-domain inputs transfer poorly. We recommend per-class selection criteria measured off the early-stopping split.
221. 【2610.37867】What Constitutes Food Similarity? Visual Representations and Retrieval in Indian Cuisine
链接：https://arxiv.org/abs/2610.37867
作者：Saransh Kumar Gupta,Akshadha Sharma,Lipika Dey,Partha Pratim Das,Ramesh Jain
类目：Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
关键词：food computing tasks, computing tasks, typically evaluated, Food, identity preservation
备注： 17 pages, 7 figures, 4 tables, 31 references
点击查看摘要
Abstract:Food similarity underlies retrieval, recommendation, substitution, and other food computing tasks, yet food representations are typically evaluated by identity preservation. We ask what happens to food similarity beyond identity: how different visual representations organize foods into relational neighborhoods, what culinary relationships these neighborhoods reveal or obscure, and how consistently these structures emerge across representations. Indian cuisine provides a challenging setting in which visually similar foods may have different culinary identities, while foods within a culinary family can vary substantially in appearance. Using an expanded Khana dataset and seven pretrained visual representations spanning convolutional, transformer-based, self-supervised, and image-text aligned models, we evaluate image-to-image retrieval and image-to-dish neighborhoods, examining culinary-family structure, contrast cases, and cross-representation agreement. We also aggregate image representations to compare dish-level visual similarity across models and against recipe-derived similarity for the same dishes. Image-level neighborhoods vary substantially across representations, including similarities across culinary-family boundaries. At the dish level, aggregation reveals common visual structure despite local variation, but this structure corresponds less strongly with recipe-derived similarity. These findings show that recognition performance alone cannot characterize relational structure: food similarity depends on representation, modality, and abstraction level rather than being a fixed property of the food collection. Understanding these structures is therefore a prerequisite to combining multiple food information modalities into a coherent notion of similarity.
222. 【2610.37860】Effects of Deep Learning Architecture and Pretraining on Automated Vocal Fold Segmentation in Ex Vivo High-Speed Phonation Images
链接：https://arxiv.org/abs/2610.37860
作者：Varun Vinayak Kalaiarasan(1),Charles Farbos de Luzan(1 and 2),Khalid Malik(2),Ephraim Gutmark(3),Liran Oren(1 and 2) ((1) Department of Biomedical Engineering, University of Cincinnati, (2) Department of Otolaryngology, University of Cincinnati, (3) Department of Aerospace Engineering Engineering Mechanics, University of Cincinnati)
类目：Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV); Fluid Dynamics (physics.flu-dyn)
关键词：medial surface dynamics, fold medial surface, experiments quantifies vocal, vocal fold segmentation, quantifies vocal fold
备注： 37 pages, 8 figures, 7 tables
点击查看摘要
Abstract:Two-dimensional (2D) intraglottal particle image velocimetry (PIV) of ex vivo phonation experiments quantifies vocal fold medial surface dynamics and the intraglottal flow field. However, it relies on manual segmentation because residual glare, low tissue-to-flow contrast, and tracer-particle speckle obscure the vocal fold boundaries. This study assessed how deep learning architecture and ImageNet pretraining affect automated vocal fold segmentation in these high-speed images. Four established architectures from three categories were benchmarked: the convolutional neural network (CNN)-based UNet and UNet++, the hybrid CNN-transformer-based TransUNet, and the pure transformer-based Swin-UNet, the latter two trained from scratch and from ImageNet-pretrained weights. Using 16,344 expert-annotated images from 227 phonation trials of excised canine larynges, the six pipelines were trained with five seeds under a standardized protocol and compared on 50 held-out test trials using trial-level Wilcoxon signed-rank tests with Holm correction. Mean Dice similarity coefficient (DSC) spanned only 0.9734 to 0.9764, so the pipelines differed mainly in boundary fidelity. Pretraining significantly improved the boundary F1 score (BF1) and 95th percentile Hausdorff distance (HD95) of the TransUNet and Swin-UNet, about four times more for the Swin-UNet. Once pretrained, the two performed similarly, although the Swin-UNet has only about a quarter of the trainable parameters. Without pretraining, the TransUNet had a significantly higher BF1 and a lower HD95 than both CNN-based pipelines and about half their share of images with HD95 50 pixels. To our knowledge, this is the first study to apply deep learning to vocal fold segmentation in 2D intraglottal PIV, and it offers a reference framework for segmenting experimental flow-visualization images with noisy optical conditions or highly deformable objects.
223. 【2610.37844】Convex Potential Learning for Optimal Transport Between Medical Images
链接：https://arxiv.org/abs/2610.37844
作者：Hongyu Kan,Ivan Medri,Natasha Ironside,Kristofor Pas,Richard Corp,Shinjini Kundu,Gustavo Kunde Rohde
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：Computing optimal transport, Computing optimal, optimal transport maps, plays a fundamental, fundamental role
备注：
点击查看摘要
Abstract:Computing optimal transport maps between medical images plays a fundamental role in morphological analysis. However, existing methods are either difficult to scale to large 3D medical images or rely on regularized formulations that provide only approximate transport maps. To improve the accuracy of downstream medical image analysis, we propose a novel approach for estimating optimal transport maps by parameterizing the convex potential function. By construction, the resulting transport map is guaranteed to be curl-free, consistent with Brenier's theorem. We evaluate the proposed method on both 2D and 3D medical imaging datasets. Experimental results demonstrate that our approach achieves lower transport estimation errors and produces higher-quality image reconstructions than existing methods.
224. 【2610.37823】Learning Without Clocks: Partition-Invariant Neural Network
链接：https://arxiv.org/abs/2610.37823
作者：Han Wang,Jingyuan Bao,Juntao Wu,Saiao Zhou,Yuman Nie
类目：Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
关键词：similar sensors emit, sensors emit asynchronous, emit asynchronous events, spiking cochleas, frames or packets
备注：
点击查看摘要
Abstract:Event cameras, spiking cochleas and similar sensors emit asynchronous events, yet most learning systems first bin them into frames or packets, so the output depends on how the stream was sliced. Networks whose state is updated exactly at every event and evolves in closed form between events are invariant to this slicing. We ask what more makes the output (i) independent of the processing order of simultaneous events and (ii) continuous in the time stamps, even when two events cross in time, while (iii) staying sensitive to temporal order. For arbitrary nonlinear event models we prove that (i) and (ii) are each equivalent to pairwise commutation of the event updates, and that once updates commute, all order dependence comes from their non-commutation with the inter-event dynamics and vanishes linearly in the time lag. This identifies a class of time-mediated non-commutative (TM) models and yields a resolution-robustness bound for any model. We construct the minimal TM layer, a learnable, provably Lipschitz generalization of the Hassenstein-Reichardt correlator; prove that no single additive layer can uniformly represent ordered pair interactions, whereas a three-dimensional TM state can; show that each TM layer adds two levels of temporal nesting versus one for additive layers; and give a single-layer higher-order construction (TM-k) and a stacking rule preserving all three properties. We instantiate these results as the Clock-Free Event (CFE) layer, linear in the number of events, give a topology-free variant for temporal point processes with closed-form likelihoods, and specify a preregistered clock-robustness protocol with a physical-time threat model for spike-retiming attacks. In preliminary development-set experiments, the TM state beats an otherwise identical additive state by 3.5-9.1 points on DVS128 Gesture and by 24.5 and 30.6 points on SHD and Speech Commands.
225. 【2610.37818】GOFish: Full Field-of-View Gaussian Splatting and Mesh Extraction
链接：https://arxiv.org/abs/2610.37818
作者：Allen Tu,Meher Gitika Karumuri,Shifan Zhu,Hailing Wang,Yen-Liang Lin,David Held,Matthias Zwicker,Tom Goldstein
类目：Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR); Robotics (cs.RO)
关键词：fisheye Gaussian Splatting, Gaussian Splatting methods, Gaussian Splatting surface, autonomous vehicles, Gaussian Splatting
备注： Project Page: [this https URL](https://3dgofish.github.io/)
点击查看摘要
Abstract:Robots, autonomous vehicles, and consumer 360° cameras use fisheye lenses for wide fields of view (FoVs). Simulation-ready digital twins require photorealistic rendering across the full FoV and an aligned collision mesh. However, prior fisheye Gaussian Splatting methods only rasterize up to 180° and do not produce meshes, while Gaussian surface reconstruction methods assume pinhole training images. We present GOFish, the first 3D Gaussian Splatting surface reconstruction method to train directly on fisheye images, rasterize FoVs beyond 180°, and extract an aligned mesh from the same scene. Calibrated densification recovers the surface across the entire FoV, while Calibrated Mip Filtering and Wide-Angle Appearance Modeling keep pixel blur and lens falloff out of the scene and mesh. Across the 200° FIORD and 205° FullCircle benchmarks, GOFish improves PSNR over the best prior method by 2.28 and 1.97 dB, respectively. It also outperforms fisheye rendering baselines within their own FoVs and achieves the highest mesh F-score on FIORD and the main ScanNet++ set. Together, the Gaussian scene and mesh form a LiDAR-free, simulation-ready digital twin that renders in real time, ray traces the full FoV, and provides collision geometry.
226. 【2610.37806】ACOS: Terrain-Aware COntact Supervision for Visual Footstep Planning
链接：https://arxiv.org/abs/2610.37806
作者：Jean-Michel Fortin,William Larrivée-Hardy,Gabriel Jeanson,Olivier Gamache,Daniele De Martini,François Pomerleau,Philippe Giguère
类目：Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
关键词：Humans navigate, vision system, navigate their environment, ground will respond, respond before committing
备注： 9 pages, 8 figures, Submitted to ICRA 2027
点击查看摘要
Abstract:Humans navigate their environment based on their vision system, anticipating how the ground will respond before committing to a step. Self-supervised learning could provide legged robots the same ability, but the supervisory signal is typically chosen on intuition, and its effect on what the model learns is rarely examined. We present Terrain-Aware COntact Supervision (TACOS), which proposes a meaningful and consistent footstep instability score, based on the residual of a factor graph in legged state estimation. It projects the continuous score into preceding images to supervise a dense visual predictor. We propose two variants, a hard score that exploits hindsight and a causal soft score available immediately. Both metrics are compared against state-of-the-art alternatives in two independent evaluations that share no data or labels. On 1716 manually annotated contacts, both variants outperform every evaluated alternative at detecting instability and ranking its severity, improving pooled severity ranking by 28 % over the strongest baseline. On three outdoor test runs, our metric also trains the most accurate visual predictor, improving rank correlation by 58 % over the strongest alternative, and by a further 25 % when supplemented with onboard context. In a real-world experiment over a 50 m forest trajectory, the trained predictor handles non-geometric terrain properties such as mud and vegetation, while stepping around geometric obstacles like fallen branches. Code is available for reproducibility at this https URL.
227. 【2610.37784】Sparse tensor networks for individual tree diameter estimation in mixed forests
链接：https://arxiv.org/abs/2610.37784
作者：Clara Lößl,Fabian Ewald Fassnacht,Viacheslav Barkov
类目：Computer Vision and Pattern Recognition (cs.CV)
关键词：breast height, regions are impractical, ULS, Quantitative Structure Models, diameter at breast
备注：
点击查看摘要
Abstract:Measurements of tree diameter at breast height (DBH) are fundamental to forest inventories, but manual surveys across entire stands or regions are impractical. UAV-borne laser scanning (ULS) enables scalable automated forest surveys, yet canopy occlusion makes accurate DBH estimation from ULS point clouds non-trivial. Limited stem-level point density of ULS data constrains approaches relying on detailed stem reconstruction, such as Quantitative Structure Models (QSMs) and circle fitting. Empirical machine learning methods based on predefined point cloud metrics are an alternative. New deep learning approaches offer an even more principled solution, allowing models to learn directly from point clouds. Sparse tensor networks have shown particular promise, transforming forestry tasks such as tree species classification. We investigate whether these strengths can be extended to estimate individual tree DBH from ULS data across four mixed-species temperate forest plots in Germany. We further advance existing sparse tensor networks by introducing TreeNext, a lightweight deep learning architecture for direct individual tree parameter estimation from voxelized LiDAR point clouds. We present a fully automated forest inventory pipeline from individual tree instance segmentation to DBH regression, and quantify the impact of segmentation errors by comparing performance on automatically segmented and ground truth instances. TreeNext achieves an $R^2$ of 0.8355 on automatically segmented trees and 0.9380 on ground truth segments, outperforming strong tabular baselines including Random Forest and XGBoost. Segmentation quality emerges as the dominant error source in the automated pipeline. TreeNext proves highly data-efficient, performing well even in small data scenarios. Our findings establish sparse tensor networks as a practical foundation for automated individual tree DBH estimation from ULS.
228. 【2607.21628】Wavelet Phase Diffusion for Structurally and Semantically Consistent Sim-to-Real Translation
链接：https://arxiv.org/abs/2607.21628
作者：Kaiwen Wang,Frank Bieder,Yinzhe Shen,Carlos Fernandez,Jan-Hendrik Pauls,Omer Sahin Tas
类目：Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
关键词：semantic consistency, methods, methods achieve, semantic, Conditioning-based methods achieve
备注： Accepted by WACV 2027. 20 pages, 15 figures. Project page: [this https URL](https://kit-mrt.github.io/Wavelet-Phase-Diffusion/)
点击查看摘要
Abstract:Simulation-to-reality translation must bridge the appearance gap between synthetic and real domains while preserving structural and semantic consistency. Conditioning-based methods achieve spatial alignment but introduce computationally expensive control modules. Paired-data methods achieve realism but rely on complex synthesis pipelines, often altering scene geometry and semantics. Training-free editing methods avoid both constraints but lack a learned appearance prior, limiting their perceptual quality. Recently proposed phase-preserving diffusion presents a promising alternative, but Fourier-domain formulations are constrained by global spectral coupling. This coupling induces spatial artifacts such as ringing and boundary leakage, thereby degrading structural and semantic consistency. We introduce Wavelet Phase Diffusion, which addresses this through two components. First, we operate in the Dual-Tree Complex Wavelet Packet Transform domain, whose localized wavelet packets enable spatially adaptive phase injection without global spectral interference. Second, Low-Frequency Randomization (LFR) replaces the low-frequency packet, decoupling the model from the synthetic illumination prior and enabling in-distribution real-world appearance. Both components train on unpaired open-domain data, and introduce negligible inference overhead. The spatial locality further enables instance-level translation, where individual objects or regions are translated to photorealistic appearance independently while the surrounding scene remains untranslated. On vKITTI $\to$ KITTI image translation, $\psi$-PD outperforms prior methods in realism and semantic consistency while maintaining competitive structural alignment. For CARLA video translation, $\psi$-PD approaches the realism of paired-data methods while reducing VLM planner ADE and FDE by $5.4\%$ and $5.1\%$, respectively.
229. 【2607.03487】owards Diverse and Comprehensive Benchmarks for Mutual Information Estimation
链接：https://arxiv.org/abs/2607.03487
作者：Alberto Foresti,Ivan Butakov,Alexander Tolmachev,Giulio Franzese,Alexey Frolov,Pietro Michiardi
类目：Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Information Theory (cs.IT)
关键词：Mutual information, data largely unexplored, realistic data largely, existing benchmarks typically, low-dimensional distributions
备注：
点击查看摘要
Abstract:Mutual information (MI) estimation is a central problem in machine learning and statistics; however, existing benchmarks typically evaluate estimators on simplified, low-dimensional distributions, leaving their performance on complex, realistic data largely unexplored. We address this gap with a comprehensive benchmarking framework grounded in a unified copula-theoretic perspective that subsumes existing benchmarks as special cases. Within this framework, we propose two complementary families of tests: a copula-first family that systematically varies ground-truth MI, dimensionality, and marginal complexity using synthetic and flow-based transformations; and a marginals-first family that couples real-world image data with controlled dependency structures, extending the classic same-class-pairing paradigm. We use this suite to extensively evaluate three classes of estimators: non-parametric, discriminative, and generative. Contrary to prevailing assumptions, our results indicate that there is no universal winner: each category can systematically outperform all other estimators under specific setups. By analyzing these cases, we identify fundamental estimation barriers and propose new tests that more effectively stress these specific limitations. We share the open source code at this https URL.
230. 【2606.15768】LaWAM: Latent World Action Models for Efficient Dynamics-Aware Robot Policies
链接：https://arxiv.org/abs/2606.15768
作者：Jialei Chen,Kai Wang,Kang Chen,Shuaihang Chen,Feng Gao,Wenhao Tang,Zhiyuan Li,Weilin Liu,Zhuyu Yao,Boxun Li,Yuanbo Xu,Chao Yu
类目：Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
关键词：leverage large-scale vision-language, large-scale vision-language pretraining, lack explicit foresight, leverage large-scale, large-scale vision-language
备注： Accepted at CoRL 2026. Project website: [this https URL](https://rlinf.github.io/LaWAM/)
点击查看摘要
Abstract:Vision-Language-Action models (VLAs) leverage large-scale vision-language pretraining for semantic robot control, but often lack explicit foresight into how robot actions change the scene. World-Action Models (WAMs) address this limitation by conditioning policies on predicted futures, yet existing approaches typically rely on computationally expensive video generation with substantial pixel-level redundancy. We present LaWAM, a Latent World Action Model that exposes predictive dynamics to robot policies through compact latent visual subgoals instead of reconstructed future video. At the core of LaWAM is a latent-action-conditioned Latent World Model (LaWM). We obtain LaWM by training a latent action model in the latent space of a pretrained vision foundation model and repurposing its forward decoder to predict future observation features for scene evolution. LaWAM then conditions action generation on these predicted latent visual subgoals to enable dynamics-aware robot control. LaWAM achieves state-of-the-art or competitive success rates (SRs) across LIBERO (98.6% SR), RoboTwin (91.22% SR), and real-world manipulation tasks while retaining low-latency inference. LaWAM runs in 187 ms per action-chunk prediction and achieves up to 24x lower wall-clock latency than pixel-space WAMs.
231. 【2610.39972】Blind Video Denoising for Vocal Tract rtMRI using a Self-Supervised Ensemble
链接：https://arxiv.org/abs/2610.39972
作者：Julian Andres Hernandez-Potes,Tomás Arias-Vergara,Daiqi Liu,Andreas Maier,Paula Andrea Perez-Toro
类目：Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
关键词：Real-time magnetic resonance, articulation directly observable, magnetic resonance imaging, makes speech articulation, speech articulation directly
备注： 7 pages, 5 figures
点击查看摘要
Abstract:Real-time magnetic resonance imaging of the vocal tract makes speech articulation directly observable, but the trade-off between temporal resolution and signal quality introduces substantial noise and obscures fine articulatory detail. Since clean references cannot be acquired for a moving vocal tract, supervised denoising approaches are not applicable in this setting. We propose a self-supervised denoising framework that combines spatiotemporal super-resolution learning with pseudo-supervision derived from a variance-stabilized non-local means filter. A dual-model fusion strategy balances noise suppression and preservation of high-frequency articulatory structure. Experiments on the USC-75 multi-speaker dataset under natural noise conditions show that the ensemble raises the signal-to-noise ratio by 11.9 dB on unseen subjects while attaining more high-frequency articulatory content in the frames. Among tuned implementations of anisotropic diffusion, block-matching filtering, non-local means and pre-smoothed non-local means, every configuration that achieves comparable noise suppression does so at the cost of high-frequency content or of structural fidelity. Under severe added Gaussian, Rayleigh and Rician noise, peak signal-to-noise ratio improves by 7.6 to 11.8 dB and structural similarity roughly triples. An ablation shows that the variance-stabilized pseudo-supervision is an essential element for denoising, and that the effect is attributable to the hard thresholding contained in the stabilization transform.
232. 【2610.38530】From the Superderivative to a Relative-Gradient Edge Operator
链接：https://arxiv.org/abs/2610.38530
作者：Alejandro Garnung Menéndez
类目：Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
关键词：Roberts and Scharr, Scharr measure absolute, absolute grey-level differences, First-derivative edge detectors, measure absolute grey-level
备注： 15 pages, 11 figures, 6 tables
点击查看摘要
Abstract:First-derivative edge detectors such as Sobel, Prewitt, Roberts and Scharr measure absolute grey-level differences. We revisit the log-gradient magnitude, a relative-gradient edge operator, and motivate it by analogy with the superderivative, a multiplicative derivative in which the independent variable is perturbed multiplicatively rather than additively. In one and two dimensions the logarithm of the superderivative is a radial logarithmic derivative; its local, origin-free, isotropic analogue is the magnitude of the logarithmic gradient. The resulting operator measures relative rather than absolute change, giving an edge-strength map that is invariant under global intensity rescaling. We present the theory, show how the operator can be implemented in two steps---a classical gradient kernel plus a pointwise division---and compare it with standard first-derivative filters on synthetic images with ground-truth edges and on the official BSDS500 benchmark. We also study a parametric family that interpolates between the absolute gradient and the log-gradient; empirically, the best members are small intermediate exponents, which edge the classical kernels by a small but statistically detectable margin in the per-image metric and, with the same Gaussian smoothing as the single-$\sigma$ Canny, match it in ODS and lead it in OIS and AP, while the tuned Canny remains ahead in ODS. The single-threshold margin over the kernels does not replicate on the held-out test split, and the fully relative endpoint is never the best member, its ranking depending on the edge-map normalisation. The contribution is the construction and the invariance it buys, not a new state of the art.
233. 【2610.37709】owards Myocardial Infarction Diagnosis: An Investigation into Machine Learning and Deep Learning Algorithms using ECG Data
链接：https://arxiv.org/abs/2610.37709
作者：Rafid Ahnaf,Husne Mubarak,Fahim Ahsan,G. M. Refatul Islam,Rafeed Rahman,Nabuat Zaman Nahim
类目：Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
关键词：myocardial injury due, potentially fatal cardiac, fatal cardiac condition, cardiac condition caused, Myocardial Infarction
备注： 18 pages, 7 figures, 3 tables. Based on the authors' undergraduate thesis at BRAC University (2023)
点击查看摘要
Abstract:Myocardial Infarction (MI) is a severe and potentially fatal cardiac condition caused by myocardial injury due to ischemia, and timely detection can substantially reduce its mortality. Electrocardiography (ECG) is the primary diagnostic modality for MI, but manual interpretation is susceptible to human error. In this work, we propose Visionary V2, a multi-step, adaptive filtration method for ECG signal normalization, and use it to study machine learning (ML) and deep learning (DL) approaches to MI detection on the PTB Diagnostic ECG Database using all 15 available leads. On statistical features extracted from the filtered signals, such as the R-R interval, a Bagging ensemble of decision trees reaches 96.6% accuracy, compared with 87.7% when the signals are processed with BioSPPy. We further generate an image dataset from the filtered signals, on which a custom ResNet-50 misclassifies only 1 of 7,000 validation images and a custom CNN reaches an F1 score of about 98.7%. Finally, we apply LIME, Grad-CAM and Integrated Gradients to verify that the image models attend to the ECG waveform rather than the background. These results indicate that careful signal normalization combined with ML and DL models can support fast and accurate automated MI screening.
