# Research Papers — 2026-10 · bundle page 73 of 156

> Full-text markdown of papers #3601–3650 of 7799 published in 2026-10, newest first. AI-readable mirrors of the full 7799-paper 2026-10 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-10/page/73
[← Previous bundle](https://aipod.dev/papers/archive/2026-10/page/72.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-10/page/74.md)

### From 2.1M Hours of Raw Broadcast Audio to Finnish Speech SSL: Confidence Filtering and Architectural Analysis at Scale

Authors: Yaroslav Getman, Tamás Grósz, Tommi Lehtonen, Mikko Kurimo
Published: 2026-10-09
Categories: cs.CL, cs.SD, eess.AS
arXiv: https://arxiv.org/abs/2610.13102v1

Broadcast archives offer scale and diversity for self-supervised learning (SSL) but contain substantial noisy, non-speech, and multilingual content. Prior work typically relies on voice activity detection or language identification (LID) to extract speech, ensuring presence rather than quality. Whether and how to filter unlabeled speech by quality for SSL, particularly from noisy broadcasts, remains underexplored. We present a large-scale Finnish SSL study starting from 2.1M hours of unconstrained broadcast audio. We propose automatic speech recognition (ASR) confidence as a proxy for pretraining data quality and show 3-7% relative WER improvements over LID-based filtering despite a smaller training set. We further address the lack of controlled architectural comparisons in speech SSL by evaluating Transformer, Macaron-Net, Conformer, and E-Branchformer at matched parameter budgets. Our models improve upon prior Finnish SSL baselines, particularly at Base scale and out-of-domain.

---

### VETR: Execution-Grounded Self-Improving Agent for Symbolic Graphics Reasoning

Authors: Yingbo Ma, Kuangshi Ai, Gordon Dai, Chenglong Yu, Chaoli Wang, Hua Shen
Published: 2026-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.13098v2

Visual agents for symbolic graphics reasoning depend on executable world representations, yet continued improvement remains bottlenecked by ambiguous supervision from final renders. Recent agentic systems refine scenes through repeated interaction, but they lack mechanisms to recursively extract actionable evidence from execution traces, overlooking which proposed symbolic edits survived, disappeared, or interfered during program execution. To address this, we propose VETR (Verified Execution Trajectories for Refinement), a framework that enables agentic recursive self-improvement over executable world representations for symbolic graphics reasoning. Within each task, VETR retains aligned program/render records with bounded runtime observations and introduces an execution refiner that compiles observed writes and state changes into constrained alternative programs. The refiner validates their endpoints and preservation conditions against captured runtime state; the ordinary visual verifier then assesses goal relevance. Repair construction makes no additional language-model calls, but can require additional executions. Evaluations on BlenderGym (280 tasks) and BlenderBench (27 released tasks) with GPT-5.6-luna and Qwen3.8-27B show that VETR consistently outperforms existing refinement strategies, including BlenderAlchemy and VIGA. Controlled ablations evaluate the contributions of execution refinement and role-asymmetric evidence organization under fixed generator-interaction caps.

---

### Stochastic Finite-Transition Learning and Exploration for Accelerated MRI Reconstruction

Authors: Muhammad Usama Mirza, Esin Öztürk Işık, Tolga Çukur
Published: 2026-10-09
Categories: eess.IV, cs.CV, eess.SP
arXiv: https://arxiv.org/abs/2610.13095v1

Generative MRI reconstruction maps an initial source distribution to the target image distribution along a learned recovery trajectory. Common generative methods, such as diffusion and flow, typically learn fine-grained local dynamics whose integration defines this global mapping, hence requiring many network evaluations at inference. Existing few-step sampling approaches can lower computational burden by approximating large transformations from local dynamics, albeit at the expense of reconstruction fidelity. FLEX-MRI instead partitions recovery into a small number of finite intervals between retained states defined by progressive changes in k-space support, and directly learns the reconstruction change across each interval. Because such finite transitions can admit multiple plausible outcomes, FLEX-MRI combines support-relaxed finite-transition learning with stochastic exploration. During training, randomized k-space sampling masks define finite-transition targets with varying support, while the prediction loss remains unrestricted by each individual mask. At inference, FLEX-MRI perturbs a random subset of currently estimated k-space coefficients, allowing the learned predictor to stochastically explore alternative transition paths. Each predicted state is then reconciled with the acquired measurements through data consistency. We theoretically show that this stochastic exploration induces nonzero transition-level variation through the learned recovery map and data consistency without requiring an auxiliary generative sampler. Experiments on multi-coil MRI datasets show that FLEX-MRI outperforms competing methods while maintaining computational efficiency comparable to few-step baselines.

---

### When Faster Decisions Do Not Make Better Agents: An Empirical Study of Jev in Agent Workflows

Authors: Yuchen Ling, Shengcheng Yu, Zhenyu Chen, Chunrong Fang
Published: 2026-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.13092v1

AI agents depend on repeated decisions to accomplish tasks, with large language models (LLMs) commonly serving as their core decision makers. The recent emergence of Jev offers an opportunity to make these decisions faster and at lower cost. However, it remains unclear whether such local savings can be translated into efficient and reliable task completion. In this paper, we investigate this efficiency transfer across five agents in two representative application settings, embodied tasks in ALFWorld and mobile app use in AndroidWorld. Across 4,500 task executions, we jointly evaluate task effectiveness, decision-level and task-level resource use, and execution behavior. Experimental results show that Jev integration substantially reduces decision cost and latency across all five agents, with task-level gains varying by workflow. Task success remains close to baseline levels on ALFWorld and declines more substantially in two mobile agents. Auxiliary reasoning and additional interactions absorb decision savings, while failed recovery and premature completion limit successful delivery. Mean cost per successful outcome increases in one agent even as spending per attempt falls. These findings highlight the importance of preserving progress, supporting recovery, and verifying completion when integrating efficient decision models into agent workflows. Their practical value should be assessed through successful task delivery, recovery behavior, and resources per delivered outcome.

---

### Volume Rendering with Metric--Phase Fields for Thin Structures and Sign-Ambiguous Geometry

Authors: Junkai Deng, Jiayi Kong, Xuhui Chen, Guanli Hou, Fei Hou, Junhui Hou, Wenping Wang, Ying He
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13088v1

Neural implicit surface methods that couple volume rendering with signed distance functions (SDFs), such as VolSDF and NeuS, achieve high-quality multi-view reconstruction for watertight objects. However, these approaches rely on a globally consistent inside-outside partition of space. This assumption becomes fragile for thin structures (e.g., sheets, tubes, and thin walls) and can break down for sign-ambiguous geometry (e.g., surfaces with boundary, non-manifold junctions, and closely spaced surface components), often resulting in spurious closures or loss of fine structures. Unsigned distance field (UDF) formulations relax the global sign requirement and better support non-watertight geometry, but are substantially harder to optimize. We propose PhaseSurf, a neural volume rendering framework built on Metric-Phase Fields (MPFs). MPFs decompose geometry into a nonnegative metric field capturing unsigned proximity and a continuous phase field providing a soft sign cue. By coupling the two through residual phase injection, MPFs yield a signed composite implicit function with stable near-surface gradients while retaining UDF-like robustness in sign-ambiguous regions. This design enables direct iso-surface extraction using standard Marching Cubes and avoids the failure modes of both pure SDF- and UDF-based formulations. We further introduce MPF-specific training losses that enforce metric consistency, stabilize phase transitions near the interface, and suppress spurious geometry in free space. Experiments on multi-view datasets with thin structures and sign-ambiguous geometry show that PhaseSurf produces more stable reconstructions and higher-quality geometry than UDF-based methods and recent 3DGS pipelines, while avoiding artifacts typical of SDF-based approaches.

---

### The Limits of Query Monitoring against Distributed Model Extraction

Authors: Snigdha Chandan Khilar
Published: 2026-10-09
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2610.13083v1

Model providers now face distillation campaigns run through thousands of accounts, each of which looks like an ordinary user. When network and payment metadata cannot link those accounts, the only evidence left is what they ask. We study what a provider can and cannot learn from those queries alone, using one million real chat turns for development and 17 later weeks that no design choice touched for a single frozen evaluation. The answer is set less by anything specific to extraction than by benign coordination: a campaign is visible only if it is more distinctive than the most distinctive group of ordinary users in the same week. We make this precise, show that a small but positive difference between campaign and users buys a detection time rather than detectability, and identify the campaign that is hardest to detect. Empirically, a calibrated monitor catches naive distributed extraction that per-account monitors miss (0.86-1.00 of weeks), and one variant - selecting accounts that move together after discounting how topical each one is - still catches campaigns padded with natural text (0.50-0.96 of weeks) where the other two spectral detectors catch none. Distilling each campaign into a small student shows that, in our testbed, campaigns that evaded every detector kept at most 42% of the capability of a detectable one. The limits are equally concrete: a campaign that ramps up slowly poisons every detector's calibration, benign evaluation pipelines look exactly like distillation, we found no deployable detector at colluder shares of 2.5% or below, and campaigns that relay real users' requests may be invisible to any query-only monitor.

---

### Human and LLM Facilitation in Collaborative Design Meetings: A Computational Behavioral Comparison

Authors: Mohammed Azeez Khan, Aaron Dsouza, Ashutosh Mishra, Ummul Faiz Z. Bibi, Neha K. Nair, Arshaan Khan
Published: 2026-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.13082v2

Large language models (LLMs) are increasingly being explored as facilitators of group discussion and collaborative work, but the behavioral differences between LLM-generated and human facilitation remain insufficiently characterized. We compare human Project Manager facilitation turns from scenario-based collaborative design meetings with Llama 3.1 8B Instruct responses across 199 paired contexts from the AMI Meeting Corpus. The analysis covers five confirmatory dimensions: contextual semantic distance, a lexical elaboration proxy, empathyand-hedge lexical density, topic divergence, and phase-vocabulary alignment. An exploratory directiveness measure failed domain validation and is not interpreted. The sampled human Project Manager turns were higher on contextual semantic distance (r = 0.644) and on empathyand-hedge lexical density (r = 0.203). The second contrast is carried mainly by hedge terms, so we read it as lexical hedging rather than psychological empathy. Llama-generated facilitation was higher on topic divergence (r = 0.703), the lexical elaboration proxy (r = 0.288; lengthconfounded), and instruction-conditioned phase-vocabulary alignment (r = 0.382). A secondary no-phase check left the Llama-higher phase contrast in place. The system prompt still requested a facilitation move, so that check is not evidence of spontaneous phase awareness. These differences should not be interpreted as a ranking of facilitation quality. The resulting behavioral profile provides a reference point for later comparisons of LLM facilitators, human facilitators, prompts, models, and collaborative settings.

---

### SWEAR: Section-Wise Efficient Audio Reasoning via Self-Rated Difficulty

Authors: Yixuan Li, Minghao Lv, Xuenan Xu, Zihao Zheng, Mengyue Wu
Published: 2026-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.13081v1

Recent work on large audio-language models (LALMs) explores adaptive reasoning to balance accuracy and efficiency. However, existing approaches often rely on coarse-grained difficulty estimation, overlooking the diverse factors that determine reasoning effort. We address this limitation by deriving five-dimensional difficulty scores and compressed reasoning traces from complementary offline assessments. Based on this characterization, we propose SWEAR, a difficulty-aware reasoning framework that learns difficulty signals during supervised fine-tuning and uses reinforcement learning to adapt reasoning allocation across reasoning stages. Experiments on MMAU-test-mini and MMAR show that SWEAR improves accuracy while substantially reducing generated tokens, shortening reasoning for easy instances without sacrificing performance on difficult ones.

---

### Attention Sees, MLP Refuses: Tracing the Attention-to-MLP Safety Pathway in LLMs

Authors: Zixuan Weng, Haoqi He, Weifei Jin, Yue Huang, Yang Bai
Published: 2026-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.13078v1

Safety-aligned large language models (LLMs) remain vulnerable to jailbreaks and can over-refuse benign requests. Understanding these failures requires distinguishing harmfulness recognition from the computation that writes refusal. We reframe this as a pathway question and make it measurable on residual-stream updates in decoder-only Transformers, separating the harmfulness information a module update exposes from the change it produces in readable refusal evidence. Across 18 Base-Instruct pairs spanning Qwen, Llama, and Gemma, harmfulness is decoded more readily from attention updates, particularly in early layers, whereas MLP updates contribute more strongly to harm-selective refusal writing in aligned models. Post-training strengthens MLP refusal writing in all 18 pairs without systematically improving harmfulness decodability, while positive MLP writing becomes more concentrated in deeper layers after post-training. Across six reasoning-enabled models, harmfulness remains decodable during reasoning while strong refusal writing concentrates at answer onset, a pattern that persists with reasoning disabled. Causal interventions further support this pathway: resetting downstream MLP updates to their clean values substantially reduces the refusal shift caused by upstream attention interventions. Building on these findings, we introduce Sentinel, which combines an attention-derived perception score P with an MLP-derived decision score D. Their joint profiles characterize jailbreaks and over-refusals, and conditional routing with both signals achieves a lower aggregate error rate than single-signal routing and evaluated fixed reminders across eight models.

---

### LongHarness: Advancing Agentic Reasoning over 100M-Token Corpora via Stateful Evidence Management

Authors: Songtao Huang, Zhijie Zhong, Xiaohan He, Shuaiyu Zhang, Zhiyin Yu, Li Kang, Heng Zhou, Yutao Fan, Xiangchao Yan, Bo Zhang, Shiyang Feng
Published: 2026-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.13075v1

Corpus-level long-context reasoning requires agents to acquire and integrate evidence across document collections that exceed their context windows. Existing agent harnesses enable incremental access to these corpora, but acquired evidence and evolving conclusions often remain entangled in a growing interaction history. This implicit state management can lead to two coupled failure modes: unbounded exploration, where retrieval expands the interaction history without resolving decision-relevant information gaps, and evidence-state drift, where previously acquired evidence or subsequent corrections fail to inform the current conclusion. We present LongHarness, a model-agnostic harness that makes the interaction between acquired evidence and evolving conclusions explicit. Its core mechanism, Evidence-Conclusion Coevolution, uses unresolved conclusions to guide exploration toward missing information, while newly acquired evidence strengthens or revises the conclusions that directed the search. LongHarness couples this process with Evidence Lifecycle Management, which keeps accumulated evidence current and traceable throughout the interaction. Together, these mechanisms focus retrieval on remaining information needs and keep conclusions aligned with current evidence. We evaluate LongHarness on OfficeQA Pro V1 and V2, AA-LCR, and LongBench v2 using GLM-5.3-Flash and Qwen3.8-Flash. Compared with direct reasoning using the same backbone, LongHarness improves GLM-5.3-Flash exact match on OfficeQA Pro from 15.8% to 60.2% on V1 and from 4.4% to 61.1% on V2. Against state-of-the-art agent harnesses, it surpasses Claude Code by 12.0 percentage points on V1 and Codex by 18.9 points on V2. With Qwen3.8-Flash, LongHarness reaches 61.7% on V1, outperforming Claude Code by 8.3 points and Codex by 12.0 points.

---

### Double-Take: Improving Capacity, Binding, and Selectivity In Long-Context Within Bidirectional Memory

Authors: Vakul Raghavan, Utkarsh Grover, Mingyang Mao, Chengwen Wang, Xiaomin Lin
Published: 2026-10-09
Categories: cs.LG, q-bio.GN, q-bio.QM
arXiv: https://arxiv.org/abs/2610.13073v1

Processing a long sequence does not guarantee that distant information remains usable. As context grows, a fixed-size memory must retain many associations, connect information across positions, and preserve relevant structure without being overwhelmed by everything it observes. These are distinct requirements: a memory may store more information without representing the relationships needed for retrieval, and it may transport those relationships without preserving the ones that matter. Thus, the challenge is not simply extending context length, but constructing a memory that remains usable as context grows. We study this problem in large-chunk test-time training (TTT) and identify two basic failures: single-step writes accumulate interference, limiting capacity, while self-associative writes cannot form cross-position bindings. We introduce Double-Take (DT), which repairs these failures with multi-step unpinned updates and cross-position mixing, and adds bidirectional change-sensitive writes for selective retention. Capacity alone leaves end-to-end recall at the $0.10$ chance floor; adding binding raises it to $0.92$. Trained only on $32\,\mathrm{kb}$ genomic contexts, DT reaches $0.927$ AUROC at $131\,\mathrm{kb}$ and $0.856$ near $1\,\mathrm{Mb}$, while transferring zero-shot from human to mouse. Together, these results identify capacity, binding, and selectivity as three requirements for usable long-range memory.

---

### AV-FlowEdit: Training-Free Prompt-Driven Audio-Visual Editing with Event-Aware Spectral Trajectory Preservation

Authors: Shaohua Pan, Xiao Liu
Published: 2026-10-09
Categories: cs.SD, cs.CV, eess.AS
arXiv: https://arxiv.org/abs/2610.13072v1

Joint audio-video diffusion models generate synchronized content, but editing existing clips requires target changes without disrupting non-target structure or cross-modal correspondence. We present AV-FlowEdit, a training-free and inversion-free framework for prompt-driven audio-video editing. Our key insight is that source--target flow differences provide both cross-modal editing directions and preservation cues. Under shared perturbations, they jointly update the video and audio branches, supporting visual-only, audio-only, and joint edits while retaining pretrained cross-modal interactions. To suppress drift accumulated during sampling, Cross-Modal Event-Aware Spectral Trajectory Preservation combines flow saliency, aligned audio activity, multiscale low-frequency displacement, and diffusion-stage scheduling. This adaptively preserves unsupported regions while allowing intended audio-visual events. AV-FlowEdit requires no model training, external masks, or optimization-based inversion. Experiments show strong prompt alignment and temporal consistency, together with competitive audio quality and cross-modal coherence.

---

### Evolving Creatives with Clicks: CTR-Driven Advertising image Editing via Attribute Disentanglement and Preference Alignment

Authors: Xu Zhang, Haohan Wang, Zheng Zhang, Yingjie Xia, Wei Feng, Honghe Zhu, Jingjing Lv, Changping Peng, Xuheng Qin, Zehua Zhang, Zheng Luo, Qicheng Li
Published: 2026-10-09
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2610.13066v1

The optimization objective for e-commerce advertising creatives has shifted from aesthetic quality to online business metrics like click-through rate (CTR). Recent methods generate appealing ad images from product images by constructing CTR-based reward signals and refining prompt generation via Reinforcement Learning. Such paradigms equally penalize all attributes within an underperforming image, even though some are actually beneficial to clicks. In this work, we propose Evolving Creatives with Clicks (ECC), which reformulates creative optimization as fine-grained editing rather than generation from scratch. ECC selectively refines attributes that hurt CTR while preserving those that contribute positively. Specifically, faced with a rich attribute space across massive images, ECC first identifies 28 attributes significantly correlated with CTR via ANOVA and Random Forest regression. Based on attribute contrasts between images with disparate CTRs, we instruction-tune a prompt model to generate editing suggestions, then refine it with real click preferences since reference-derived suggestions are not uniformly beneficial. Furthermore, we introduce Consistent Preference Optimization (CPO) as ECC's preference optimization strategy, which enforces consistency across semantically equivalent shuffled suggestions to mitigate DPO overfitting and improve preference generalization. Extensive experiments demonstrate that ECC significantly improves the CTR of ad images, outperforming state-of-the-art models by a substantial margin. The complete data pipeline, including attribute extraction and analysis, the resulting dataset, and source code will be released upon acceptance.

---

### CB-Orchestrator: Adaptive Workflow Optimization for LLM Agents via Contextual Bandits

Authors: Jiahang Sun, Zhiwei Shang, Zhipiao Liu, Hongwei Yang, Guoqing Xie, Shuang Qiu, Zhongxiang Dai
Published: 2026-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.13064v1

Large Language Model (LLM) agents have demonstrated remarkable capabilities in tackling complex tasks through agentic workflows. However, manually designing these workflows is labor-intensive and lacks flexibility to adapt to diverse queries. While automated workflow optimization is a promising direction, existing methods often incur expensive API costs during training-phase feedback collection. Furthermore, they fail to adaptively reuse high-quality workflows, which creates a bottleneck for both performance and cost-effectiveness. To address these limitations, we propose CB-Orchestrator, an adaptive framework that decouples workflow generation from selection. In the first stage, we construct a diverse pool of candidate workflows via evolutionary search. In the second stage, we formulate workflow selection as a Contextual Bandit problem, which enables sample-efficient learning by balancing exploration and exploitation, thereby significantly reducing the feedback required for training. During inference, the model adaptively selects the optimal workflow tailored to each query. Evaluations across five diverse benchmarks demonstrate that CB-Orchestrator consistently outperforms all baselines. Notably, on the MATH dataset, CB-Orchestrator achieves better performance than strong recent baselines while reducing training-phase API token overhead by no less than 40.55% and total end-to-end training (including workflow pool generation) token overhead by no less than 29.73%.

---

### KV-Repurpose: A Testbed for Evaluating Non-Prefix KV-Cache Reuse

Authors: Taiga Ikeda, Daisuke Miyashita, Jun Deguchi
Published: 2026-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.13060v1

KV-cache reuse in large language model serving is commonly limited to exact prompt prefixes, preventing reuse when the same retrieved document appears after different system prompts, conversation histories, or dynamic metadata. We present KV-Repurpose, an open-source, vLLM-based system for developing and evaluating segment-wise KV-cache reuse beyond exact prefixes. The system combines cache-aware segment scheduling, prompt construction, capacity management, workload generation, and runtime monitoring. Experiments on long-context benchmarks show that naive segment reuse largely preserves output quality for single-document question answering and summarization, while exposing clear limitations in cross-document reasoning. Under a Zipfian RAG-style workload in which 20% of requests share a fixed prefix, KV-Repurpose achieves a substantially higher token-level cache hit rate than prefix caching at the same 30 GB cache capacity, 63.9% versus 3.3%, resulting in lower TTFT. Even with 1 TB of cache capacity, prefix caching reaches only a 17.9% hit rate in this workload. Source code is available under the Apache-2.0 license at https://github.com/taiga333/eval-kv-repurpose.

---

### CliQ-RRG: Clinical-Knowledge Guided Disease-aware Visual-Textual Alignment for QA-Style Radiology Report Generation

Authors: Chandravardhan Singh Raghaw, Shahid Shafi Dar, Arnav Nishith Sharma, Shrey Modi, Nagendra Kumar
Published: 2026-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.13059v1

Radiology reports are essential for diagnostic reasoning and patient care, yet their manual preparation is time-consuming and cognitively demanding. Automatic radiology report generation (RRG) offers a scalable alternative, but existing models often produce lengthy, unstructured narratives that overlook diagnostic cues and multi-view information. We present Clinical-Knowledge Guided Disease-aware Visual-Textual Alignment for QA-Style Radiology Report Generation (CliQ-RRG), a unified two-stage framework for interpretable and clinically grounded reporting. In Stage 1, CliQ-RRG employs a Disease-aware Visual-Textual Alignment module that aligns image and text representations using predicted disease embeddings, reinforced by a Prior-Guided Attention Module (PrAM) to capture multi-view dependencies across current and prior scans. In Stage 2, GraphRAG-guided retrieval augments intermediate textual representations with graph-grounded clinical evidence, and a large language model restructures them into concise, interpretable question-answer (QA) pairs with diagnostic summaries. Experiments on two public chest X-ray benchmarks demonstrate that CliQ-RRG consistently outperforms prior methods across both lexical and clinical metrics, generating accurate and clinically coherent QA-style radiology reports. Code is available at: https://github.com/chandravardhan-singh-raghaw/CliQ-RRG.

---

### Travel Back To Move Forward: Retrospective Exploration and Evolving for Long Horizon Agents

Authors: Ziang Ye, Wentao Shi, Lei Chen, Zhengzhou Cai, Yu Wang, Zhiyuan Yao, Zihan Wang, Xierui Song, Fuli Feng, Yixin Cao
Published: 2026-10-09
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.13058v1

Large language model agents increasingly operate in long-horizon environments where early mistakes can trigger cascading failures and trap agents in dead-end trajectories. Agents must therefore be able to proactively recover from such states and turn their exploratory experience into guidance for subsequent decisions. We introduce ReTRAIL, a framework that enables agents to rewind the environment to earlier states while preserving valid progress and carrying forward knowledge learned from abandoned exploration. The environment state moves backward while the agent's knowledge continues to move forward. ReTRAIL trains a shared policy with two complementary signals: State-Anchored GRPO compares alternative continuations from revisited states to improve execution, while a branch-advancement reward guides reflection through the progress of subsequent exploration. Experiments across four interactive benchmark MineSweeper, Sokoban, ALFWorld and ScienceWorld show that simply adding a rewind primitive provides limited benefit and can even degrade performance. In contrast, ReTRAIL enables agents to rewind from unproductive states and use insights from abandoned branches to guide subsequent exploration. These results demonstrate the value of coupling environment rewinding with experience reuse within an ongoing task for long-horizon task solving. The code is available at https://github.com/whi497/ReTRAIL.

---

### CoBe: A Benchmark For Conversational Counterfactual Text Editing

Authors: Arvind Raghavan, Inwoo Hwang, Shreyas Havaldar, Kai-Zhan Lee, Aurghya Maiti, Yushu Pan, Jeffrey Wu, Mingxuan Li, Elias Bareinboim
Published: 2026-10-09
Categories: cs.CL, cs.AI, cs.ET, cs.LG
arXiv: https://arxiv.org/abs/2610.13057v1

Counterfactual text editing is the ability to revise a given text in light of a hypothetical change while preserving facts not causally affected by that change. When asked what would have happened if a doctor had not recommended surgery, the model output should retain the patient's medical history and modify only the consequences of the decision. This requires causal understanding: a model must determine, without being explicitly instructed to, what events are affected by the hypothetical change and what should remain fixed. This skill is central to human reasoning, especially in safety-critical domains like healthcare or law. By contrast, the task of associational text generation merely generates text statistically associated with the altered condition. In the medical example, an associational response might make the patient's symptoms less severe because milder cases correlate with no surgery. Such responses produce a plausible scenario, but not the counterfactual version of the given one. Prior counterfactual-text datasets do not clearly separate these modes of reasoning. We introduce the Counterfactual Benchmark (CoBe), a testbed for models' zero-shot ability to counterfactually edit a given textual scenario when prompted to do so in a conversationally natural way. CoBe contains 2000 questions spanning healthcare, science and engineering, business, and everyday scenarios. We test a suite of frontier and smaller models, with frontier systems achieving an average score of $58.05\%$. Our findings indicate that models frequently default to associational generation rather than preserving the causal structure of the original scenario. Moreover, performance does not scale consistently with either model size or output token length, suggesting that the failure may reflect model training objectives rather than insufficient model capacity.

---

### A4-MR: Abnormality-aware Anatomical Alignment for MR

Authors: Noel D'souza, Zhijian Yang, Nishank Singla, Pranay Doshi, Krisztian Koos, Fei Tan, Bruno Nunes, Laszlo Rusko, Bernadett Kolozsvari, Eszter Sabo, Emanuele Valeriano, Vikash Gupta et al.
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13056v1

Recent advances in large-scale foundation models have enabled strong generalization across a wide range of clinical and research tasks in magnetic resonance imaging (MRI). However, existing vision-language MRI models remain limited in anatomy-specific localization and pathology-aware representation, particularly in real-world settings. We introduce A4-MR, a 3D vision-language foundation model trained on approximately 50,000 MRI studies comprising 491,000 series, roughly half of which are paired with radiology reports. A4-MR is designed to: (1) learn anatomy localization through large-scale pretraining with noisy labels; (2) align image and text representations at the anatomical level; and (3) learn distinct representations of healthy and abnormal anatomy. Across a diverse set of downstream tasks, A4-MR outperforms existing foundation models, including a +1.7 AUROC improvement over MedSigLIP for abnormality detection. Using a frozen encoder, A4-MR also surpasses an end-to-end task-specific MedNeXt model for knee segmentation by +1.4 Dice and performs comparably to the best task-specific model for abdominal organ segmentation on AMOS, within 0.4 Dice. Finally, without anatomical prompts, A4-MR localizes 3D anatomy comparably to MedSAM2 provided with ground-truth bounding boxes. Together, these results demonstrate that anatomy-level multimodal alignment and abnormality-aware representation learning can enable a single MRI foundation model to support localization, segmentation, and pathology-related tasks.

---

### UniMergeTrack: Hierarchical Model Merging for General RGB-X Visual Object Tracking

Authors: Shiao Wang, Xiao Wang, Yuhe Ding, Jiarui Xu, Zikang Yan, Bo Jiang, Yonghong Tian, Jin Tang
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13053v1

RGB-X tracking leverages complementary modalities, such as thermal infrared, event, or depth, to improve robustness under challenging conditions. Existing general-purpose RGB-X trackers typically follow two paradigms: independent task-specific training and joint training across modality combinations. However, the former suffers from knowledge isolation and model redundancy, while the latter introduces optimization interference across heterogeneous tasks. In this work, we present UniMergeTrack, a new model merging paradigm that preserves task-specific optimization while enabling effective cross-task knowledge integration. UniMergeTrack first trains modality-specific trackers independently and then performs hierarchical model merging. Specifically, transferable backbones are merged across RGB-X tasks via high-pass-filtered merging, and complemented by lightweight task-specific experts that preserve modality-specific cues. Tracking heads are merged within each task using optimal transport-guided channel alignment to mitigate hidden channel mismatch. A subsequent distribution alignment fine-tuning stage further improves compatibility between the unified backbone and task-specific heads. Extensive experiments on the LasHeR, RGBT234, VisEvent, VOT-RGBD22, COESOT, and CDTB benchmarks demonstrate that UniMergeTrack achieves state-of-the-art or competitive performance across RGB-Thermal, RGB-Event, and RGB-Depth tracking tasks, while exhibiting strong generalization to unseen datasets. The source code will be released upon acceptance. The source code of this paper will be released on https://github.com/Event-AHU/OpenEvTracking.

---

### Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors

Authors: Arnab Dey, Mohammad Mahfuzul Islam
Published: 2026-10-09
Categories: cs.CR, cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2610.13051v1

Some backdoor attacks never touch an image. FLIP only changes labels: on CIFAR-10, flipping 3% of them makes a ResNet-32 send 99.7% of triggered source-class images to the attacker's class. The usual defenses look for the poison inside a model trained on those same labels. So when they struggle, is the problem the test they run, or the space they run it in? We answer with a controlled swap: we keep the test, the labels and the metric fixed, change only the feature space, and judge each defense by retraining the victim on what it keeps. The space decides. On dog$\to$cat the same nearest-neighbor check scores AUROC 0.546 on the poisoned model's features, 0.967 on frozen DINOv2 features that never saw a label, and 0.995 on the same network trained on clean labels, an oracle control rather than a defense. We reproduce SPECTRE, the defense FLIP's authors recommend, to 90.94 / 0.97 against its published 91.47 / 1.7, then price it: on truck$\to$deer it deletes 20% of every class, while a prototype audit on the frozen features deletes 6% for higher clean accuracy and lower attack success. On three harder pairs that audit leaves about a quarter of the attack standing. A count read off encoder geometry and clean labels, before any attack is run, separates the pair where the audit removes the attack from those where it only partly does, though on a single zero-count pair. Three results warn against the easy story. A better detection score does not mean a safer model: on two of four pairs Confident Learning scores the higher AUROC, and on dog$\to$cat leaves 2.6 times the attack success. The prototype family, our best static performer, is the first an adaptive attacker slips past. And past a point, catching more poisons stops reducing attack success. Check labels in a space that never learned them, and judge a defense by the retrained model, not by its detection score.

---

### NOFACE3D: NOise-scheduled Focused Anchoring and Controlled Enhancement for 3D Generation

Authors: James Rose, Jingwen Liang, Wenchao Ma, Oren Jacob, Haomiao Jiang
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13050v1

Recent image-to-3D models generate complete, high-quality assets from a single image, with global shape and appearance that are already convincing; what remains is the fine scale, where small, thin, elongated, or detail-rich regions are reconstructed softly and under-resolved. This gap can arise in generators with fixed-resolution spatial grids: the same spatial sampling density is used for a large, smooth, low-curvature surface and for a thin, high-curvature structure that carries most of the perceived detail. To this end, we introduce NOFACE3D, a training-free framework that refines these detail-critical regions while preserving the rest of the asset. Given a generated mesh, its input image, and a refinement box, NOFACE3D extracts the selected submesh, scales it up to fill the generator's input volume, and re-encodes it into the full-resolution latent grid. Conditioned on the image crop, it refines the region through flow-based sampling with a spatially varying noise schedule that anchors boundary latents. The refined region is then integrated back into the original asset, preserving the surrounding geometry and appearance. We implement NOFACE3D on TRELLIS.2 and its sparse voxel representation; all of it operates purely at inference time on the frozen pretrained generator. We evaluate NOFACE3D on PAniC-3D, TexVerse and BigDetail, a new benchmark of 339 assets with annotated detail-critical regions. On all three, it generates sharper geometry and finer texture detail than the baselines. Project page: https://idkcouscous.github.io/NoFACE3D/

---

### Redirecting Teacher Supervision: Temporal-Response-Guided Latent Distillation for Video-Language Models

Authors: Xinkai Lu, Junli Zha, Haolan Wei, Jiahui Wang, Chao Zhou, Xiang Song
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13045v1

On-policy distillation (OPD) transfers capabilities through teacher predictions on student-generated sequences, as studied by GKD and MiniLLM for language models and Video-OPD for video tasks. We ask how internal supervision can further support temporal learning. Direct hidden-state alignment does not explicitly use the student's temporal responses to choose a supervision direction. We propose Temporal-Response Guidance (TRG), which measures representation responses to local temporal perturbations, adds the teacher correction's projection onto their span back to the original correction, and restores its original norm. The projection is not normalized separately, and correction components outside the response space are retained. With an orthogonal tangent projector and a detached target, TRG changes the supervision direction while preserving the cosine-loss value and aligned-representation gradient norm at target construction. On TempCompass, Video-MME, MVBench, and Video-MMMU, TRG improves the 4B student's macro accuracy by 2.86 percentage points over the reported output-only OPD baseline.

---

### RAMP: Recurrent Adaptive Mixed Precision for Efficient Prefill in Recurrent-Depth Language Models

Authors: Ke Wan, Chen Chen
Published: 2026-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.13041v1

Recurrent-depth language models improve inference-time reasoning by repeatedly applying a shared network to refine internal representations, but the additional recurrent computation also increases inference cost. This overhead is particularly pronounced during long-context prefill, where the full input sequence is repeatedly processed across recurrent depth. Weight-and-activation quantization offers a natural way to accelerate this computation through lower-precision arithmetic, yet repeatedly applying quantized computation introduces numerical perturbations that accumulate along the recurrent trajectory. We systematically analyze how such errors propagate across recurrent depth and show that quantization errors introduced earlier are attenuated by more subsequent recurrent steps and therefore have a smaller effect on the final representation. Matched-budget experiments strongly support this prediction, showing that placing quantized computation earlier in recurrent depth consistently yields the best fidelity. Based on this insight, we introduce RAMP (Recurrent Adaptive Mixed Precision), which performs early recurrent refinement using low-precision weight-and-activation computation and restores high precision for terminal refinement. Across context lengths up to 4K, RAMP preserves near-full-precision prefill quality while achieving up to 1.16x end-to-end first-prefill speedup on an NVIDIA H100, corresponding to a 13.9% latency reduction.

---

### Lightweight Time-Frequency Vibration Sensing for Consumer Device Condition Monitoring

Authors: Po-Heng Chou, Jen-Yu Chiu, Chun-Yu Yeh, Wei-Lung Mao, Walid Saad
Published: 2026-10-09
Categories: eess.SP, cs.LG
arXiv: https://arxiv.org/abs/2610.13040v1

Consumer rotating devices, including washing machines, air conditioners, electric fans, robotic vacuum cleaners, and home fitness equipment, increasingly rely on vibration sensing for intelligent condition monitoring and predictive maintenance. Ensuring reliable operation in these resource-constrained appliances requires efficient, lightweight artificial intelligence (AI) solutions capable of processing non-stationary signals under real-world operational interference. This paper proposes a lightweight AI-assisted vibration sensing framework that employs continuous wavelet transform (CWT) and Hilbert-Huang transform (HHT) to convert one-dimensional (1-D) vibration time-series signals into two-dimensional (2-D) time-frequency representations. These representations are subsequently analyzed using lightweight deep learning architectures, specifically MobileNetV3-Small, ConvNeXt-Tiny, and the You Only Look Once version 11 classification model (YOLOv11-cls). Public bearing benchmark datasets, including Case Western Reserve University (CWRU), Paderborn University (PU), and Intelligent Maintenance Systems (IMS), are utilized as representative rotating components to investigate model generalization. The proposed framework is designed with eventual embedded deployment in consumer electronics in mind. Experimental results demonstrate that CWT combined with YOLOv11-cls achieves high diagnostic accuracy, while MobileNetV3-Small provides an attractive low-latency alternative for computationally constrained edge platforms. Conventional random splitting is also compared with dataset-specific structured isolation. The results highlight the importance of avoiding data leakage and quantify tradeoffs in classification accuracy, parameter count, floating-point operations (FLOPs), model footprint, preprocessing cost, and central processing unit (CPU) inference latency.

---

### Towards Easier-to-Tune Training Algorithms: New Results from AlgoPerf

Authors: Priya Kasimbeg, Ahmed Khaled, Aahlad Chandrabhatta, Yifan Wang, Andy Liu, Niccolò Ajroldi, David Tweedle, Xingyu, Yang, Isaac Patole, Frank Schneider, George E. Dahl
Published: 2026-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.13036v1

Evaluating deep learning optimizer is challenging because confounding factors such as hyperparameter tuning and optimizer-model co-design often obscure true algorithmic performance. We present results of the updated AlgoPerf V1 benchmark, in which we evaluate a new set of training algorithms on wall-clock time-to-target across nine workloads on fixed hardware. This report focuses on the results of the benchmark's self-tuning ruleset, which requires submissions to use a single configuration across all workloads. Out of the 13 submissions in PyTorch and JAX, Schedule-Free AdamW ranked first. In addition to the wall-clock scores, we also analyze the trade-offs between step efficiency and wall-clock execution, revealing that matrix-preconditioned and two-loop methods achieve good sample efficiency but require more aggressive computational optimizations to be competitive in runtime. Finally, our hyperparameter saliency and retention analyses show that the learning rate and momentum are highly sensitive across domains and algorithms, highlighting the need for methods that are more robust against hyperparameter tuning and can generalize well to new tasks.

---

### GeoGR: Spatial-Semantic Token Compression and Group Routing for Efficient 3D VLMs

Authors: Yongxin Zheng, Hao Wu, Fang Li, Hu Fu, Peiran Yin, Xudong Wang, Haozhe Hu, Xinghao Chen, Yunpu Ma, Wei Zhang, Xiaoyu Shen
Published: 2026-10-09
Categories: cs.CV, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.13035v1

Multi-view representations enable pretrained image and video vision-language models to understand 3D scenes, but their long visual token sequences make inference costly. Redundancy arises both from repeated observations of the same scene regions across views and from processing all visual tokens at every LLM layer, regardless of their relevance to the current computation. We propose GeoGR, a two-stage framework that jointly compresses scene representations and visual computation. At the projector level, GeoSemZip consolidates cross-view observations into 3D voxel tokens and combines semantic saliency with spatial coverage to construct a compact scene representation. Unselected voxel features are aggregated into contextual anchors to retain scene information that would otherwise be discarded. At the LLM level, we identify an effective visual computation window and observe that query-conditioned token relevance remains locally stable but changes over longer layer intervals. Guided by these findings, GroupRoute combines late entry and early exit with recoverable group-wise skipping. Low-relevance tokens bypass complete LLM layers while retaining their hidden states and can rejoin the active group at subsequent anchor layers selected under a computation budget. For compression-aware post-training, we construct a compact dataset of 67K samples by retaining 30% of the original training pool. With this protocol, GeoGR retains 95.2% and 96.8% of the dense post-trained models' performance on LLaVA-OneVision-7B and Video-3D LLM-7B, respectively, while accelerating LLM prefill by $6.8\times$ and $6.1\times$. These results demonstrate the consistent effectiveness of GeoGR across 3D VLMs with and without explicit 3D spatial encoding, highlighting a favorable performance-efficiency trade-off. The code is available at https://github.com/EIT-NLP/GeoGR.

---

### Target-Constrained Pooling for Trajectory Extrapolation

Authors: Ning Wang, Claire Cardie, Sainyam Galhotra
Published: 2026-10-09
Categories: cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2610.13034v1

Trajectory extrapolation from a partial target trajectory is underdetermined, but completed related trajectories provide information about the unresolved continuation. We propose target-constrained pooling (TCP), which separates the finite-horizon endpoint from continuation shape and estimates that endpoint from two complementary source signals. Uniform pooling captures final response levels, while matched-increment pooling captures the remaining change after aligning source and target states. TCP selects representative sources, projects both endpoint estimates onto a target-consistent region, converts them into corrections of the free forecast, and controls their influence through strength calibration and degradation-aware weighting. Across 25,245 targets from five domains and both forecasting directions, TCP reduces mean RMSE by 35.8% to 42.1% for all ten target-only forecasters and improves all 100 forecaster, dataset, and direction comparisons. The two endpoint signals consistently outperform either alone, while representative selection remains within 0.64% to 1.02% of the full donor pool using 12.7x fewer donor inputs and 138.8x fewer matching distance computations.

---

### Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning

Authors: Minjong Cheon
Published: 2026-10-09
Categories: cs.LG, physics.comp-ph
arXiv: https://arxiv.org/abs/2610.13028v1

Neural operators that learn physical laws from data should generalize beyond their training distribution and expose the learned law in closed form. The Kolmogorov-Arnold neural operator achieves both for position-dependent dynamics by learning a dense pseudo-differential symbol over all positions and frequencies, at a cost that grows with the square of the resolution and with closed forms that must be read out by hand. This dense symbol is unnecessary: a finite-order variable-coefficient differential operator has a symbol whose separation rank is bounded by the number of its coefficient functions, independently of the resolution, and the symbols that this network learns on its own benchmarks have rank two. We introduce the Kolmogorov-Arnold flow (KAFlow), which learns the generator of a physical operator as a low-rank symbol built from univariate Kolmogorov-Arnold functions, fixes the gauge of these functions so that every term has a unique place, integrates the dynamics with exact exponential integrators, and distills its closed form automatically. On the synthetic benchmarks of the Kolmogorov-Arnold neural operator, KAFlow lowers the out-of-distribution loss by 3.4 to 11 times and recovers every coefficient to within 10^-4 without symbolic templates, reaching the precision floor of the data. On long-horizon quantum dynamics, it lowers the state infidelity by up to 6 orders of magnitude, and by 30 times when trained only on position and momentum measurements. KAFlow also learns a variable-order fractional operator 15 times more accurately than existing operator networks and discovers nonlocal, fractional and non-polynomial dispersion relations on which sparse regression fails.

---

### Dataset-Specific Encoding of Image Content Underlies Cross-Dataset Failure in Brain MRI Foundation Models

Authors: Minh Sao Khue Luu, Bair N. Tuchinov
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13027v1

Frozen features of brain MRI foundation models lose accuracy on new datasets, a failure usually blamed on the features encoding where each scan came from. We show that the failure lies elsewhere. Across 17 public datasets and eight frozen representations, every representation recognizes the MRI sequence within each dataset, yet most fail on an unseen one. The cause is neither dataset encoding nor a shift of each dataset: the content itself is encoded along different directions in each dataset. This explains which datasets fail, for the sequence and for sex, and why removing the dataset signal with INLP, LEACE, CORAL or adversarial training never improves sequence transfer. Only the broadly pretrained 3DINO and SAM-Med3D encode the sequence consistently enough to transfer; the brain-specific BrainIAC and BrainFM do no better than an untrained network. Code is available at https://github.com/BrainFM/brainfm-profiler/tree/main/representation_utility.

---

### Evo$^2$: Evolving the Evolver through Weight-Harness Co-Evolution

Authors: Chejian Xu, Shujian Zhang, Wenxuan Zhou, John Lambert, Beck Chen, Bo Li, Andrew Hard, Rajiv Mathews
Published: 2026-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.13025v1

Self-evolving agents are increasingly used to improve language model systems without human intervention through either weight updates or harness improvements. Existing approaches typically pursue these directions separately: some methods update model weights using self-generated experience, while others rewrite the harness that defines how an agent uses prompts, tools, context, and execution feedback. However, these approaches overlook the relationship between model weights and the agent harness. To connect these processes and ensure effective and smooth agent self-evolving, we propose Evo$^2$, a framework for weight-harness co-evolution. For harness updates, we design a unified dual-role pipeline that a single policy serves as both a Proposer, which writes harness edits, and a Reflector, which diagnoses their effects and guides revision. For weight updates, harness modifications and their execution outcomes guide the training of both the Proposer and the Reflector. We further develop EvoTIDE, a training pipeline that recovers usable trajectories through edit extraction and prefix continuation, then combines SFT with interleaved GRPO across the two roles. Extensive experiments across three coding benchmarks demonstrate that Evo$^2$ significantly outperforms existing self-evolution methods. It achieves 82.1% on Polyglot and 48.3% on SWE-bench Verified, compared with 62.6% and 41.0% for the strongest baseline, respectively. We also show that the evolved harnesses transfer to different tasks without further evolution and rank first in cross-benchmark evaluations. Our dual-role harness improvement framework is also able to extend to self-evolution with proprietary models, improving a GPT-5-mini agent's performance from 24.4% to 48.3%. These results demonstrate the applicability of the framework Evo$^2$ across domains and its broad transferability across models and tasks.

---

### What Visual Representations Do World Action Models Need?

Authors: Kai Liu, Jiangran Lyu, Tingyu Deng, Yixin Zheng, Jingtian Wan, Wenbo Cui, Mi Yan, Yizhou Zhou, Haoran Li, Dongbin Zhao, Zhizheng Zhang, He Wang
Published: 2026-10-09
Categories: cs.CV, cs.RO
arXiv: https://arxiv.org/abs/2610.13024v1

World Action Models (WAMs) inherit powerful priors from pretrained video generation models, but their variational autoencoder (VAE) representations are optimized for visual reconstruction rather than manipulation. External visual representations can improve manipulation, but existing methods couple representation choices with different integration mechanisms, obscuring their individual effects. We systematically compare visual representations, five integration mechanisms, and feature combinations under controlled settings. Our study yields three main findings. First, strong general-purpose visual representations provide broader robustness gains and can match or outperform representations pretrained for specific visual capabilities. Second, channel-wise joint prediction achieves the highest average success rate; further analyses suggest that future-feature prediction and the action expert's effective use of added visual information contribute to its gains. Third, additional sources should be selected for the task-relevant information they contribute beyond existing representations, as the same source can help one combination but harm another. However, even the best evaluated composition provides a smaller additional gain than the initial improvement from adding a strong general-purpose representation to the VAE-only baseline. We instantiate the resulting Representation Recipe as R$^2$-WAM. It achieves state-of-the-art performance on RoboCasa GR1 Tabletop with an 84.7% success rate and reaches 92.8% on RoboTwin 2.0, improving over the VAE-only baselines by 5.9 and 2.2 percentage points, respectively. Taken together, these findings suggest that future work should prioritize stronger general-purpose visual representations and more effective mechanisms for integrating them into WAMs, with feature composition serving as a supplementary route to improvement.

---

### Unified Generative Video Compression with Autoregressive and Bidirectional Video Diffusion

Authors: Tianyu Zhang, Shuai Huo, Dengchao Jin, Hewei Liu, Meng Lei, Tianwen Wu, Jiawen Gu, Chao Zhou, Li Li, Dong Liu
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13023v1

Video diffusion models can reconstruct visually compelling content from severely compressed representations at ultra-low bitrates, but they impose a rigid deployment trade-off. Bidirectional (BI) models achieve high quality through dense full-sequence attention, yet require the entire sequence before reconstruction and incur substantial latency. Autoregressive (AR) models enable low-delay streaming, but their restricted temporal context reduces compression efficiency. This tension is particularly relevant when a video stream must be decoded in real time yet is also stored for later, quality-oriented playback, making AR and BI models natural choices for the live and replay stages, respectively. To support both stages, we present UniGVC, a Unified Generative Video Codec that supports two decoding modes from a single bitstream. UniGVC-BI employs a BI-DiT with dense attention for quality-oriented random-access decoding, whereas UniGVC-AR uses windowed AR-DiT for low-delay streaming. Both paths share a hierarchical latent codec that recurrently produces the same conditions with ultra-low bitrates, allowing UniGVC-BI to serve as a conditional teacher for single-step distillation of UniGVC-AR. A tiny conditional video decoder further reduces streaming latency while preserving most of the reconstruction quality. On standard 1080p benchmarks, UniGVC-BI achieves the best overall rate-distortion-perception trade-off among the evaluated methods and produces temporally stable video at rates as low as 0.0005 bits per pixel. Meanwhile, UniGVC-AR remains competitive while decoding 1080p video at 36.85 FPS, over $250\times$ faster than UniGVC-BI. Code and models will be released later.

---

### Don't Guess, Just Ask: Evaluating Interactive Requirement Clarification for Repository-Level Software Development

Authors: Mingyu Chen, Zhiqing Zhong, Manyi Wang, Linxi Liang, Rouyu Chen, Pinjia He
Published: 2026-10-09
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2610.13022v1

As LLMs become increasingly capable of implementing software, a critical bottleneck emerges: whether they truly understand what users want. Users often begin with vague intent, leaving important requirements to be clarified through interaction. In repository-level development, misunderstandings can propagate and cause costly rework. However, existing benchmarks typically evaluate implementation from complete specifications or underspecified requests, overlooking pre-coding requirement clarification. Prior clarification studies mainly focus on function-level tasks. It therefore remains unclear how well LLMs can transform vague intent into clear requirements through interaction and how this capability affects subsequent code generation. We introduce ReqClarifyBench, an interactive requirement clarification benchmark for repository-level software development. It contains 341 tasks spanning game, miniapp, and web development, each with a vague user intent, a human-validated requirement document, and code evaluation artifacts. We employ LLM-based user simulators with diverse personas derived from real-world conversations. We evaluate six LLMs in terms of clarification effectiveness and generated requirement document quality. Results show that stronger clarification effectiveness does not necessarily translate into higher document quality, and performance varies substantially across user personas. We also find that coding-agent harnesses do not further improve requirement document quality compared with their base LLMs. These findings offer insights for building LLM-based end-to-end software development systems.

---

### EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agents

Authors: Jie Huang, Yanan Chen, Ruixun Liu, Kaichen He, Xinyi He, Rui Huang, Zilong Zheng, Zhenliang Zhang, Yiwu Zhong
Published: 2026-10-09
Categories: cs.AI, cs.CL, cs.CV
arXiv: https://arxiv.org/abs/2610.13021v1

Continual learning requires models to accumulate past experience and transfer it to new tasks. Despite substantial research on continual learning, existing work remains largely confined to non-embodied settings, without extending to embodied environments. To address this gap, we introduce EmbodiedCLBench, the first benchmark in embodied simulation for evaluating the continual learning capability of self-evolving agents. With compositional generalization as the core evaluation principle, our benchmark evaluates whether agents can first learn from basic tasks and then recombine the acquired skills to solve complex, unseen advanced tasks. Based on EmbodiedCLBench, we conduct extensive experiments across multiple agent harnesses and learning methods, revealing consistent limitations and distinctive behaviors. Our results show that while advanced tasks can be hardly solved under zero-shot setting, the experience from basic tasks enables consistent improvement. Such benefit from the experience remains stable even if the task complexity increases. However, this improvement does not scale accordingly with the number of learning samples, and current agents cannot extract meaningful information from additional experience. Collectively, our findings highlight continual learning as a fundamental yet underexplored capability, and our benchmark offers valuable resources for designing effective self-evolving agents in embodied environments. Our code and dataset are available at https://pku-value-lab.github.io/EmbodiedCLBench-homepage.

---

### Trajectory-Aware Policy Optimization: Closing the State Distribution Gap in Long-Horizon LLM RL

Authors: Zhicheng Cai, Weinan Dai, Yuyan Zhou, Wei-Ying Ma, Ya-Qin Zhang, Hao Zhou
Published: 2026-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.13020v1

Reinforcement learning (RL) has become a dominant paradigm for enhancing LLMs' reasoning capabilities. Yet long-horizon autoregressive rollouts introduce a fundamental limitation: local policy mismatch can compound into severe state distribution shift over trajectories. Existing LLM RL methods, such as GRPO and GSPO, control policy mismatch at the token or sequence level, but overlook this trajectory-induced shift, causing training instability and eventual collapse as the horizon grows. To close the state distribution gap, we propose Trajectory-Aware Policy Optimization (TAPO), a principled framework that theoretically derives the state-distribution correction from the off-policy policy gradient and incorporates it into the RL objective. TAPO further employs a geometric-mean estimator for variance reduction, together with a length-aware dynamic clipping that prevents overly deviated updates across the trajectory. Extensive experiments demonstrate that TAPO significantly surpasses existing LLM RL algorithms across eight competition-level benchmarks, achieving up to 48% and 12% improvement over GRPO on mathematical reasoning and algorithmic coding, with substantially enhanced training stability.

---

### FlyHack: Stress-Testing Mechanistic Claims in a Connectome-Constrained Visual Model

Authors: Aashish Dhawan
Published: 2026-10-09
Categories: q-bio.NC, cs.CV
arXiv: https://arxiv.org/abs/2610.13018v1

Connectome-constrained neural networks make biological components directly accessible to intervention, but their mechanistic conclusions can still depend on the comparison model, intervention, circuit representation and trained solution. We stress-test these dependencies in flyvis, a \emph{Drosophila} motion-vision model, using its 50-model ensemble, task-matched causal CNN controls and replicated interventions. At the level of model behaviour, three-level reverse-phi reversal in flyvis is strongly period- and velocity-dependent and also appears in task-matched CNNs; neither family reproduces the fly's stationary-sawtooth response. Adversarial direction reversal requires a perturbation bound of 0.21--0.27 times stimulus contrast in flyvis versus 0.13--0.25 in CNNs, but the gap over a parameter-matched CNN largely closes for larger CNNs. At the level of mechanism, matching attribution to the intervention matters: for held-out-mean replacement, first-order edge attribution patching reaches pooled AUROC 0.970 and Spearman 0.872, versus 0.933 and 0.126 when its zero-aligned score is evaluated against the same ground truth. Circuit conclusions also vary with representation and ensemble member. Greedy cell-type sets meet a model-relative T4/T5 tuning criterion in 25 of 50 models, while T4/T5 output silencing reduces the tested optic-flow behaviour to at most two of eight clips in 40. In six models where the cell-type search first succeeds, the corresponding connection sets can be pruned to 13--31 added classes; under an additional two-sided response-amplitude check, only two of those six first-success cell-type sets still meet the tuning threshold. These results support a narrower use of connectome-constrained models: they make mechanistic hypotheses directly testable, but the resulting claims must be scoped to the model family, intervention, representation and trained solution.

---

### Generalized Proximal Policy Optimization: Unifying LLM RL Algorithms from the Perspective of Divergence

Authors: Zhicheng Cai, Xinyuan Guo, Hanlin Wu, Weinan Dai, Yuyan Zhou, Wei-Ying Ma, Ya-Qin Zhang, Hao Zhou
Published: 2026-10-09
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.13015v1

Reinforcement learning has become the foundational paradigm for enhancing the reasoning capabilities of LLMs. Yet the dominant approach, Proximal Policy Optimization (PPO), relies on a geometrically degenerate clipping proxy that cannot adapt its resistance to policy shifts, leading to premature clipping and degraded exploration. Subsequent variants alleviate these issues through fragmented and heuristic modifications. We propose Generalized Proximal Policy Optimization (GPPO), a principled and unified framework for understanding and designing LLM RL algorithms through the lens of divergence. We establish a generalized $f$-divergence trust-region theory with a monotonic performance improvement guarantee, and show that different $f$-divergences induce distinct geometries and policy update dynamics. We further introduce novel estimator topologies that provide more faithful divergence estimates than the conventional single-sample Monte Carlo estimation. Under this framework, PPO and recent variants such as DAPO, DPPO, and RIPO emerge as special cases, establishing a principled two-dimensional design space of {Divergence $\times$ Estimator}. Extensive experiments on eight competition-level math and coding benchmarks show that both dimensions are critical. The best-performing GPPO instantiation surpasses standard GRPO by up to 50% and 14% on math and coding, delivering SOTA performance.

---

### Evidential Multimodal Transformers with Dempster-Shafer Fusion for Emotion Recognition

Authors: Muhammad Usama, Yeoreum Jung
Published: 2026-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.13013v1

Softmax-based multimodal emotion recognition methods often assign high confidence even when sensor data are corrupted or contradictory, for example when facial expressions suggest positive valence while vocal prosody indicates negative arousal. We present an Evidential Multimodal Transformer with Dempster-Shafer Fusion (EMT-DS) that addresses this problem through formal uncertainty quantification. EMT-DS models predictions as Subjective Opinions (Dirichlet distributions), separating epistemic uncertainty (lack of evidence) from aleatoric uncertainty (sensor noise). Modality-specific transformer encoders produce belief mass distributions that are fused with Dempster's Rule of Combination, which normalizes the joint belief mass by a conflict coefficient 1/(1-c), so that contradictory evidence is explicitly quantified as conflict and the fused prediction becomes diffuse rather than being forced into an unjustified consensus. Under subject-independent and speaker-independent evaluation on four benchmarks, EMT-DS achieves 84.3% accuracy on IEMOCAP versus 81.7% for the best baseline (p < 0.01), 82.4% on DEAP under leave-one-subject-out evaluation versus 78.1%, and 96.8% on the highly imbalanced KEMDy20, and it improves accuracy on CMU-MOSEI by 2.4 points over the strongest evidential baseline. In controlled ablation, it reduces Expected Calibration Error from 0.153 (softmax transformer) to 0.042. It maintains 71.2% accuracy under severe Gaussian noise on audio features (sigma = 1.0), where attention baselines drop to 45.3%, and degrades gracefully when entire modalities are absent. The fusion step has O(MK^2) complexity, compared with O(M^2 d^2) for cross-modal attention. The learned conflict metric also flags likely errors: reviewing 20% of samples captures 80% of misclassifications, which supports selective prediction in clinical and safety-critical systems.

---

### Learning Features, Not Layouts: Feature-Axis Attention as a Transfer Prior for Medical Imaging

Authors: Aon Safdar, Mohamed Saadeldin
Published: 2026-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.13012v1

Natural-image-pretrained foundation models are increasingly adapted to medical imaging, yet transfer remains unreliable under domain shift, modality heterogeneity, and label scarcity. We trace part of this fragility to an overlooked design choice. Standard vision transformers place all global modeling on self-attention over spatial tokens, an object-layout prior well suited to natural scenes but not to medical images, where the discriminative signal lies in feature correlations and texture that recur across scanners while layout does not. We ask whether the axis of global attention, spatial versus feature, is itself a lever for transferable medical representations. Existing channel-aware hybrids confound this question, adding channel attention to a spatially-mixed backbone and leaving the axes coupled and the feature axis unmeasurable in isolation. We design FAD-ViT, a backbone that strictly decouples the two axes, making feature-axis attention the primary global operator and the attention axis a controlled variable. Evaluating comprehensively across 23 classification, segmentation, and low-data tasks under matched tokenization, capacity, and protocol, we find that shifting global attention from the spatial to the feature axis leaves from-scratch performance essentially unchanged, yet markedly strengthens transfer. Identical ImageNet pretraining yields up to roughly three times the transfer gain, with consistent improvements on every held-out segmentation task at up to a quarter the parameter count and far lower compute than standard and channel-attention baselines. The advantage is largest where texture rather than geometry is discriminative, tracking greater reuse of pretrained features. The axis of global attention is thus a design lever for medical transfer, and making it channel-primary is a simple, cheap, and principled choice.

---

### Provenance-Aware Belief Updating in Human-AI Feedback Loops

Authors: Tianyi Huang, Eric Liu, Jay Chong, Emily Yin, Wanqi Zhao
Published: 2026-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.13008v1

In repeated human-AI interaction, a single observation can be quoted, summarized, and returned to a model as multiple reports, making repetition look like independent corroboration. We study whether language models can use supplied report lineage to update according to evidence provenance rather than report count. We introduce Latent Quotient Evidence Training (LQET), a relational objective that combines sparse Bayesian supervision with constraints that encourage stable predictions when reports add no new evidence and consistent updates to the same new evidence across relay histories. On a controlled binary benchmark, LQET reduces new-evidence update RMSE by 41-44% and log-odds drift after eight relay steps by 89-91% relative to task-only training across Qwen3-4B and Phi-4-mini, with three paired training seeds per backbone. The gains extend to a closed-loop simulation in which model confidence controls how many relayed reports re-enter the model's context in later rounds. By grounding belief updates in supplied provenance, LQET offers a route toward dynamic human-AI systems in which influence follows new evidence, not the number of times the same evidence returns.

---

### Scaling Empowered Agents

Authors: Catherine Ji, Vivek Myers, Joshua Lin, Alexandra Wingate, Sergey Levine, Benjamin Eysenbach
Published: 2026-10-09
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.13002v1

Empowerment measures the capacity of an agent to actively control its environment. In this work, we explain the connection between skill learning and empowerment. While skill learning is not empowerment, we introduce a simple scalable recipe for empowerment maximization: empowerment corresponds to the number of executable skills from a state, so maximizing empowerment corresponds to reaching states that are good starting points for many skills. Our recipe applies to arbitrary underlying skill learning methods. Whereas most prior work on empowerment looks at tabular settings, this recipe enables us to study empowerment at a scale where we can realize theoretically-predicted but empirically-elusive properties of empowerment (e.g., gathering resources, environment modification, survival). Experiments show that empowerment maximization unlocks interesting emergent properties, such as building a house in Crafter. Website and code at \href{https://empowerment-geometry.github.io/scaling-empowerment/}{this https url}.

---

### Hindsight Structured Advice Distillation: Steering Executor Agents with Advisors Trained on Their Failures and Task Characteristics

Authors: O. Ozan Koyluoglu
Published: 2026-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.13000v1

Language-model agents are used for specialized work such as financial research, healthcare and law. Such agents are often capable enough for a task but still miss its details. In finance, for example, these details include which filing reports a figure or how a company discloses a metric. Retraining the agent to learn these details is expensive, and often impossible when the agent is a closed or shared system. Writing the missing know-how into the agent's prompt makes the prompt long and identical for every input, although different inputs need different guidance. We propose to learn to generate this guidance instead. A small advisor model reads each input and writes structured advice, which is appended to the prompt of a frozen executor. The advisor is trained with hindsight, following work on distillation with privileged information. On a training pool with graded answers, we analyse the executor's failures, run it under several strategies to see which one worked on each input, and let a teacher that sees the grades and the traces write the advice that would have helped. The advisor then learns, by supervised fine-tuning, to write that advice from the input alone. Because the student is a separate small model rather than the executor, it is cheap to train and adds little latency. We call this approach hindsight structured advice distillation (HSAD). On FrontierFinance, a public benchmark of expert-written financial research queries, an 8B advisor raises the qualification rate of Samaya's proprietary research agent, which already leads the benchmark's leaderboard, from 52.9% to 56.6%. This is above the agent's own high-effort configuration, at 37% lower cost per query. We study the generalization of the proposed approach, and in particular show that HSAD also improves results on WebShop, BIRD text-to-SQL and ExpertLongBench.

---

### Evaluating LLM Creativity through Long-Tail Performance

Authors: Yichen Wang, Tenghao Huang, Vishakh Padmakumar, Mina Lee
Published: 2026-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.12998v1

Creativity benchmarks for large language models (LLMs) vary widely in task and metric design. This raises the question of whether they capture a common notion of creativity. We propose that this shared target can be formalized as task-specific performance in the long tail of a reference input-output distribution, which describes how common input-output pairs are relative to a chosen source of text. We ground this proposition in the novelty-quality product formulation and evaluate it empirically across 15 benchmarks and 26 open-weight models, using n-gram absence from the pretraining data as a novelty proxy. Our results show that creativity benchmarks lie further in the tail regions than standard benchmarks. Moreover, on both creativity and standard benchmarks, model rankings from higher-tail subsets of each benchmark align more closely with human creativity judgments in LM-Arena. Building on these findings, we introduce q@Tail($α$), a lower-cost creativity diagnostic that measures mean task quality on the tail fraction of an existing benchmark. Applying it to creativity-promoting methods reveals substantial room to improve LLM creativity through stronger tail performance.

---

### Benchmarking Embedding Geometries and Clustering Algorithms for Unsupervised Patent Claim Classification

Authors: Sujith Subramanian, P N Bhargav Teja, Susen Sarkar, Rajasekhara Reddy Duvvuru Muni
Published: 2026-10-09
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.12997v1

Prior-art retrieval systems are structurally organized around classification codes, making cross-code discovery difficult. This paper asks if an unsupervised representation trained on patent text can recover relevant connections that code-restricted search cannot, and if hyperbolic space helps. We trained a hyperbolic and a matched Euclidean Variational Autoencoder (VAE) on 45,000 US patents (96 runs, 6 seeds). The hyperbolic model retrieves genuinely cross-category prior art 52% more often than the Euclidean model (18.76% vs. 12.34% of top-10 results, 3,000 queries). This represents a practical gain for the motivating search problem. The hyperbolic model agrees less with official classification (NMI 0.2392 vs. 0.3206), which we explain geometrically: the two measures capture different goals, and only one aligns with prior-art search needs. In a study of 10 technology domains (50,000 more patents), we find that a domain's classification breadth--the number of distinct CPC subgroups spanned--predicts approach performance (Pearson r=-0.642, p=0.045, N=10), and that patents per subgroup is an even stronger predictor (r=0.715, p=0.020). This provides a checkable signal for practitioners to determine where the method is most valuable. We report all findings with statistical transparency and release a complete, 91-test automated implementation.

---

### What Should SFT Learn? Conditional Information Gain for Generalizable Post-Training

Authors: Yuxuan Gu, Kequan Niu, Xiaocheng Feng, Yun Li, Yiqiao Yang, Bing Qin
Published: 2026-10-09
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.12995v1

Supervised fine-tuning (SFT) remains a widely used approach for post-training large language models, because fixed demonstrations can directly introduce desired reasoning strategies and behavioral patterns. Yet forcing a model to imitate externally provided trajectories may also interfere with behaviors already supported by pretraining, limiting generalization beyond the training distribution. Recent dynamic SFT methods mitigate this issue by reweighting token-level supervision according to model-side signals such as predictive probability, prior support, or distributional compatibility. However, model compatibility is not equivalent to learning value: a token that is weakly supported by the current model may either be an undesirable mismatch or precisely the task-specific decision that the demonstration is intended to teach. We propose CIG-SFT, an information-theoretic framework that estimates the task-conditioned information carried by each demonstrated token. Under teacher forcing, we compare the token likelihood from a frozen reference model with and without access to the task input while keeping the gold solution prefix fixed. The resulting conditional information gain quantifies how much the task itself reduces uncertainty about the demonstrated decision. We then convert this information into a KL-regularized projection of the current policy and derive an equivalent scalar-weighted SFT objective whose logit gradient exactly matches that of the projected target. Across mathematical reasoning and code-generation benchmarks, CIG-SFT consistently outperforms standard SFT and recent adaptive fine-tuning baselines, with particularly strong improvements in generalization to challenging evaluation distributions, while requiring neither additional training data nor online rollouts.

---

### Gaussian-Width Bounds for Offline-to-Online Linear Bandits

Authors: Kushagra Chandak, Volodymyr Tkachuk, Xiaoqi Tan, Csaba Szepesvári
Published: 2026-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.12990v1

We study how offline data can help online learning in stochastic linear bandits. We focus on partially covered offline data, where only $k$ of the $d$ feature directions are poorly covered. Existing analyses of algorithms based on phased elimination and optimism give regret bounds of order $\widetilde O(\sqrt{dkT})$ in this regime. We show that this dependence arises from loose control of the reward prediction error, and obtain tighter bounds using concentration of Lipschitz functions of Gaussian vectors. Based on this analysis, we develop a phased elimination algorithm with regret $\widetilde O(k\sqrt T)$. For the unit Euclidean ball, we prove a matching regret lower bound under partial coverage, as well as matching upper and lower bounds under uniform offline coverage. We also study pure exploration with offline data, providing algorithms and matching upper and lower bounds on simple regret for fixed-design algorithms and also for adaptive algorithms under partial and uniform coverage of offline data.

---

### Language Models Have Interpretable Narrative Mechanisms

Authors: David Yiyang Liu, Aditya Joshi, Paul Dawson, Andrew Piper
Published: 2026-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.12989v1

We investigate whether language models contain internal components that contribute to representing narrativity, the degree to which a text exhibits narrative properties. Across GPT-2-Small, Gemma-2-2B, and Llama-3.1-8B, we identify attention heads, MLP neurons, and sparse autoencoder features whose activations correlate with narrativity. Ablating selected neuron and feature sets increases next-token prediction loss more for high-narrativity than for low-narrativity texts, and reduces the narrativity of open-ended generations. On downstream benchmarks, the largest effects occur in story completion and narrative question answering, although they vary across models and languages; results on confabulation and hallucination benchmarks are mixed. These findings provide causal evidence that selected sets of narrative-associated components contribute to model behaviour, while their internal organisation, relation to narrative theory, and broader functions remain open questions.

---

### Forecasting Corpus-Sampling Disagreement in Masked Text Reconstruction

Authors: Hare Krishna
Published: 2026-10-09
Categories: cs.CL, cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2610.12987v1

If a second text corpus were collected from the same population, how much would a predictor's outputs change? We study this question for masked-text reconstruction using a smoothed predictor built from nearby token-pair counts and evaluated over the full 50,257-token GPT-2 vocabulary. Treating each source document as one observation, we derive a first-order formula for the expected disagreement between predictors fitted to two independent corpora. The formula accounts for variable document length, random context counts, and correlations across token offsets, and it can be evaluated from a separate reference corpus without access to either target corpus. On TinyStories, WikiText-103, and CNN/DailyMail, growing each corpus from 512 to 2,048 chunks of 64 tokens lowers disagreement by only 5--40\%, far short of the 75\% implied by $1/n$ scaling. The forecast tracks this slow decline, which we trace to smoothing of rarely seen contexts. Forecast accuracy depends less on how these correlations are modeled than on the size of the reference corpus: small references systematically underpredict disagreement. A source-split jackknife that uses only reference data reduces this calibration error by 57--59\%. After this correction, a simpler smoothed-multinomial approximation is most accurate when the reference corpus is smaller than the target corpus, and a nonlinear source bootstrap is most accurate when it is larger. Finally, small masked diffusion language models trained on the same corpora show that the count-based forecasts partly identify which inputs are most sensitive to the corpus, but miss the neural models' much faster decline in disagreement.

---

### Scan-Parallel Variational Inference for event-driven Latent Jump Models

Authors: Vansh Bansal, Min Chen, James G. Scott
Published: 2026-10-09
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2610.12986v1

Continuous-time latent state-space models are widely used for irregularly observed time series, typically representing uncertainty through a random initial state followed by smooth ordinary differential equation (ODE) dynamics, or Brownian perturbations accumulated by a stochastic differential equation (SDE). While these constructions are natural for smooth or diffusive systems, they do not directly encode a finite set of unobserved shocks whose timing, magnitude, and effect on memory are themselves uncertain. We instead introduce latent jump models (LJMs): the latent state follows closed-form deterministic dynamics between Poisson-driven event times and undergoes a learned instantaneous jump at each event, with state-dependent intensity and mark distributions. This defines a time-homogeneous Markov process whose jumps are intrinsic state transitions rather than numerical integration steps. For training, we derive a tractable event-augmented path-space evidence lower bound (ELBO) and introduce Scan-Parallel Variational Inference (SPVI): an observation-conditioned variational posterior samples the complete latent event configuration before the state recurrence is evaluated using triangular affine scans. This enables fast and scalable training for LJMs in logarithmic temporal depth. Across synthetic stochastic systems and real irregular time series, LJMs remain competitive with continuous-time latent baselines in forecast accuracy, often with fewer latent updates than numerical solver evaluations, while training one to two orders of magnitude faster than the sequential baselines.

---