# Research Papers — 2026-09 · bundle page 21 of 266 > Full-text markdown of papers #1001–1050 of 13298 published in 2026-09, newest first. AI-readable mirrors of the full 13298-paper 2026-09 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-09/page/21 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/20.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/22.md) ### TAEC: Trajectory-Aware Evidence Coordination for Multi-Step Visual RAG Authors: Yalun Wu, Bingzhou Wang, Boyang Wang, Peiying Wang, Shaojie He, Yunhan Wang, Shaozu Yuan, Jiawei Wang Published: 2026-09-29 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.37349v1 Multi-step visual retrieval-augmented generation (RAG) answers complex questions by repeatedly retrieving visual evidence, updating an intermediate state, and deciding whether to continue searching or answer. Yet retrieving relevant evidence does not ensure its effective use throughout the reasoning trajectory. As multi-step reasoning progresses, redundant sources occupy context capacity needed for missing evidence, observations tied to resolved requirements or unproductive searches linger in context, and visual sources are revisited with insufficient detail for fine-grained reading. We term this loss of usable evidence over a reasoning trajectory trajectory-level evidence utilization degradation. To address it, we propose Trajectory-Aware Evidence Coordination (TAEC), a training-free framework that coordinates evidence use around unresolved answer requirements. TAEC tracks these requirements in a shared trajectory state to guide which evidence enters the context, how accumulated memory is retained, and at what level of detail visual evidence is examined. Under a unified evaluation protocol on ViDoSeek, SlideVQA, and MMLongBench-Doc, TAEC achieves the best overall performance against leading training-free visual RAG baselines, with the highest average accuracy across multiple proprietary vision-language models. These results demonstrate that aligning evidence with evolving reasoning needs improves evidence use throughout multi-step visual RAG. --- ### When to Retrieve, When to Stay: Uncertainty-Aware Temporal Evidence Allocation for Streaming Video-LLMs Authors: Xiang Hu, Jiazuo Yu, Lu Zhang, Yunzhi Zhuge, Huchuan Lu Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37345v1 Streaming video understanding requires Video Large Language Models (Video-LLMs) to reason over continuous visual streams under causal constraints. As the visual history grows, a bounded visual?processing budget requires evidence selection that balances temporal recency with query relevance. Recent-only selection excludes potentially relevant historical evidence, whereas Semantic-only retrieval can displace useful recent context when relevance scores are ambiguous. We introduce WRWS (When to Retrieve, When to Stay), a training-free framework for uncertainty-adaptive evidence allocation. A lightweight external vision-language encoder scores query relevance across the observed history, while an adaptive allocation module uses the normalized entropy of the similarity distribution as a proxy for retrieval uncertainty. WRWS favors semantic retrieval when relevance cues are reliable and strengthens the recency prior under uncertainty. Following a retrieve-first, encode-later pipeline, WRWS selects evidence before target-model visual encoding, such that only the selected observations are processed by the costly target Video-LLM. Experiments across four Video-LLM families and multiple model scales demonstrate competitive accuracy on StreamingBench and OVO-Bench. In our efficiency evaluation, WRWS reduces average vision-to-answer time to 47.93% of the state-of-the-art method. Code will be released. --- ### A Sharp Transition in Data Reconstruction under Differential Privacy Authors: Max Cairney-Leeming, Simone Bombari, Marco Mondelli Published: 2026-09-29 Categories: cs.LG, cs.AI, stat.ML arXiv: https://arxiv.org/abs/2609.37344v1 Data reconstruction attacks have empirically been successful in recovering training samples from learned models, raising privacy concerns and motivating defenses with guarantees that remain valid against future threats. While differential privacy (DP) provides formal protection, choosing the privacy budget remains a challenge: small budgets severely reduce utility, but it is hard to quantify how large the budget can be without allowing accurate reconstruction. In this work, we study informed attackers who aim to reconstruct a single $d$-dimensional training sample from a $ρ$-zero-concentrated DP model, knowing all other training data. Our main contribution is to establish a sharp transition at $ρ\asymp d$ for data reconstruction: on the one hand, we derive entropy-based lower bounds for any private mechanism and any attack, characterizing a set of target priors for which reconstruction is information-theoretically impossible for $ρ\ll d$; on the other hand, we analyze a simple attack on private linear regression with output perturbation, showing that reconstruction is practically feasible for $ρ\gg d$. Remarkably, the transition moves to $ρ\asymp s$ for data lying in an $s$-dimensional subspace, demonstrating that the privacy budget guaranteeing adequate protection must be assessed in terms of the effective dimension of the data. We validate our findings via experiments on synthetic data and natural images (CIFAR-10, ImageNet). --- ### HyperSAM: A Promptable Foundation Model for Hyperspectral Remote Sensing Authors: Li Pang, Xinqiao Wu, Jing Yao, Pedram Ghamisi, Jun Zhou, Zhengchao Chen, Deyu Meng, Xiangyong Cao Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37340v1 Hyperspectral remote sensing provides dense spectral measurements that are indispensable for material-level Earth observation, yet the construction of a general-purpose hyperspectral foundation model remains difficult. Two bottlenecks are especially limiting. First, large hyperspectral corpora rarely provide high spatial resolution together with reliable dense annotations. Second, many hyperspectral models are still trained almost from scratch, so the geometric and interactive priors learned by modern vision foundation models are not fully reused. To alleviate these issues, we \highlight{present} \textbf{HyperSAM}, a promptable hyperspectral foundation model that couples a data-centric hyperspectral synthesis pipeline with a spectral adaptation architecture based on Segment Anything Model 3 (SAM3). On the data side, HyperSAM synthesizes full-spectrum hyperspectral cubes from high-resolution SpaceNet multispectral imagery through a physics-informed abundance-transfer generator, while SAM3-derived pseudo-masks provide object-centric supervision. On the model side, the latest implementation uses a frozen SAM3 RGB image branch, a trainable hyperspectral side encoder initialized from the RGB vision transformer (ViT), ControlNet-style zero-initialized feature injection, and a lightweight mixture-of-experts mask refiner. To enhance training robustness against noisy pseudo-labels, Cross-modal Sample Selection (CromSS)-style confidence selection is incorporated for noisy-label weighting. Extensive experiments show that HyperSAM obtains strong generalization on diverse hyperspectral tasks (e.g., classification, anomaly detection, change detection, target detection, and airborne oil-spill mapping) and that high-quality synthetic hyperspectral data can be more effective than simply scaling noisy hyperspectral supervision. --- ### UGO: Unified Architecture for General Multi-Object Tracking by Segmentation Authors: Jer Pelhan, Alan Lukezic, Matej Kristan Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37339v1 General multi-object tracking (GMOT) tracks all instances of a user-specified category from a single first-frame exemplar. Prior work relies on bounding boxes and surrogate training, and struggles with non-rigid objects, crowded scenes, and distractors. We introduce UGO, a unified GMOT tracker that pairs a pretrained exemplar-conditioned detection head with an instance-propagation head in a common architecture. A novel training-free, energy-minimization consolidation method converts overlapping proposals into exclusive pixel-wise masks and detections, resolving over-segmentation, duplicates, and conflicts. A hierarchical memory spanning global and instance levels improves recall and per-instance segmentation accuracy using a new memory management protocol. UGO sets a new state-of-the-art on GMOT benchmarks and video object counting, and is competitive with specialist MOT methods, establishing a strong paradigm for unified, open-category multi-object tracking. --- ### OFBD: Object-Focused Background Debiasing for Long-Tailed Learning Authors: Shenghan Chen, Yiming Liu, Zhipeng Deng, Haolin Wang, Jiale Zhou, Zhijian Wu, Xiankai Lu, Yafei Ou, Yefeng Zheng Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37331v1 Balancing performance trade-offs on long-tailed data distributions remains a long-standing challenge in visual recognition. Existing methods mainly improve tail classes through re-balancing, representation learning, or data augmentation, but the underlying cause of tail class degradation is still insufficiently explored. In this paper, we find that standard long-tailed training induces background-biased representation and optimization: tail classes suffer larger background distribution shifts and become increasingly driven by background gradients. This reveals that tail degradation is not merely caused by insufficient samples, but also by the learning of irrelevant background features. To tackle this issue, we propose Object-Focused Background Debiasing (OFBD), a framework that mitigates background bias from both distribution and optimization perspectives. Specifically, Foreground-guided CutMix preserves target-related foregrounds while diversifying complementary backgrounds, and Background-guided Feature Rectification suppresses background-biased features without learnable parameters or additional training. Extensive experiments show that our method improves overall accuracy, achieves significant tail-class gains, and can serve as a plug-in for mainstream long-tailed methods without external data or pretrained recognition models. The code is available at: https://ofbd-neurips2026-longtail-learning.github.io/ --- ### The Domain Is a Residue: Adapting Self-Supervised Features, Not Generators Authors: Thomas Deixelberger, Markus Steinberger Published: 2026-09-29 Categories: cs.CV, cs.GR, cs.LG arXiv: https://arxiv.org/abs/2609.37330v1 Clearing fog, rain or snow from footage, or turning renders into photographs, must remove the source domain and keep the scene. Unpaired translators carry it through because their generator sees the source appearance (pixels, a near-invertible latent or a control map) and keeps it. A DINO feature map fixes what is in the scene and carries weather, lighting and rendering style as a residue of 13 to 14% of the feature norm. We propose the Representation Feature Adapter (RFA), a 2.9M-parameter network that moves this residue. We train only the adapter and its discriminators; the encoder and a feature-conditioned decoder, trained once for all conditions, stay frozen. Against CycleGAN-Turbo it is ahead on both metrics on fog and on KID on night, and level within noise on snow, rain and haze. On sim-to-real it leads REGEN and HyPER-GAN on both metrics. Only the RFA removes the rain while keeping the scene. The removal costs scene structure: CycleGAN-Turbo keeps more on every condition but fog. On VAE latents the identical adapter collapses to the identity, and decoders from other groups that never saw it render its output. The RFA has about 160 times fewer trainable parameters than CycleGAN-Turbo and under a fifth of its per-condition training time. --- ### Solving Without Stopping: On-Policy Distillation at Small Scale Authors: Hongyang Li, Yiming Zhu, Xiao Li, Caesar Wu, Said Mammar, Pascal Bouvry Published: 2026-09-29 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.37326v1 On-policy distillation, where a student learns from a stronger teacher's feedback on its own outputs, is a common way to pass reasoning to smaller models. We analyze what it transfers at small scale, distilling Qwen3-8B into Qwen3 4B, 1.7B and 0.6B students, in thinking mode (reason at length, then end the reasoning and answer) and, for comparison, in non-thinking mode (no separate reasoning phase). Long reasoning needs two abilities, solving a problem and knowing when it is solved, and we find that distillation transfers the first, but in thinking mode not the second. Solving improves at every size, up to two ceilings, which we measure comprehensively across both modes and all student sizes: a student's single attempt never exceeds what it could already reach in many attempts before training, and the smaller the student, the further it stays below the teacher. Stopping is where the modes part. In non-thinking mode every student keeps stopping; in thinking mode students stop ending their reasoning early in training, and the smaller the student, the less of this ability survives: the teacher signals a stop almost only where a student already ends its reasoning, so distillation teaches no new stops; it only keeps the student's existing stops that land on a right answer, and a weak student has few such stops. The smallest students often reach the right value but do not commit to it: they either rarely mark it or mark it and write past it. Together, these results describe how small students behave under on-policy distillation, and a diagnostic that separates answer marking, correctness and stopping. --- ### VISTA: Value-Informed Event Appraisal for Multimodal Emotion Conflict Authors: Jiale Dai, Liuxian Ma, Xiaoke Niu, Wenjing Zhang, Huiying Zhao, Zhaoxiang Liu, Shiguo Lian, Guojie Song Published: 2026-09-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.37324v1 Conflicting emotional cues can be individually valid: a subdued voice may reflect a blocked goal while a smile satisfies a social obligation. Their interpretation depends on what the event means to the person. We introduce VISTA (Value-Informed Semantic Trust Arbitration), a learned seven-field appraisal interface that conditions modality arbitration on concerns, event relations, and expression conditions while retaining a joint-evidence residual. A log-odds decomposition separates emotion expectation from cue diagnosticity, motivating an interface that lets appraisal change how evidence is interpreted. With a shared Qwen2.5-Omni-7B backbone and matched training examples and steps, VISTA reaches 64.5% conflict accuracy on CA-MER, improving on modality gating by 2.5 percentage points on conflict and 0.2 on consistency. Shuffling appraisal across scenes or removing its decision connection reduces this benefit. A common frozen-backbone probe reaches 0.600 macro CCC for appraisal readout, compared with 0.505 for emotion-only fine-tuning. Evaluations across five benchmarks connect recognition under increasing conflict with appraisal readout and downstream decision use. Together, the analyses and experiments support scene-specific appraisal as an intermediate representation that helps interpret conflicting emotional evidence. --- ### Parallel Tempering for Diffusion-Based Combinatorial Optimization Authors: Arman Mielke, Uwe Bauknecht, Thilo Strauss, Mathias Niepert Published: 2026-09-29 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.37323v1 Discrete diffusion models have emerged as a powerful paradigm for solving combinatorial optimization (CO) problems on graphs by learning to sample high-quality solutions. A common inference-time approach is to generate multiple candidate solutions independently and return the best-performing sample, improving solution quality at the expense of an increase in computational cost. In this work, we introduce PT-Denoise, an inference-time procedure that allows these concurrent denoising trajectories to interact through parallel tempering, without requiring retraining or fine-tuning of the underlying denoiser. Our method assigns a temperature to each diffusion process and allows processes to swap temperatures based on their relative performance. This dynamically reallocates promising, low-energy trajectories to colder, more concentrated sampling regimes while allowing higher-energy states to escape local minima through randomized exploration. Experiments on canonical graph-structured CO problems show that our approach consistently improves the quality of the best solution found, while only adding minimal computational overhead. --- ### Mubric: Mutation Testing-Guided Rubric Generation for LLM Evaluation Authors: Jiayuxuan Yang, Jie M. Zhang, Yiling Lou, Zhenpeng Chen Published: 2026-09-29 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2609.37322v1 Rubric-based evaluation is widely used to assess LLM-based systems by decomposing response quality into task-specific scoring criteria. However, automatically generating rubrics that reliably capture task-specific quality requirements remains challenging. We introduce Mubric, a mutation testing-guided approach to rubric generation. Mutation testing, a classic software testing methodology, evaluates a test suite by injecting faults into programs and checking whether the tests detect them. We draw an analogy between test suites and rubrics: if a rubric captures an important quality requirement, introducing a corresponding defect into an otherwise high-quality response should reduce its score. Mubric first mines common defects from real pairs of preferred and dispreferred responses and abstracts these defects into reusable mutation operators, each specifying how to introduce a particular type of response defect. For a new task, it applies relevant operators to a reference response, checks whether the injected defects reduce response quality, and uses insufficiently penalized defects to refine the rubric. We evaluate Mubric on 703 tasks across four representative domains against six advanced rubric generation methods. Mubric achieves the highest overall evaluation accuracy, outperforming the strongest baseline by 7.48 percentage points. --- ### PowerMarketJax: A JAX Benchmark Suite for Multi-Agent Reinforcement Learning in Power Markets Authors: Zhanhua Pan, Xin Qin, Xiao Liu, Zhilong Cao, Jianhong Wang, Dawei Qiu Published: 2026-09-29 Categories: cs.MA, cs.AI, eess.SY arXiv: https://arxiv.org/abs/2609.37321v1 Power markets are a natural testbed for multi-agent reinforcement learning (MARL), where multiple self-interested participants repeatedly submit bids. A market-clearing mechanism then determines dispatch and prices subject to power grid constraints and market settlement rules. However, existing MARL environments typically focus on a single market setting, implement simplified clearing mechanisms, or rely on CPU-based optimization solvers that slow large-scale training and limit the systematic study of bidding strategies and market behavior. We introduce PowerMarketJax, a benchmark suite for MARL across five power markets: day-ahead wholesale, real-time balancing, ancillary services, peer-to-peer double auctions, and local flexibility. Each environment implements its own clearing, pricing, and settlement rules while providing a common framework for learning and evaluation. We find that learned bidding behavior depends strongly on the market design: independent learners can miss better strategies when gains require many agents to change together, when more profitable strategies lie beyond a region of lower profit, or when profits disappear as more agents adopt the same strategy. PowerMarketJax implements both market simulation and policy training in JAX, allowing the entire pipeline to run on the GPU with 1,024 X 1,200 parallelisms across both environments and market participants, achieving up to 33X speedup over CPU-based baselines. Our open-source benchmark is available at: https://github.com/powermarketjax/PowerMarketJax. --- ### What Comes Next? Omni-StoryBench for Evaluating Story-Grounded Omnimodal Generation Authors: Sieun Hyeon, Yejoon Lee, Mintaek Lim, Woojin Kim, Jaeik Kim, Jaeyoung Do Published: 2026-09-29 Categories: cs.CV, cs.MM arXiv: https://arxiv.org/abs/2609.37317v1 Omnimodal evaluation should go beyond independent text, image, and speech production: individually plausible outputs may not express a coherent shared event. We introduce Omni-StoryBench, a story-grounded omnimodal benchmark evaluating whether models can coherently continue stories across image, narration, and speech. Each instance provides a current storybook page and structured next-page conditions, requiring models to generate the next illustration, narration, and spoken character utterance. Omni-StoryBench contains 900 rigorously validated story transitions from openly licensed children's books, with ground-truth next-page references and speech metadata. We evaluate systems with modality-specific metrics and consistency-centered LLM-as-a-judge rubrics for context preservation, condition following, reference consistency, and cross-modal coherence. Across 32 baseline configurations spanning orchestration, semi-orchestration, and native any-to-any paradigms, we find orchestration with strong VLM planning most reliable, while current native omnimodal models often struggle with output completeness and controllability. Our analysis shows text-side performance is associated with image and speech quality, but image generation and visual continuity form the clearest observed bottleneck among the evaluated configurations. These results position Omni-StoryBench as a system-level benchmark measuring coherent omnimodal generation beyond isolated modality quality. --- ### Do Agent Benchmarks Do What They Say? An Executable-Contract Audit of Tool-Using Agent Environments Authors: Rohith Reddy Bellibatlu, Zichong Wang, Wenbin Zhang Published: 2026-09-29 Categories: cs.SE, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.37315v1 Tool-using agents are entering settings where a wrong action carries real cost, and the benchmarks certifying them grade what each simulated tool call reports having done, assuming the tool did what its interface advertises. The audit taxonomies we survey publish no category for that assumption, and a defect beneath a score is present on every rerun. We treat a tool's advertised surfaces as an executable contract, check the implementation against it, and trace each score's provenance through the task files and evaluator code to the verdicts that derive from state a defective tool should have written. Across 34 audited mutating tools in four benchmarks we confirm seven tool defects and one evaluator property at pinned commits. On injected defects the checker raised no false positive in 25 flags, flagged 2 of 5 negative controls, and missed most: in 29 of 33 scored misses a clause covered the defect but no probe revealed it. The checker's own static half, run alone, flags 14 of 17 confirmed sites, so on these findings the dynamic half confirms and traces rather than discovers. Twelve further AgentDojo tools, with six held-out tools and the seven audited first, complete its 25-tool mutating surface, on which at least 5 tools diverge from their advertised surface as our contracts read it, a rate for AgentDojo alone. No gold trajectory reaches either tau2-bench defect; on 1,120 paths built to isolate the telecom defect, a number fixed by construction, the evaluator rewards a refuel of a suspended line and fails the repaired tool. The clearest case is a clinical benchmark whose tool tells the agent each write executed under a documented no-write design its interface does not disclose; its grader takes that message as evidence, so its action success rate records whether a request carried the expected payload, not whether any record changed. --- ### FLASH: A "Generate Once, Synthesize Many" Framework for Synthetic Anomaly Generation in Industrial Anomaly Detection Authors: Abhay Kumar Das, Rajesh Gangireddy, Ashwin Vaidya, Samet Akcay Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37314v1 Synthetic anomaly generation helps expand industrial anomaly datasets when real defects are scarce or unavailable. Existing approaches lie at two extremes: procedural approaches are fast but struggle to represent complex anomalies, while generative approaches produce diverse defects but require costly per-sample generation. We present FLASH, a framework that decouples defect generation from anomaly synthesis under a ``generate once, synthesize many'' paradigm. Given only normal images, FLASH uses Vision-Language Model (VLM) guidance and an image-generation model to produce a small set of defect images, from which it extracts, validates, and banks reusable defect patches. For synthesis of anomalous images, Object Boundary Suppression (OBS) first identifies the probable foreground object-aware region of the host image, while Multi-Resolution Spectral Pyramid (MRSP) noise generates diverse, size-controllable masks that determine the defect location and spatial extent. It then composes a large and diverse synthetic anomalous image set by localizing the defect region, sampling size-controllable placement masks and seamlessly blending retrieved defects onto new defect-free images without further need for image generation. Experiments on the MVTec AD 2 dataset show that FLASH-generated anomalies nearly close the calibration gap on real defects, reaching 78.1% image-level F1 against an 83.6% real-anomaly upper bound and providing the most consistent calibration transfer across detectors among procedural and generative alternatives. Moreover, FLASH synthesizes anomalies more than 11.95x faster than per-sample generative approaches. --- ### Hidden Reasoning Must Leak, but Need Not Be Readable: Fundamental Opportunities and Limits for Chain-of-Thought Monitoring Authors: Mohammadali Mohammadkhani, Madhava Krishna, Yash Sarrof, Michael Hahn Published: 2026-09-29 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.37312v1 Can reasoning models trick chain of thought (CoT) monitors and perform hidden computation without revealing it in their thinking traces? We show that the answer depends on the underlying task difficulty and the model size. Simple computations can be performed covertly; however, beyond a threshold depending on model size, successfully solving the task necessarily leaks a near-linear amount of information about the covert task input into the CoT. Therefore, sufficiently complex hidden computation always leaves an information-theoretic footprint. However, concerningly, this leakage need not be readable: Under plausible cryptographic assumptions, even a one-layer Transformer can encrypt its reasoning online so that no polynomial-time monitor can extract information about the hidden computation. Overall, our theoretical and empirical results provide a holistic view of both the opportunities and the limitations of CoT monitoring. --- ### ReMem: Rethinking Perception and Memory in Long-Context Recommendation Agents Authors: Haohao Qu, Yongcheng Jing, Chun Hin Chan, Shanru Lin, Wenqi Fan, Dacheng Tao Published: 2026-09-29 Categories: cs.AI, cs.IR arXiv: https://arxiv.org/abs/2609.37311v1 Recent Recommendation Agents (RecAgents) offer a promising alternative by shifting recommendation to an active, user-side paradigm, where generative agents autonomously perceive external platforms, reason over user preferences, and execute decisions. However, existing RecAgents still suffer from two critical limitations: brittle item perception based on noisy and heterogeneous item pages, and inefficient long-context reasoning over extended user histories and multi-step interaction traces. To address these challenges, we propose a novel recommendation agent framework, termed as ReMem, that combines OCR-based multimodal perception with time-evolving dynamic memory. Instead of parsing raw HTML, ReMem observes item pages through screenshots and extracts structured multimodal information via an OCR tool, enabling a more humanoid and platform-agnostic perception mechanism. To support long-horizon preference modeling, ReMem further introduces a chunk-wise sequential memory update strategy, where the agent selectively maintains a fixed-size memory of informative historical interactions while processing arbitrarily long contexts with linear inference complexity and bounded context length. This design allows the agent to preserve evolving user preferences without relying on external memory modules or disrupting the standard autoregressive generation process. To enhance the dynamic memory instruction, we further develop a multi-memory GRPO variant, which propagates the final-answer advantage to all intermediate conversations that contribute to the final response. Extensive experiments on three datasets demonstrate that ReMem consistently outperforms state-of-the-art baselines, achieving an average improvement of 5.16\% across three recommendation agent tasks, namely searching, ranking, and judging. --- ### AutoMark: Enabling Autoresearch to Discover Better LLM Watermarks Authors: Thibaud Gloaguen, Robin Staab, Martin Vechev Published: 2026-09-29 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2609.37310v1 With LLM watermarking being deployed commercially and now required by regulations, improving its reliability and effectiveness has become crucial. Yet, recent progress in the field of LLM watermarking has increasingly been driven by improving details of existing methods, an effort fundamentally limited by the pace of human researchers. In this work, we enable for the first time the autonomous discovery of new distortion-free state-of-the-art watermarking schemes. To enable this, we (i) establish strict criteria to ensure that watermarks are reliable (e.g., they do not have an unexpectedly high false positive rate), (ii) propose rigorous statistical tests to automatically evaluate whether a watermarking scheme satisfies our criteria, and (iii) design an evaluation suite to rank watermarks along three key dimensions: detectability, quality, and robustness. By running our framework with 3 frontier models (GPT-6 Astra, Opus 5, Gemini-3.8 Flash), we discover over 50 different watermarking schemes, including several that outperform prior works along all key dimensions. We complement this by a manual study of the discovered schemes, distilling the key ideas into smaller components, and individually studying the impact of each component across dimensions (detectability, quality, robustness) to better understand how the proposed schemes operate. Importantly, we find that the agents, on top of improving existing ideas, also discover fundamentally new ideas (e.g., aligning watermark scores with random per-request direction). Overall, our work establishes the first steps of fully autonomous watermarking research, enabling the discovery of more reliable and effective watermarks. Our code is available at https://github.com/eth-sri/automark, and a blogpost to visualize our results at https://www.sri.inf.ethz.ch/blog/automark. --- ### Hybrid Joint-Selective Optimization: Reduced-Space Levenberg-Marquardt Refinement of Low-Dimensional Parameters of Interest Authors: Muhammad Luthfi Shahab, Gabriella Alfa Indahsari, Imam Mukhlash, Hadi Susanto Published: 2026-09-29 Categories: math.NA, cs.LG, cs.NE, math.OC arXiv: https://arxiv.org/abs/2609.37308v1 This paper introduces a hybrid joint-selective optimization (HJSO) framework for large-scale numerical problems in which a small subset of trainable quantities is of primary interest. We partition the full parameter vector into a high-dimensional remaining block and a low-dimensional block of parameters of interest (POIs), perform joint first-order optimization over the full parameter set, and then freeze the remaining variables while applying a reduced-space Levenberg-Marquardt (LM) refinement to the POIs. The method is designed for settings in which the POIs are low-dimensional but strongly influence the quality of the computed solution, while the full parameter space remains too large for full-space second-order methods. The framework is evaluated on three representative problems: a matrix eigenvalue problem, an inverse Bratu problem solved with a physics-informed neural network, and a 100-dimensional nonlinear Black-Scholes problem solved with the DeepBSDE method. In each test, HJSO reaches prescribed POI-error thresholds faster than the corresponding joint first-order baseline and improves the final POI accuracy for the reported solver configurations. The contribution is therefore not a universal optimizer, but a practical reduced-space strategy for problems with known low-dimensional parameters of interest and expensive high-dimensional training variables. --- ### MetaCtrl: Your Large Language Models Can Reason Better and More Concisely with a Metacognitive Controller Authors: Zhibin Wen, Tao Han, Lei Bai, Can Li, Yang Xu Published: 2026-09-29 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.37304v1 Large reasoning models improve performance on challenging problems by allocating additional computation before answering, but longer reasoning does not always lead to better results and can introduce substantial redundant reasoning on simple problems. Conversely, aggressively shortening reasoning can degrade performance on difficult ones. Effective reasoning therefore requires dynamically deciding when additional computation is useful based on the reasoner's capabilities and evolving solution state. Existing approaches often rely on predefined budgets or intervention rules, retrain the target reasoner, or require additional supervision. We introduce MetaCtrl, a lightweight controller that adaptively regulates a frozen reasoner without predefined token budgets or reasoner retraining. We formulate reasoning regulation as a sequential metacognitive control problem: MetaCtrl observes the evolving reasoning trace and decides whether to continue, simplify, skip redundant steps, or conclude reasoning. It is trained directly with reinforcement learning using a reward that prioritizes correctness while favoring shorter trajectories among correct solutions, requiring neither supervised intervention trajectories nor problem-specific budgets. Across seven benchmarks spanning mathematics, science, and code, MetaCtrl consistently improves the accuracy of LRMs while reducing their reasoning length. On DeepSeek-R1-Distill-Qwen-7B, it improves average accuracy by 4.7 points while reducing generation length by 53.3%. Without further training, the same controller transfers to an unseen reasoner (e.g., Qwen3-14B), improving average accuracy by 2.9 points and reducing generation length by 50.3%. These results establish MetaCtrl as a plug-and-play controller for improving reasoning accuracy while substantially reducing inference-time generation. The code is available at https://github.com/binbin2xs/MetaCtrl. --- ### Technical note on: Zero-Training Feature-Space Alignment via Information Geometry Authors: Behraj Khan, Tahir Qasim Syed, Syed Ahmad Chan Bukhari Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37302v1 Deep vision models often degrade under distribution shift. Test-time adaptation can improve robustness but typically requires iterative optimization, hyperparameter tuning, and multiple forward-backward passes. We propose Zero-Training Fisher Geometry Alignment (ZFGA), a closed-form method that improves robustness under covariate shift without modifying model parameters. ZFGA is based on the observation that distribution shifts distort feature-space geometry. It estimates the Fisher information matrix of the predictive distribution with respect to feature embeddings and applies a linear transformation that aligns test-feature Fisher geometry with a reference geometry computed from clean data. This provides a natural-gradient-inspired preconditioning step in feature space. We evaluate ZFGA on CIFAR-10-C and ImageNet-C using ResNet-50, DINO ViT-S/16, and CLIP ViT-B/32. ZFGA consistently improves over zero-shot inference across all three models, although it is not the strongest method for every model. Covariance whitening performs better on ResNet-50, while Fisher whitening is statistically indistinguishable from ZFGA on CLIP. Across six training-free and gradient-based alternatives (covariance whitening, Fisher whitening, TENT, T3A, LAME, and AdaNPC), ZFGA is the only method that does not substantially harm any of the three model families. The Fisher geometry distortion is also positively correlated with ZFGA gain (Pearson r = 0.366, p = 0.017), providing preliminary evidence that geometric misalignment contributes to robustness degradation. ZFGA requires only forward passes and matrix operations at inference time, offering a lightweight and deterministic alternative to optimization-based test-time adaptation. --- ### Scaling Full Conformal Image Classifiers Authors: Julio Silva-Rodríguez, Ender Konukoglu Published: 2026-09-29 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2609.37298v1 Conformal prediction provides set-valued predictions with distribution-free coverage guarantees, making it attractive for high-stakes image classification. However, split conformal prediction is data-inefficient, while full conformal prediction (FCP), despite its stronger statistical efficiency, is computationally prohibitive at scale because it requires candidate-specific model refits at test time. We address this limitation by leveraging zero-shot vision-language models (VLMs) to guide scalable FCP in large label spaces. We introduce Targeted Full Conformal Prediction (T-FCP), which uses a lightweight inductive conformal predictor to prune unlikely labels and applies FCP only to the remaining candidates, reducing computation while retaining the formal guarantee of the combined conformal procedure. We further propose Stabilized Online LDA (SO-LDA), an efficient VLM adaptation solver based on rank-one inverse-covariance updates. Across multiple benchmarks, including ImageNet, T-FCP enables practical full-conformal image classification with modest test-time overhead, yielding efficient prediction sets and more stable empirical coverage than split conformal alternatives. --- ### Why Cross-Skeleton Retargeting Is Non-Identifiable: Structural Limits of Generative Motion Models Authors: Zhiyuan Li, Wenyan Yang, Pekka Marttinen, Joni Pajarinen Published: 2026-09-29 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2609.37297v1 Cross-skeleton motion generation trains generative models to carry action structure and motion intention from one body to another. Yet a target motion that shows the right action has two explanations that the training data cannot tell apart: the model transferred the source clip, or it recovered a typical motion for the requested action. We show that this ambiguity is structural rather than incidental: under standard generative objectives, the source-conditioned retargeting map is non-identifiable in sparse heterogeneous motion domains. Unpaired distribution matching yields gauge non-identifiability: the latent spaces of different skeletons can be transformed relative to one another without changing the training evidence, so different source-conditioned maps fit it equally well. Sparse paired supervision admits the complementary failure mode, \emph{conditional-mean degeneration}: when clips are paired only by action, squared-error training converges to an average target motion that ignores the source clip. To make the missing evidence observable, we introduce Source-Instance Fidelity (SIF), a diagnostic that tests whether outputs differ from one another the way their source clips do, with the target skeleton and action held fixed. Under this diagnostic, methods that succeed at the standard action-level test on animal motion data often sit at the source-blind floor, while the methods that rise above it retain only a partial relational signal. Retargeting therefore needs objectives and evaluations that can identify the source-conditioned map it claims to learn. Project page: https://cross-skeleton-retargeting.netlify.app/. --- ### SafeLLM4SE: Statistical Evaluation and Reporting for LLM-based Software Engineering Systems Authors: Francisco Ortin Published: 2026-09-29 Categories: cs.SE, cs.LG arXiv: https://arxiv.org/abs/2609.37294v1 Large language models (LLMs) are increasingly used for software engineering tasks, yet their stochastic behavior challenges the validity, reproducibility, and comparability of their evaluations. Conventional practices such as reporting a single output, an average score, best-of-N, or pass@k performance can obscure variability and estimation uncertainty, potentially leading to misleading conclusions about system reliability. This article presents SafeLLM4SE, a practical methodology and reporting standard for statistically principled evaluation of LLM-based software engineering systems. Rather than treating generated outputs as deterministic artifacts, SafeLLM4SE treats them as realizations of a stochastic process and distinguishes quality, stability, and estimation uncertainty. It combines adaptive sampling with confidence intervals, distribution-aware statistical comparisons, effect sizes, and a minimum reporting standard covering model configuration, reproducibility, evaluation procedures, and resource usage. SafeLLM4SE is also provided as an open-source software package available on PyPI, enabling researchers and practitioners to reproduce and extend the methodology. We illustrate its application by comparing two LLMs on HumanEval, a benchmark of programming problems assessed through functional tests. --- ### VISTA-Bench: Benchmarking Multilingual Image Translation with Image-Specific Rubrics Authors: Bo Lv, Mao Zheng, Zheng Li, Fangxu Liu, Mingrui Sun, Tao Chen Published: 2026-09-29 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.37287v1 Image translation is a fundamental capability of multimodal models for multilingual applications, requiring visual understanding and meaning preservation across languages. However, existing benchmarks have limited language coverage and often lack explicit image-specific evaluation criteria, making it difficult to comprehensively assess this capability. To systematically evaluate this capability, we introduce VISTA-Bench, covering 22 languages and 10 domains, and develop an image-specific rubric evaluation protocol. The benchmark combines sampling for language and scenario coverage with model-assisted, human-verified annotations that group related text into coherent semantic units and provide multilingual reference translations. The rubrics specify essential content, semantic relations, and acceptable translation variants, yielding separate output-based scores for translation quality and the preservation of visual and knowledge-dependent information. We conduct extensive evaluations of 16 mainstream models, including 12 multimodal models and four text-input models, and provide systematic analyses across languages, domains, and evaluation dimensions. --- ### AssayRouter: Historical Utility Priors for Frozen Molecular Predictor Routing Authors: Dong Xu, Zhangfan Yang, Jiantao Wu, Shipeng Zhang, Zexuan Zhu, Jiangqiang Li, Jun Zhang, Junkai Ji Published: 2026-09-29 Categories: cs.AI, q-bio.BM, q-bio.QM arXiv: https://arxiv.org/abs/2609.37285v1 Laboratories often face a new molecular assay with 16-64 labels and a bank of predictors whose training data and parameters are unavailable. The practical question is which frozen outputs to include in a small local model. AssayRouter treats completed assays as pseudo-targets and labels each candidate by its post-fit utility: the reduction in held-out discovery loss when the candidate is added to the local target predictor. A shared regressor learns to predict this utility from candidate behavior on the support set, without source identity; on a new assay, one frozen ranking selects four sources and separate labels fit a convex combiner. We train only on completed ChEMBL-MT assays and evaluate 24 external regression assays across six frozen interface families. AssayRouter-C lowers strict four-call negative log-likelihood (NLL) by 0.0409 relative to Support-CV@4. Frozen candidate-label permutations confirm that candidate-utility correspondence carries the transferred information, and leave-one-interface-out training shows that the mapping generalizes to unseen predictor families. Completed assays therefore provide transferable supervision for scarce-label routing through frozen prediction interfaces. --- ### SAM Meets VLM: Parameter-Decoupled Full-Parameter Training for Unified Medical Reasoning and Segmentation Authors: Xuyang Cao, Enyou Liu, Jun Zhao, Zhuoyun Liu, Jintao Fei, Leo Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37283v1 Medical multimodal large language models (MLLMs) are increasingly expected not only to answer clinical questions, but also to localize the visual evidence behind their predictions. A common strategy connects a vision--language model (VLM) with SAM-style segmentation through a special <SEG> token, yet full-parameter training of this unified architecture is difficult because image-level reasoning and pixel-level segmentation impose different requirements on the shared representation space. To address this issue, we propose a parameter-decoupled training framework for unified medical reasoning and segmentation. The framework treats the <SEG> hidden state as a semantic-to-spatial prompt for the mask decoder and encourages it to become separable from generic language states, reducing ambiguous segmentation prompts and potential disruption to reasoning representations. It first performs medical shallow alignment to adapt visual features to clinical language without disturbing the LLM; then controlled instruction tuning shapes separable <SEG> prompt states, monitored by the Davies--Bouldin Index (DBI), while scaling segmentation gradients entering the language backbone; finally, the SAM branch is specialized with the VLM frozen to improve mask precision without altering reasoning parameters. Experiments on medical referring segmentation, grounding, visual QA, and textual QA benchmarks show that our framework achieves strong language-conditioned segmentation while preserving competitive reasoning ability. Ablations show that two-phase instruction tuning, gradient scaling, and segmentation specialization all contribute to the model. --- ### Transolver-$σ$: Joint Spectral-Physical Subspace Modeling for Neural PDE Solving Authors: Haonan Shangguan, Hang Zhou, Haixu Wu, Yuezhou Ma, Jianmin Wang, Mingsheng Long Published: 2026-09-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.37279v1 Neural solvers offer efficient surrogates for numerical simulation of partial differential equations (PDEs). For time-dependent problems, strong one-step accuracy does not necessarily translate into reliable autoregressive rollout. We observe that a solver based only on physical-state modeling can achieve lower one-step error, whereas its spectral-only counterpart can become more accurate at later rollout steps. Motivated by this observation, we present Transolver-$σ$, a neural PDE solver based on joint spectral--physical subspace modeling. Within each block, adaptive physical-state interactions and spectral transformations are modeled in dedicated latent subspaces, whose responses are recomposed to enable information exchange between the two representations. Within the physical subspace, we introduce Slice-Residual Physics-Attention (SRPA), which preserves an explicit slice-space identity path while retaining learnable cross-slice interaction. In parallel, an axis-factorized Fourier operator captures global spectral structure. Across five well-established PDE benchmarks spanning steady-state prediction and time-dependent dynamics, Transolver-$σ$ achieves state-of-the-art with a benchmark-averaged relative error reduction of 33.4% over the strongest baseline for each metric, while consistently improving autoregressive rollout over single-operator counterparts. Transolver-$σ$ further delivers strong gains on coupled multiphysics systems and real-world fluid and combustion measurements from RealPDEBench, demonstrating its effectiveness beyond standard simulation benchmarks. --- ### Geometry-Aided Channel Deduction with Partial Channel Estimates and Uncalibrated Digital Twin Authors: Hongning Ruan, Zhaoyang Zhang, Zirui Chen, Ziqing Xing, Zhaohui Yang, Mérouane Debbah Published: 2026-09-29 Categories: eess.SP, cs.LG arXiv: https://arxiv.org/abs/2609.37277v1 The acquisition of high-dimensional channel state information (CSI) in wireless MIMO-OFDM communications usually requires high pilot overhead, or relies on accurate and complete positional or environmental information. In this paper, we propose a geometry-aided channel deduction (GCD) approach, which utilizes an uncalibrated digital twin (DT) with only approximate environmental geometry and positions to assist the channel acquisition. The key rationale behind is that, even imprecise geometric information, which can be easily obtained in advance through radio sensing technologies or existing geographic databases, provides certain structural features about the current channel; meanwhile, the coarse instantaneous channel estimates using only a small amount of pilots provide dedicated information that aligns with the channel structure and further compensates for the geometry inaccuracy and other channel unknowns. To this end, we first extract geometric features from the DT, which contain only simple structural information of the channel. Then we propose random prompt augmentation, a novel method to generate an appropriate prompt that converts geometric multi-path structure into a CSI-like representation while suppressing the disturbance of other unknown channel parameters. The prompt is then fused with the pilot-based instantaneous channel estimate via a channel deduction network. To further enhance the network's versatility, we incorporate pilot configurations into the existing learning architecture to support variable pilot patterns. Comprehensive experiments validate the superiority of the proposed method, which demonstrates high channel acquisition quality, low pilot overhead, and strong robustness. Furthermore, the structural prompt also serves as scenario-related context, enabling our approach to generalize well in new scenarios. --- ### Task-Relevant Null-Space Residuals for Non-Injective Neural Mappings Authors: Bizu Feng, Zhimu Yang, Shuming Wang, Yuan Cheng, Shaode Yu, Xiaojun Qian, Zixin Hu Published: 2026-09-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.37272v1 Non-injective mappings in neural networks map distinct inputs to the same representation, thereby implicitly inducing equivalence relations in the input space. However, the input differences eliminated by these mappings may still be required by downstream tasks, creating a mismatch between operator-induced indistinguishability and task-required distinctions. For non-injective linear operators realized in the current forward pass, their null spaces exactly characterize these invisible input variations. We propose Task-Relevant Null-Space Residuals (NSR), a general residual framework for non-injective linear mappings. NSR combines null-space component extraction from pre-mapping representations, member-level encoding and gating, and application-specific integration to exploit potentially task-relevant information under downstream supervision while preserving the original aggregation or merging rules. We evaluate NSR in two structurally different settings: token merging and graph aggregation. In token merging, NSR achieves higher semantic segmentation performance than the corresponding compressed baselines in 34 out of 36 evaluated configurations, with a maximum observed gain of 31.51 mIoU points under strong compression. In graph aggregation, NSR achieves 100% training accuracy on Tree-NeighborsMatch at depths d=2--6 across three backbones, alongside gains on heterophilic node classification and molecular graph regression. Together, these results support null-space residuals as a practical complement to non-injective linear mappings, enabling downstream models to learn from input distinctions invisible in the original operator's output. --- ### Foundations of Proactive Agents: Principles, Technical Layers, and Proactivity-Gym Authors: Jio Oh, Seunghyun Do, Young-Jun Lee, Steven Euijong Whang, Dongyeop Kang Published: 2026-09-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.37267v1 Proactive LLM agents can turn idle compute into useful support before users ask. Yet even correct work can misread user context, impose review costs, or undermine trust. This work proposes foundations for designing, realizing, and evaluating proactive LLM agents around three joint principles (3T): Task Capability, anticipating relevant needs and correctly performing useful work; Temporal Allocation, allocating compute according to resource availability and when results are needed; and Trust, sustaining users' confidence and appropriate reliance on the agent. We connect these objectives to a design space organized around five dimensions: task scope, anticipation horizon, activation trigger, processing timing, and intervention depth, and specify the situation and system modeling needed to support its choices, including user and environment representations, backbone LLMs, and agent harnesses. Lastly, we propose PROACTIVITY-GYM, a simulation-based evaluation testbed including multi-day scenarios, stateful environments, and persona-conditioned simulated users that can evaluate the consequences of proactive assistance across interactions. Evaluations across 23 model-harness configurations uncover substantial performance gaps across 3T and reveal that LLM judges often conflate task capability and trust. A human study with 30 participants demonstrates the importance of the joint 3T optimization: participants show sharp trust declines after intervention misalignment despite correct outcomes, and prefer sleep-time assistance, even when imperfect, to preserve ongoing focus. Together, these findings support designing and evaluating proactive agents through the joint consideration of useful work, compute allocation, and evolving user trust. --- ### UniAfford: Token-Routed Multitask Learning for Generalizable 2D-3D Affordance Perception Authors: Yuhao Liu, Yiming Zhong, Hanqing Wang, Shaocheng Yan, Yuhang Zhang, Wenzhou Lyu, Ziyang Ding, Wei Zhang, Xue Zhao, Jin Pan, Yuexin Ma, Xinge Zhu Published: 2026-09-29 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.37264v1 Affordance perception aims to localize actionable regions supporting embodied interaction, yet 2D and 3D affordance grounding have evolved as separate problems, with different task definitions, supervision formats, datasets, and evaluation protocols. This fragmentation limits the learning of transferable object-affordance semantics across visual and geometric spaces. We propose Token Router for Tasks, a multitask training paradigm for MLLM-based systems that routes contextual hidden states to task-specific branches without requiring the language head to generate predefined markers. Routed states are supervised directly by branch-specific objectives, enabling dense prediction losses to shape shared MLLM representations. We instantiate this paradigm as UniAfford, a unified framework for generalizable 2D-3D affordance perception, together with UniAfford-Data, a dataset integrating pixel-level 2D annotations, point-level 3D annotations, and language instructions under a shared object-affordance taxonomy, supporting heterogeneous supervision through semantic-level 2D-3D pairing. UniAfford adopts an MLLM as a shared semantic hub and a modality-aware token router to produce image- and point-cloud-affordance queries. These queries respectively condition a SAM-style pixel decoder and a SONATA-based point decoder, enabling flexible 2D, 3D, and joint affordance inference from image-only, point-cloud-only, or paired multimodal inputs. Experiments demonstrate strong zero-shot generalization across 2D and 3D affordance benchmarks without target-specific fine-tuning, alongside state-of-the-art branch-wise performance under modality-isolated protocols. Ablations validate token routing, joint 2D-3D supervision, and decoder coupling, while language-head diagnostics show that routed latent states carry meaningful object-affordance semantics. Project page: https://4dvlab.github.io/UniAfford --- ### Beyond Attention Imbalance: Mitigating Hallucinations via Spectral Surgery Authors: Siqi Lu, Suo Wei, Yongbin Zheng, Jianhang Yao, Wanying Xu, Peng Wang Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37263v1 While Large Vision-Language Models (LVLMs) achieve remarkable success, hallucinations remain a significant barrier to their reliable deployment. Recent studies primarily attribute these issues to cross-modal attention imbalances; most solutions therefore focus on reweighting visual tokens or suppressing language priors. However, such approaches often overlook the spectral characteristics of the visual information flow and frequently rely on Contrastive Decoding (CD), which doubles inference time. Instead of following conventional approaches, we identify two distinct hallucination patterns-Perceptual-Semantic Dissociation and Localized Fixation-and propose FLASH (Frequency-Localized Attention SHaping), a training-free and CD-free framework. FLASH utilizes a Spectral Vortex Score to detect vision heads within multi-head attention layers and applies adaptive spectral modulation to rectify the visual information flow during decoding. Empirical results demonstrate that FLASH achieves a superior balance between performance and efficiency compared to SOTA methods. --- ### Efficiently Approximating Attention Is Hard Authors: Lukas Haverbeck, Carmen Amo Alonso, Andres Felipe Posada-Moreno, Sebastian Trimpe, Marco Pavone Published: 2026-09-29 Categories: cs.LG, cs.CC arXiv: https://arxiv.org/abs/2609.37261v1 Softmax attention is ubiquitous in modern machine learning, but its quadratic scaling with sequence length makes it costly. To reduce this cost, attention is often approximated with fast algorithms, which incur error but can still perform well in practice and on some inputs. At the same time, the growing diversity of attention applications makes approximation guarantees that do not depend on particular input structure a compelling target. For such uniform guarantees over all inputs, known runtime lower bounds rule out fast algorithms for near-exact attention, but leave open the practically important regime: is there an efficient algorithm with even a modest uniform approximation guarantee? We answer this question negatively. Under standard complexity-theoretic assumptions, no truly subquadratic algorithm can approximate attention with any nontrivial additive or relative guarantee uniformly over all inputs. This impossibility holds in the mildest parameter regime for which known algorithms do not already achieve strong approximation guarantees in near-linear time, and extends to practically relevant relaxations: even after polynomial preprocessing of the KV cache, no efficient algorithm can obtain a nontrivial uniform approximation guarantee, or identify a small set of keys receiving substantial attention under sparsity. Overall, our results settle the computational limits of uniform attention approximation. --- ### Collision-Aware and Observation-Aligned Object-Centric Scene Reconstruction from Point Cloud Authors: Yuxuan Xie, Xuan Yu, Rong Xiong, Yue Wang Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37260v1 Object-centric scene reconstruction requires completing partial object observations while preserving metric alignment and avoiding collisions with the surrounding. Existing generation-based methods are often image-conditioned and suffer from scale ambiguity and insufficient geometric constraints. We propose COOL, a framework for COllision-aware and Observation-aLigned reconstruction. Based on an object generation model, COOL conditions the generation on instance and background point clouds. Instance geometry anchors generation in scene coordinates, while background geometry provides local context for scene-consistent completion. We further introduce an explicit collision loss and use joint optimization and resampling to reduce collisions during inference. Experiments on 3D-Front and Scan2CAD demonstrate strong scene-level fidelity, observation alignment, and collision reduction. Moreover, additional studies validate its robustness to mask errors and its applicability to real-world scene replicas. --- ### Loss-Guided Pretraining Data Selection for Time-Series Foundation Models Authors: Yike Li, Shaoxu Song, Jianmin Wang Published: 2026-09-29 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.37255v1 Time series foundation models (TSFMs) are pretrained on heterogeneous collections containing billions of observations, yet their training windows are typically sampled without estimating whether they provide useful learning signal. We introduce a static data-selection framework that scores each window with a reference forecaster and retains an intermediate interval within every source dataset. Specifically, we connect forecasting loss to optimization difficulty by showing that normalized squared loss controls the per-sample gradient norm under a local Jacobian condition. We then define a reference loss score and apply dataset-stratified selection to preserve the diversity of samples. Across various TSFM architectures, our method outperforms random selection by an absolute margin and even improves both relative MASE and CRPS over full-data pretraining by retaining fewer candidate pretraining windows. Further analyses show strong cross-scale and cross-architecture score correlations, indicating that a small reference model can often select data for larger targets, provided that the reference and target share compatible difficulty orderings. --- ### V-JEPA Policy: Building Effective World-Action Models on Predictive Visual Latents Authors: Yang Zhang, Jiangyuan Zhao, Chenyou Fan, Jiayu Hu, Xiu Yuan, Chenjia Bai, Xiu Li Published: 2026-09-29 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2609.37250v1 World-action models (WAMs) couple future visual-state prediction with action generation. By adapting video generators or image-editing models pretrained at scale, a prominent line of recent WAMs inherits both predictive knowledge and the models in which it was learned. We ask whether a predictive visual latent space induced by large-scale predictive pretraining can instead provide a sufficient foundation for effective WAM learning without inheriting a complete pretrained visual generative model. To answer this question, we introduce V-JEPA Policy, a simple framework that builds a WAM on the latent space of a frozen V-JEPA 2.1 encoder. An instruction-conditioned future-latent predictor and a flow-matching action expert are jointly learned from scratch in a single downstream stage, with the predictor's future-informed context key--value states conditioning action generation. With 0.9B total parameters, of which 0.6B are trainable, V-JEPA Policy achieves competitive performance with representative WAM and vision-language-action baselines across LIBERO, LIBERO-Plus, and RoboCasa-GR1. Comparing visual foundations under the same downstream framework and training budget identifies V-JEPA latents as more effective than the discriminative, reconstructive, and video-understanding-oriented alternatives, particularly under distribution shifts. Beyond task-specific learning, pretraining the predictor on DROID video--instruction pairs without action labels and adapting it into a WAM yields substantial gains in downstream control and out-of-distribution generalization. Together, these findings establish predictive visual latents as a foundation for effective WAM learning from task-specific demonstrations and for transferring future-modeling knowledge acquired from broader in-the-wild videos. Our code is available at https://github.com/breez3young/VJEPA-Policy. --- ### Codebook-Guided Cross-Modal Knowledge Distillation for Structurally Heterogeneous Features Authors: Dae Ung Jo, Jongin Lim, YoungJoon Yoo, Daeho Um Published: 2026-09-29 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.37243v1 Cross-modal knowledge distillation transfers knowledge from a teacher modality to a student modality. Existing feature-level alignment methods typically assume that teacher and student features reside in structurally alignable representation spaces. However, this assumption does not hold when cross-modal features are structurally heterogeneous and lack clear unit-level correspondence, such as 2D spatial visual grids and 1D temporal audio sequences, thereby limiting the applicability of feature-level alignment. To address this challenge, we propose a cross-modal distillation framework that enables effective knowledge transfer across structurally heterogeneous feature spaces via a vector-quantized codebook. Specifically, teacher features are abstracted into a set of vector-form codes regardless of their original feature structure, and the selected codes serve as concept-level anchors for student learning. Code selection is guided by both task relevance and student compatibility, allowing the student to receive transferable teacher knowledge without requiring direct unit-level feature alignment. Experimental results across diverse cross-modal distillation scenarios demonstrate the effectiveness of the proposed framework on classification and semantic segmentation tasks. --- ### Trident: Unifying Guarded Dispatch and Host Execution for PyTorch Triton Workloads Authors: Jinjie Liu, Xiaoyan Liu, Shuhan Zhang, Wenjia Sun, Ruilin Yang, Chunlei Men, Yonghua Lin, Shaohua Li Published: 2026-09-29 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.37241v1 User-written Triton kernels enable high-performance GPU computation within PyTorch, but their end-to-end latency can remain dominated by host-side orchestration, especially when device execution is short. Although torch.compile can generate native host wrappers for captured graphs, each invocation still passes through runtime-managed specialization lookup, guard evaluation, and preparation before reaching the wrapper. We present Trident, a compiler backend that removes this recurring overhead from the specialization cache-hit path. Trident introduces the Specialization Cache Module (SCM), which compiles guarded specialization selection, argument and execution-environment preparation, and host execution for multiple specializations into a single executable module. An invocation enters the SCM once, remains in compiled code when a specialization matches, and returns to Python only when a new specialization must be compiled. Built on Torch-MLIR, Trident lowers guards and host-side orchestration to native code while retaining calls to optimized runtime implementations of supported ATen operators. Our evalu- ation on two LLMs shows that Trident achieves up to a 1.47x speedup in model-level end-to-end latency over eager execution and up to 1.68x over torch.compile. --- ### Differentiating Bisimulation Metrics: A Framework for Parametric Markov Chain Fitting via Bicausal Optimal Transport Authors: Sergio Calo, Amy Zhang, Javier Segovia-Aguas, Anders Jonsson Published: 2026-09-29 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.37239v1 Many problems in sequential decision-making, such as imitation learning from observations, state-space compression, world-model learning, and sim-to-real transfer, can be reduced to learning a model such that a notion of distance with respect to the target process is minimized. We consider this general framework and consider the bisimulation metric, equivalently Bicausal Optimal Transport (BOT), as the notion of distance to minimize. We show that BOT, since it can be formulated as a linear program (LP), is differentiable with respect to the model dynamics. We then derive an exact closed-form gradient via the envelope theorem applied to the LP saddle point. The result is a general algorithm, Differentiable Bicausal Optimal Transport (D-BOT), that can be applied to each of the problems above. The proposed algorithm learns the best model by alternating between distance computation and gradient steps. We apply D-BOT for three different settings: state-space compression, parametric model learning, and imitation learning from observations (ILfO). We show empirical results that confirm the viability of all three instantiations. --- ### Asking for What Was Never Requested: Horizontal and Vertical Proactivity in Agents Authors: Ido Levy, Asaf Yehudai, Segev Shlomov, Asaf Adi, Leshem Choshen Published: 2026-09-29 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.37236v1 An agent that uses tools typically responds to what the user explicitly asks, yet completing the task may require information the user never requested. Work on proactive agents mainly studies whether and when an agent should act on its own, not what information it should pursue. We study a distinct axis of proactivity: its content. Horizontal proactivity pursues unstated information that the current context already identifies, and vertical proactivity pursues needs that only earlier evidence reveals. A need graph, recovered from a benchmark's own decomposition, records which needs depend on which, so both forms, and whether the agent stops at the right time, can be scored from a transcript without a model judge. To learn this behavior, we propose Q&D (questioner and drafter), which trains a questioner to prefer the question whose continuation retrieves more of the required evidence, with no reward model or judge. On held-out splits of three multi-hop question-answering benchmarks, at equal retrieval spend, the trained questioner improves both forms of proactivity over the same model, prompted, and outperforms a prompted model $15\times$ larger in the same role on two of the three, and the gain persists after controlling for question volume and length. Without further training, we place the questioner in an interactive customer-service agent with a simulated customer, where it completes more tasks while asking fewer questions, and in retail it outperforms the $15\times$ larger model with fewer follow-up turns from the customer. These results show that proactivity depends not only on whether an agent acts without being asked, but also on what it chooses to pursue and when it stops. --- ### DatalogBench: Evaluating Large Language Models on Text-to-Datalog Synthesis Authors: Yuan Li, Hanyun Jiang, Guowei Tian, Chengpeng Wang, Peisen Yao Published: 2026-09-29 Categories: cs.PL, cs.AI arXiv: https://arxiv.org/abs/2609.37233v1 Datalog underpins reasoning tasks such as program analysis, but its programs are hard to write. Existing synthesizers automate this task but require users to state their intent as input-output examples. Large language models (LLMs) suggest a more natural route, text-to-Datalog synthesis from a natural-language question, yet how well they do so has not been systematically evaluated. We present DatalogBench, a benchmark of 136 text-to-Datalog synthesis tasks curated from existing Datalog-based artifacts. Synthesized programs are graded by execution on held-out inputs against an oracle validated by mutation analysis. Across six LLMs and four prompting configurations, exact match peaks at 68.4%, and relation descriptions or an input-output example have only modest, model-dependent effects. Under direct prompting, most failures occur at compile time, typically because a model invents auxiliary predicates that it never declares or types consistently. Two coding agents reach up to 83.8% and eliminate nearly all such failures, leaving mostly semantic errors concentrated in recursive tasks. DatalogBench thus identifies recursive reasoning and decomposition as open challenges for current LLMs and agents, and offers a reliable, execution-grounded measure of both. --- ### Seeing Is Not Addressing: Auditing Linguistic Access to Frozen Visual Geometry Authors: Woosang Jeon, Jiwon Yang, Soo Chung, Taehyeong Kim Published: 2026-09-29 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.37230v1 Visual distinctions are often finer than those reflected in linguistic conceptualization. Vision-language models exhibit a similar asymmetry: a distinction can remain discriminable in frozen image geometry while being weakly addressable through the native text interface. We study this gap by separating visual discriminability from linguistic addressability in text-to-image retrieval. Using FactorAtlas, a fully crossed testbed of 23,040 images spanning shape, hue, pattern, and nuisance variation, we compare both readouts on held-out images of the same distinctions. We then derive image-side contrasts that separate each value from its alternatives for matched visual grounding, and test whether this reduces the native-text access gap across factors and models. Direction-specific and visual-absence controls tie these gains to the relevant visual contrast; the gains persist after global alignment and extend to compositional retrieval and natural images. Together, these results show that visual discriminability and linguistic addressability need not coincide, and that matched visual grounding can probe and reduce the resulting access gap. --- ### AESOP: Asymmetric Human-Camera Generation with Translation-Intensity Control Authors: Jingzhong Lin, Zhanke Wang, Heng Li, Wenxiang Liu, Zhao Zhang, Kecheng Tang, Dongdong Xiang, Changbo Wang, Di Kang, Chunchao Guo, Linchao Bao, Gaoqi He Published: 2026-09-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.37229v1 Human motion defines an action, while a camera trajectory determines how it is presented. Camera generation for a given human motion and joint human-camera generation are usually treated as separate tasks, although both share an asymmetric dependency: human motion can be generated independently, whereas the camera responds to the realized action. We introduce AESOP, a unified framework with an independent human pathway and a shared human-conditioned camera generator. Its asymmetric architecture serves both tasks while preserving the human output during camera generation. Although human context anchors the shot to the action and camera text describes its movement, translation intensity remains underspecified. We therefore construct trajectory pairs that differ in camera translation magnitude while sharing human motion and camera text, then use these pairs to learn an explicit intensity condition. Experiments on the PulpMotion dataset demonstrate strong camera distributional and framing quality in both tasks and effective control over camera translation intensity. --- ### Interacting particle guidance for sampling reward-tilted generative priors Authors: Adhithyan Kalaivanan, Zheng Zhao, Jens Sjölund, Fredrik Lindsten Published: 2026-09-29 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2609.37227v1 Inference-time steering adapts pretrained diffusion and flow-based models to new tasks, e.g., to generate samples from a conditional distribution or samples with desired properties, without retraining. This can be formalized as sampling from a reward-tilted generative prior. As exact sampling from this distribution is intractable, guidance-based methods rely on approximations producing biased samples, and sequential Monte Carlo (SMC) methods correct for this bias using importance weights. However, while exact in the large particle limit, SMC suffers from weight degeneracy and particle collapse in practice. We propose interacting particle guidance (IPG), which replaces reweighting with transport. The particles interact through an additional drift, derived from the Feynman--Kac PDE to cancel the reweighting term, and remain unweighted. Choosing the drift in a reproducing kernel Hilbert space yields a closed-form solution that is cheap to compute, with negligible overhead compared to SMC. We demonstrate the method on Gaussian mixtures with known posteriors, and on high-dimensional image inpainting and protein structure inference tasks. --- ### Follow the Entities: A Corpus Map for Agentic Search Authors: Soyeong Jeong, Sujay Kumar Jauhar, Sung Ju Hwang, Andrew Joohun Nam Published: 2026-09-29 Categories: cs.CL, cs.AI, cs.IR, cs.LG arXiv: https://arxiv.org/abs/2609.37226v1 Answering questions and completing tasks over large document collections often requires connecting evidence spread across multiple documents, such as a project's approval recorded in one, its requirements in another, and its latest status in a third. Recent LLM agents approach this by iteratively searching the full corpus rather than reading only a fixed set of top-ranked documents. However, when the corpus is exposed only as a flat collection of files, a relevant document gives no indication of how it relates to others, so the agent must rediscover these relationships for every query, often missing complementary evidence while simultaneously consuming substantial additional tokens. To address this, we introduce CorpusMap, a navigation layer that organizes the corpus around its recurring entities, which are identifiable from the documents themselves and can link a single document to many others across sources. Specifically, CorpusMap represents each recurring entity as an Entity Page that aggregates information about it and links to every document that refers to it, forming a graph between entities and documents that the agent can traverse to gather otherwise disconnected evidence. Moreover, since CorpusMap is constructed offline by resolving mentions of the same entity across documents, its links are shared across queries rather than rediscovered repeatedly at inference time. Using 7 different models with 3 benchmark datasets, we show that CorpusMap improves both evidence discovery and answer quality over raw-corpus agentic search while using fewer tokens on average, and further outperforms 4 alternative navigation layers, suggesting that entities serve as effective anchors for navigating large document collections. --- ### ResComEmb: Effective and Efficient Multimodal Embedding via Residual Homogeneity Compression Authors: Zijing Cai, Yuzhe Wang, Jingxian Zhu, Fengbin Zhu, Richang Hong Published: 2026-09-29 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.37225v1 Multimodal large language models (MLLMs) have shown strong potential for universal multimodal representation learning. However, existing methods either compress each input into a single vector, limiting fine-grained expressiveness, or retain long sequences of visual-token vectors, incurring substantial storage and interaction costs. To resolve this trade-off, we propose ResComEmb, a trainable framework for effective and efficient universal multi-vector multimodal embedding. ResComEmb first encodes each input at native dynamic resolution into ordered global, intermediate, and fine-grained views. After MLLM contextualization and embedding projection, a trainable Residual Homogeneity Compression (RHC) module reduces within-granularity redundancy and cross-granularity repetition under explicit visual token budgets. Then, ResComEmb introduces a length-adaptive Bidirectional Late-Interaction Matching mechanism for robust query-document scoring, which averages the strongest token-level matches in each direction and combines the two scores using a weight based on how many valid tokens each side has. Extensive experiments on MMEB, ViDoRe V1, and ViDoRe V2 show that ResComEmb produces higher-quality universal multimodal embeddings than VLM2Vec-V2, and outperforms ColQwen2.5 in visual document retrieval using only 37.5% of its full visual token budget, demonstrating a favorable effectiveness-efficiency trade-off. --- ### CredWise: A Controlled Agentic Decision-Intelligence Framework for Explainable and Auditable Credit-Risk Assessment Authors: Aakash Kumar Tiwari Published: 2026-09-29 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.37223v1 Credit-risk prediction is important in banking, but a prediction alone does not explain why an applicant is risky or how it should be combined with other evidence. This paper presents CredWise, a decision-support framework that integrates credit-risk prediction, probability calibration, explainable artificial intelligence, policy retrieval, SQL analytics, and controlled agent-based workflows. An XGBoost model is trained on Lending Club data (1,345,310 loans, 18 features) using a temporal split: 2007--2016 for training, 2017 for validation, and 2018 for testing. On the 2018 test set, the calibrated model achieved a ROC-AUC of 0.7109, PR-AUC of 0.2993, F1-score of 0.3714, and accuracy of 65.44\%. Calibration reduced the Brier score from 0.2157 to 0.1273 and the expected calibration error from 0.2862 to 0.0585. SHAP explanations were temporally stable, with a Spearman correlation of 0.9959 between 2017 and 2018 feature rankings. On 28 labeled queries covering nine policy sections, FAISS achieved the best Hit@1 (0.929) and MRR (0.964), while all three retrieval methods reached Hit@5 = 1.0. Agent routing achieved 95.6\% accuracy (43 of 45 cases), and the SQL benchmark scored 1.0 on exact-match, execution-success, and result-match across six cases. These results show that CredWise can combine predictions, explanations, policy evidence, and structured analytics in one controlled workflow. It is an academic research prototype, and final decisions remain with a human reviewer. --- ### OptiCom : A Unified Framework for State-Conditioned Composition in LLM-Driven Optimization Authors: Chenxing Wei, Sichen Liu, Lizhao Liu, Ningyuan Sun, Chen Bingzhou, Ying He, Bo Jiang, Fei Yu, Yao Shu Published: 2026-09-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.37221v1 Large language models (LLMs) are increasingly deployed to solve complex scientific and practical problems via iterative optimization. However, dynamically coordinating diverse search mechanisms as candidate quality, failure modes, and resource budgets evolve remains a critical open challenge. Targeted empirical diagnostics reveal that mechanism effectiveness is highly state-dependent. Motivated by this, we analyze how individual decisions drive final outcomes, decomposing the expected terminal improvement under a shared budget into cumulative decision opportunities minus cumulative selection losses. Guided by this opportunity-loss theoretical foundation, we propose OptiCom, a unified framework that represents LLM-driven optimizers within a shared configuration space: C=(A,Q,O,E,M,S), corresponding to artifact, query, operator, evaluation, memory, and strategy. Operating within this space, a fast LLM-based Optimization Controller dynamically composes immediate mechanisms through structured Action Packages, while a slower Strategy Adapter refines long-term selection preferences, operator weights, and templates based on accumulated trajectory feedback. Comprehensive evaluations across 32 benchmark groups demonstrate the superiority of framework: OptiCom achieves an average Max-score rank of 1.72 among 14 evaluated configurations, securing the top score in 23 groups. Ultimately, these results highlight the broad applicability and high extensibility of OptiCom as a general-purpose paradigm for robust LLM test-time scaling. --- ### Information Bottleneck-Guided Adaptive Hypergraph Transformer for Brain Disease Diagnosis Authors: Jingxi Feng, Xudong Chen, Yifan Zhang, Heming Xu, Hongcheng Han, Xijing Wang, Dong Zhang, Shaoyi Du Published: 2026-09-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.37220v1 Exploring high-order correlations and long-range dependencies in brain networks holds significant value for both neuroscience research and clinical diagnosis. However, previous studies have lacked a unified integration of high-order and long-range dependency information in brain networks, and there is substantial redundancy behind various types of information. These issues limit their effectiveness in the diagnosis of brain diseases. To address this, we propose an Information Bottleneck-Guided Adaptive HyperGraph Transformer (IBAHGT). By incorporating the information bottleneck (IB) principle, this approach enables adaptive learning of high-order correlations and both short- and long-range dependencies within a unified framework for brain network analysis, achieving high-precision brain disease diagnosis. IBAHGT consists of three key components: an information bottleneck-guided adaptive hypergraph convolution, which introduces a novel hypergraph information bottleneck (HIB) principle to adaptively learn hypergraph message-passing weights between nodes and hyperedges, optimizes information flow and captures high-order information in brain networks that is maximally informative and minimally redundant (MIMR). The Transformer encoder captures global information within brain networks through the attention mechanism, specifically modeling short- and long-range dependencies. An information bottleneck-guided node-level adaptive fusion employs the IB principle to learn independent weights for each node, facilitating the fine-grained integration of high-order information and global information to obtain an efficient representation for downstream tasks. Extensive experiments demonstrate that the proposed method outperforms current state-of-the-art methods and can identify biomarkers for clinical applications. ---