# Research Papers — 2026-09 · bundle page 2 of 266 > Full-text markdown of papers #51–100 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/2 [← Previous bundle](https://aipod.dev/papers/archive/2026-09.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/3.md) ### Provably Tractable NFA-Constrained Language Generation via HMMs Authors: Jialiang Sun, Kuldeep Meel Published: 2026-09-30 Categories: cs.CL, cs.FL arXiv: https://arxiv.org/abs/2609.40185v1 Constrained generation aims to sample from language models (LMs) conditioned on hard constraints. Existing constrained-generation techniques for nondeterministic finite automaton (NFA) constraints either distort the distribution or sacrifice efficiency. Theoretically, this task reduces to counting the length-$n$ sequences accepted by an NFA (#NFA), and the exact #NFA problem is #P-complete. Recent work has shown that #NFA admits a fully polynomial randomized approximation scheme (FPRAS). Inspired by this result, we propose NFA-LM, a polynomial-time engine for NFA-constrained generation with theoretical guarantees under mild assumptions. Experiments show that NFA-LM efficiently generates high-quality outputs with theoretically bounded approximation error. --- ### Index-Translate: A Multilingual Translation Model Family -- Text, Speech, Controlled Dubbing, and Long-Document Translation Authors: Tianjiao Li, Mengran Yu, Chenyu Shi, Lusheng Zhang, Qisi Chen, Yanshan Zhou, Ji Qi, Jingying Liu, Yuang Feng, Ziang Cui, Tianxing Yan Published: 2026-09-30 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.40181v1 We introduce Index-Translate, a multilingual translation model family that combines a shared multilingual foundation with specialized training for general translation, instruction following, speech translation, controlled dubbing, and long-document translation. It includes three model sizes, 2B, 9B, and 35B-A3B, and supports translation in 150 languages, with multilingual instruction following. Evaluations on general translation and complex translation instructions show that Index-Translate outperforms translation models of comparable size and achieves performance comparable to 100B-scale translation models and frontier models. Index-Echo provides end-to-end speech-to-text and speech-to-speech translation, outperforming existing end-to-end models and achieving performance comparable to frontier omni models. Index-Homura extends the family to syllable-controlled dubbing. Index-NativeLong introduces native long-document translation with a dedicated task formulation and benchmark. These capabilities support diverse translation tasks, including multilingual content production. --- ### MANET-GNN: Learned Decentralized Optimization of Power Allocation in Multi-Channel MANETs Authors: Tomer Alter, Nir Shlezinger, Michael Segal Published: 2026-09-30 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.40170v1 MANETs enable flexible infrastructure-less wireless connectivity in dynamic and resource-constrained environments. As modern MANETs exploit multiple frequency channels and support heterogeneous traffic patterns, decentralized transmit-power allocation becomes increasingly challenging. We develop a unified learned optimization framework for decentralized power allocation in dynamic multi-hop, multi-channel MANETs. We formulate a constrained end-to-end throughput maximization problem covering unicast, multicast, multicommodity, convergecast, and many-to-many communication. Although centralized and non-convex, this problem serves as an unsupervised training objective for MANET-GNN, a message-passing GNN that operates as a distributed learned optimizer. MANET-GNN uses only local, possibly noisy, CSI and a prescribed number of neighbor message exchanges, enabling low-latency decentralized inference while generalizing across topologies and network sizes. Numerical results show that MANET-GNN achieves centralized-competitive performance across communication frameworks, remains robust to channel uncertainty, and scales effectively across MANET configurations. --- ### Learning from Research: Toward Lifelong Agent Harness Evolution Authors: Jingbo Yang, Kwei-Herng Lai, Xiaowen Wang, Yaar Harari, Evgeniy Gabrilovich, Shiyu Chang Published: 2026-09-30 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.40169v1 Language agents are expected to solve increasingly complex tasks, creating a growing need for continual improvement. One promising approach is to evolve the agent harness, the software that governs tool use, memory management, and task execution, while keeping the underlying language model fixed. Recent methods automate this process by using a meta coding agent to modify the harness based on execution feedback. However, relying on that agent's existing knowledge and observed failures can restrict exploration and make adaptation reactive. Inspired by how human experts learn from the research literature for new solutions, we introduce ScholarEvolve, a framework that automatically draws on state-of-the-art research to guide harness evolution. ScholarEvolve organizes the harness evolution directions into functional modules and uses topic modeling to identify distinct improvement strategies for each module. It implements these strategies and evaluates their combinations to improve task performance. Moreover, the framework is designed to incorporate new publications over time, allowing research advances to drive proactive lifelong evolution. Experiments demonstrate improvements on AppWorld and Tau2-Bench. ScholarEvolve raises Qwen3.5-27B task goal completion from 49.6% to 63.6% on AppWorld Challenge, and raises GPT-5.4-mini pass@1 from 72.7% to 81.9% on Tau2-Bench Telecom. --- ### PrefPI: Preference-Guided Steering into Out-of-Distribution Behaviors Authors: Seungeun Rho, Wontaek Kim, Danfei Xu, Sehoon Ha Published: 2026-09-30 Categories: cs.RO, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.40165v1 We present PrefPI (Preference-Guided Policy Iteration), an iterative framework for steering pretrained generative robot policies using only relative preferences over self-generated trajectories. Unlike prior preference-learning methods that primarily sharpen modes already represented by the policy, we study steering beyond the initial effective support, where desired behaviors are rarely or never observed under the initial policy. Our key idea is to formulate preference learning as preference-conditioned generative modeling: preferred trajectories define a conditional distribution, whose density ratio with the broader behavior prior provides an implicit preference signal amplified by classifier-free guidance (CFG). Repeating this preference-conditioned modeling and guidance step yields a form of preference-guided policy iteration, turning incremental improvements toward previously inaccessible behaviors. Across diffusion policies and the PI0.5 flow- matching VLA in simulation and the real world, PrefPI produces substantial behavioral shifts with limited feedback. In particular, PrefPI increases object transport height from 10.7 cm to 19.8 cm on real hardware with only 150 preference-labeled trajectories. --- ### Reinforcement Learning-Guided Graph Transformations for SpTRSV Optimization Authors: Buse Yılmaz Published: 2026-09-30 Categories: cs.DC, cs.LG arXiv: https://arxiv.org/abs/2609.40159v1 Sparse triangular solve (SpTRSV) is a fundamental kernel in numerous scientific and engineering applications. However, the data dependencies inherent in sparse triangular matrices significantly limit the available parallelism and make efficient workload distribution challenging. Recent graph transformation techniques address these limitations by modifying the dependency graph of the input matrix to improve parallel execution. Existing graph transformation strategies, however, rely on manually designed heuristics, making their development and adaptation to different optimization objectives challenging. This work proposes a reinforcement learning-guided graph transformation framework for SpTRSV, in which graph transformation is formulated as a sequential decision-making problem and an RL agent learns matrix-dependent transformation policies. Experimental results on real-world sparse matrices demonstrate level reductions of up to 94% and reductions of up to 80% in the coefficient of variation of level costs, while modifying only 1.50% of the rows in the highest case. On average, the RL- guided graph transformation achieves a 23% reduction in the number of levels and a 29% reduction in the coefficient of variation of level costs while rewriting only 0.82% of the matrix rows. Although the heuristic strategies generally achieve more aggressive level reduction(between 31% and 46%), the RL-based approach achieves the largest average reduction in the coefficient of variation of level costs, demonstrating its ability to balance competing graph transformation objectives. The results further show that the learned policies can be transferred to previously unseen matrices through curriculum learning and fine-tuning, while zero-shot experiments provide insights into the limitations of generalizing graph transformation policies across different sparsity patterns. --- ### Partial identification with entropy regularized optimal transport Authors: Bruno N. Costa, Florian F. Gunsilius Published: 2026-09-30 Categories: econ.EM, math.ST, stat.ML arXiv: https://arxiv.org/abs/2609.40156v1 In many statistical settings, the available data and maintained assumptions do not suffice to uniquely identify the model parameters of interest. In such cases, one can only identify sets which are guaranteed to contain the true parameters. These are often characterized through linear programs that optimize over models compatible with the observed data. These programs can be infinite-dimensional in the optimizer and the number of constraints. We provide a unified way to characterize and solve such optimization problems by phrasing them as optimal transport problems on path spaces. This allows us to regularize the problem with an entropy penalty, recasting it as a multi-marginal entropic optimal transport problem, which can be solved efficiently via Sinkhorn iterations. In addition, it allows us to establish convergence of the regularized value to the sharpest bound, derive consistency rates for a plug-in estimator, and obtain asymptotic distribution for approximate bounds. The method is general and accommodates settings ranging from instrumental variable models with continuous variables to welfare estimation in heterogeneous demand models. We verify the statistical and computational properties in simulations and provide an application to demand estimation. --- ### Role-Adaptive Policy Optimization for Offline Reinforcement Learning Authors: Seonvin Cho, Soohyun Choi, Songnam Hong Published: 2026-09-30 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.40149v1 Policy regularization in offline reinforcement learning balances policy improvement against reliance on uncertain value estimates. This balance can differ between selecting actions for execution and supplying actions for critic bootstrapping, yet methods such as TD3+BC couple these roles through a shared policy. We propose Role-Adaptive Policy Optimization (RAPO), which adapts policy-update coefficients according to their roles in value learning and execution. RAPO learns these coefficients by differentiating through candidate policy updates formed using the base algorithm's actor objective. For TD3+BC, RAPO separates bootstrap and execution actors and adapts their coefficients independently: the bootstrap objective penalizes policy-induced changes in target values, while the execution objective evaluates a local policy-improvement surrogate. For IQL, whose value learning is already independent of the execution actor, RAPO preserves the original value updates and adapts only the inverse temperature in advantage-weighted policy extraction. Experiments on D4RL locomotion and AntMaze tasks show improvements over both base algorithms, with larger gains for TD3+BC, whose RAPO instantiation outperforms baselines on average. --- ### From Spectra to Joint Schedules in LLM Pre-training: 3+3(+2) Scaling-Law Regimes Authors: Yichen Wang, Fanghui Liu, Yudong Chen Published: 2026-09-30 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.40148v1 Power-law learning curves are often treated as fixed properties of a model and its data, although learning-rate and batch-size schedules can change the observed loss. We study this dependence in noisy online SGD with linear random features. Conditional on the representation, an exact Volterra equation separates two response components: a forcing term that propagates unresolved target error and a memory kernel that propagates stochastic-error injections. We prove that either component follows a power law if and only if its cumulative weighted spectral mass has the corresponding low-spectrum scaling; individual eigenvalues and target coefficients need not obey coordinatewise power laws. Under a joint schedule, intrinsic time $T_t=\sum_{s<t}η_s$ controls optimization progress, while $r_t=B_t/η_t$ controls noise injection. Their interaction yields sharp conditions under which a schedule preserves, changes, or destroys the clean power law, together with a memory ceiling on noise reduction. The power-law random-feature model realizes this mechanism in $3+3(+2)$ propagation regimes with phase-dependent compute rates. Controlled nanoGPT experiments show that (1) learning-rate and batch-size schedules with matched $B/η$ paths are nearly equivalent in intrinsic time, (2) a forcing-memory surrogate accurately predicts loss across schedules, and (3) its fitted exponents across real-world datasets identify the regime of LLMs in $3+3(+2)$ map. --- ### Policy Iteration Is Not Strongly Polynomial for Deterministic Markov Decision Processes: The Price of Algorithmic Anarchy Authors: Han Zhong, Yinyu Ye Published: 2026-09-30 Categories: cs.LG, cs.DS, math.OC arXiv: https://arxiv.org/abs/2609.40147v1 We establish an exponential iteration lower bound in the number of states for Howard's policy iteration on deterministic discounted Markov decision processes, with at most two actions per state. This rules out strong polynomiality of Howard's policy iteration when the discount factor is part of the input and yields an exponential separation from the simplex method with Dantzig's pivoting rule, which is proved to be strongly polynomial on this class. Even when each reward is restricted to logarithmic bit length, we obtain a stretched-exponential iteration lower bound. The gap between Howard's decentralized and simultaneous selfish improvements and Dantzig's coordinated selection of a single action with the largest gain across all states reveals a ``price'' of algorithmic anarchy. --- ### From DNA Design to DNA Slimming: Auditable Agentic Discovery of a Deletion-Only Designer Authors: Joel Shor Published: 2026-09-30 Categories: cs.LG, cs.NE, q-bio.GN arXiv: https://arxiv.org/abs/2609.40143v1 Compact regulatory DNA can free up space in vector payloads, reduce synthesis and assay burden, and expose which sequence features drive predicted activity. Yet most model-based nucleic-acid designers optimize fixed-length sequences through substitutions; they do not ask which bases of an existing functional element can be removed while retaining predicted activity. We define the task of sequence slimming as selecting an exact-length, order-preserving subsequence while retaining activity. Modeled on the design benchmark NucleoBench, we propose a quantitative evaluation for slimming that balances sequence reduction with maintaining function. Each slimmer must return both the subsequence and its source indices, which can be used to verify that the slimmer obeyed task requirements. To our knowledge, this is the first dedicated benchmark of this deletion-only problem. The coding agent Empirical Research Assistant (ERA) then searched over executable designer programs. ERA received the task prompt and a successful substitution-only designer GrAdaBeam as a starting program, and it modified the designer to produce GRADASLIM. We report held-out evaluations for five transcription-factor binding targets, comparing random, greedy, and ERA-guided slimming at 400 and 100 bp. ERA has the highest mean in 9/10 settings. Paired bootstrap intervals for ERA minus greedy are above zero in all five 400-bp settings, below zero in one 100-bp setting, and overlap zero in the remaining four. --- ### Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat Authors: Ahmed Marey, Henry Lu, Abhishek Gaur, Liangzhu Leon Wang, Sherif Goubran, Malek Aloui, Theodore Potsis, Alex Hernandez-Garcia, David Rolnick Published: 2026-09-30 Categories: physics.ao-ph, cs.LG arXiv: https://arxiv.org/abs/2609.40140v1 Extreme heat is where urban adaptation needs kilometer-scale data the most, but the simulations training a downscaler can cost more than they save, and how much is needed has not been identified. We measured it with CASPER, a U-Net with a structure-preserving loss downscaling 32 km reanalysis to 1 km temperature, humidity and wind, across 24 configurations of one to eight months. Held-out error grows linearly with climatological distance to the training data, RMSE = 0.83 + 2.95 d, explaining 90% of its variance against 7% for volume and predicting unseen months in advance. On held-out extreme summer weeks CASPER preserves the fine-scale structure and cross-variable physics that matched-budget baselines degrade, and matches station observations during documented heat waves to within 1.8 K. Transfer to a new region degrades geographically; 11 days of local simulation cuts Vancouver's held-out error from 3.8 to 1.3 K. Training periods should span the target climate: the same accuracy for four times less simulation, putting kilometer-scale downscaling of extreme heat within reach of groups without large computing facilities. --- ### Game-Guided Skill Discovery through Self-Play for Playable Agent Control Authors: Seungeun Rho, Jeonghwan Kim, Xue Bin Peng, Sehoon Ha Published: 2026-09-30 Categories: cs.LG, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2609.40137v1 We present Game-Guided Skill Discovery (GGSD), a framework that uses self-play in games to discover motor skills that are directly playable by humans. Playable skills provide a compact abstraction for controlling embodied agents through a small set of learned behaviors rather than low-level actions. To be effective, these skills should be semantically distinct, interpretable, and expressive; properties that existing unsupervised skill-discovery methods often fail to achieve simultaneously. GGSD achieves these desiderata by grounding skill discovery in competitive gameplay. A hierarchical agent competes against its past selves, with a high-level policy selecting from a small discrete skill set and a skill-conditioned low-level policy learning the corresponding behaviors. After training, a human can replace the high-level policy and directly control the agent through the same discrete skills. Despite the small number of high-level actions, skill transitions give rise to emergent combo behaviors, expanding expressivity beyond individual primitives. Across Ant, Franka-arm, and Unitree G1 environments, we show that GGSD produces human-playable skills that humans can compose to solve unseen tasks, such as Maze and CubePush, without additional training. An interactive demo is available at https://ggsd-demo.github.io. --- ### Tactile Curiosity Drives Robot Interaction Authors: Klemens Iten, Alexander Proshkin, Bhavya Sukhija, Stelian Coros, Andreas Krause, Pieter Abbeel, Carmelo Sferrazza Published: 2026-09-30 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2609.40134v1 Mastering robot manipulation skills via reinforcement learning (RL) remains largely sample-inefficient. The most common RL algorithms rely on random action sampling to discover new strategies, resulting in agents that allocate most of their training budget to motions in free space, away from the contacts from which manipulation skills emerge. Existing intrinsic motivation methods based on model disagreement or epistemic uncertainty improve on isotropic noise, but they can also reward uncertainty in functionally irrelevant transitions, such as erratic motions in free space. In this work, we argue that tactile feedback provides a natural signal for exploration, and introduce TacEx, a framework that incorporates touch into epistemic uncertainty-driven exploration by decomposing model uncertainty across sensory modalities and directing curiosity toward the tactile channel. By anchoring curiosity to the sense of touch, TacEx drives the robot to discover complex contact dynamics, learning to manipulate and grasp objects without task rewards or expert demonstrations during exploration. The interaction-dense dataset collected through this tactile-driven curiosity supports offline learning of downstream pick-and-place policies without additional environment interaction. We further use tactile-driven exploration to post-train vision-language-action (VLA) models. Although the VLAs are initially pre-trained without tactile feedback, post-training with TacEx substantially improves downstream performance while remaining highly sample-efficient. --- ### Prototype-Rule Neurosymbolic Regularization for Rank-Constrained Tensor Neural Networks under Label Scarcity Authors: Eftychios Protopapadakis, Konstantinos Makantasis, Konstantinos M. Giannoutakis Published: 2026-09-30 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2609.40131v1 Rank-constrained tensor neural networks reduce the parameterization of high-order inputs, but they do not explicitly constrain class geometry in the learned representation. This study investigates whether a differentiable prototype-rule can provide a complementary inductive bias for Rank-R tensor learning under limited supervision. The proposed framework augments the Rank-R objective with prototype-based regularization and optionally fuses prototype evidence with neural logits at inference. Four hyperspectral benchmarks are evaluated with four Rank-R configurations under both seven-fold stratification and spatially separated folds that mitigate leakage; a separate spatial study varies the class support budget from 2 to 20 samples. Under spatial evaluation, full neurosymbolic inference changes Macro-F1 score by +8.82 percentage points on Botswana, +5.49 on Indian Pines, +1.59 on Pavia University, and -0.62 on Salinas. Most of the benefit arises from training-time regularization, whereas inference fusion is small and dataset dependent. --- ### VR-JEPA: Learning Contrastive-State Latent Guidance for Generation-based Video Reasoning Authors: Zehua Ma, Kun Xiang, Yunshuang Nie, Quanlin Chen, Haoyuan Li, Xiuwei Chen, Jiang Ji, Haijun Wu, Zhenyu Xie, Michael Kampffmeyer, Hanhui Li, Xiaodan Liang Published: 2026-09-30 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.40129v1 Reasoning through video generation offers a promising path toward visual intelligence by modeling latent visual states and their dynamics. However, current video generation models often lack explicit guidance on how these states should evolve, leaving generated trajectories prone to physical and structural inconsistencies that undermine reasoning reliability. While the Video Joint-Embedding Predictive Architecture (V-JEPA) provides rich spatiotemporal priors learned through latent prediction, these general priors do not naturally adapt to the logical reasoning capabilities required for complex visual tasks. To bridge this gap, we propose VR-JEPA, a framework that aligns the V-JEPA predictor with task-specific reasoning logic through localized contrastive-state learning and uses its predicted latent trajectories to guide video generation for visual reasoning. Specifically, (i) we pair successful trajectories with generated alternatives under the same input conditions and use discrepancies in their V-JEPA representations to identify informative states and tokens for localized contrastive supervision. (ii) We further equip the V-JEPA predictor with skill-specific experts trained on anchor-task data, allowing the model to adaptively specialize its shared spatiotemporal priors across diverse cognitive domains. Together with skill-specific experts, this contrastive supervision enables VR-JEPA to predict latent trajectories that provide task-specific logical guidance for video generation. Comprehensive experiments on the large-scale VBVR-Pro-Bench dataset demonstrate that VR-JEPA achieves an $11.33\%$ relative improvement over the cutting-edge generation-based reasoning baseline, significantly mitigating physical artifacts and enhancing logical consistency. --- ### Learning Functional Subspaces for Neural Network Compression Authors: Massimo Bini, Anders Christensen, Stephan Alaniz, Judah Goldfeder, Ole Winther, Yann LeCun, Ravid Shwartz-Ziv, Zeynep Akata Published: 2026-09-30 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2609.40127v1 Modern transformers pair impressive capabilities with substantial memory and compute demands. Low-rank weight factorization reduces both while keeping the matrices dense, and thus efficient on standard hardware. Existing methods, however, choose the subspace to remove from each weight matrix with local closed-form criteria: activation energy, layer-wise reconstruction error, or a quadratic approximation of the loss. These criteria ignore how errors propagate through the network, so at high compression the errors compound with depth and performance collapses. We introduce Learnable Subspace Projections (LSP), which instead learns the subspaces to discard end-to-end. Each linear layer, or tied group of layers that read the same activations, is assigned an orthogonal projector. All projectors are optimized jointly against a global objective--the KL divergence to the dense model's output distribution or the model's original training loss--while the pretrained weights remain frozen. Projectors are initialized from a whitened SVD truncation, and ranks are allocated by the output KL each projector induces per parameter saved. After training, the projectors merge into standard low-rank factors, with each tied group sharing one factor. In attention, this also lets the model cache one narrow latent in place of full keys and values. Across LLMs (OPT-125M/1.3B, Qwen3-4B, Llama-2-7B) and ViT-B/16, LSP outperforms baselines, and its advantage widens as compression increases. At -70% compression, LSP brings Llama-2-7B to 10.9 WikiText-2 perplexity and 42.2% mean zero-shot accuracy, versus 13.3 and 36.0% for the strongest baseline. The factorized model decodes up to 1.6x faster than the dense model at small batch sizes, and aching the shared latent shrinks the combined memory of weights and KV cache by 13.5x at a 128k-token context, versus at most 6.5x for untied baseline factorizations. --- ### Debias It Yourself: Teaching LLMs Cognitive Bias Mitigation Interventions Authors: Chahat Raj, Sina Mansouri, Aylin Caliskan, Antonios Anastasopoulos, Ziwei Zhu Published: 2026-09-30 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.40124v1 Bias has long been studied in social psychology and cognitive science, where decades of research have produced a body of validated interventions that reduce stereotypical thinking and prejudiced responses in humans. We propose Debias It Yourself (DIY), a cognitively grounded framework that translates five such interventions into debiasing procedures for large language models and delivers them through three established paradigms: Show (in-context examples), Train (instruction tuning), and Revise (guided self-revision). Across three models, five bias benchmarks, eleven debiasing baselines, and three reasoning benchmarks, Train+Revise and Revise alone attain the top two average ranks, lead the bias-reasoning tradeoff (mean bias as low as 2% at 90% reasoning accuracy), and reduce bias on unseen dimensions by up to 14.8%. Our code and data are publicly available. --- ### On the (In)effectiveness of AMR Augmentation for Large Language Models Authors: Hoa Quynh Nhung Nguyen, Jacopo Staiano, Michael Sullivan Published: 2026-09-30 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.40121v1 While Abstract Meaning Representation (AMR) has historically improved performance on a range of NLP tasks, the benefit---or lack thereof---of AMR augmentation for modern LLMs is thus far unclear. In this paper, we attempt to reproduce recent work that reported substantial downstream gains from AMR augmentation, finding that these are likely due to specific choices in the experimental settings used: using a consistent and unified protocol for hyperparameter selection, we observe that text-only baselines consistently match or exceed the performance of AMR-augmented models. To investigate this null result, we introduce a perplexity-based probe measuring the degree to which AMR provides an LLM with supplemental relational knowledge not already available to the model. We find that AMR augmentation does not help LLMs improve their understanding of relational content in the sentence, indicating that augmenting these models with AMR offers no clear benefit on downstream tasks. --- ### Scalable Cox Regression via Grouped Risk Sets and Sharper LogSumExp Rates Authors: Elizaveta Iashchinskaia, Egor Gladin Published: 2026-09-30 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2609.40120v1 Motivated by the computational challenges of large-scale Cox regression, we study stochastic minimization of LogSumExp objectives over large sets. Mini-batch normalizer estimates generally yield biased gradients. We instead use a softplus surrogate that introduces one auxiliary scalar per normalizer and admits unbiased single-sample gradients. For smooth convex LogSumExp objectives, we prove an $O(T^{-1/2})$ averaged objective bound, improving the previous $T^{-1/4}$ analysis. With a strongly convex regularizer on the original variable, we also obtain a last-iterate squared-error rate of $\widetilde{O}(T^{-1})$ without strong convexity in the auxiliary variables. For Cox regression, the normalizers are defined over nested risk sets. We exploit this structure by grouping neighboring failures and sharing one auxiliary variable per group. The resulting compressed objective admits uniform score and curvature bounds that control the errors from grouping and softplus approximation. Together with the general optimization result, these bounds give a mean-square rate of $T^{-4/5}$, up to logarithmic factors, relative to the full Cox solution. The compressed estimator also matches the full estimator's asymptotic distribution. Experiments on synthetic and real survival datasets with slowly decreasing risk sets show a favorable performance relative to stochastic baselines. --- ### Persistent Context Graphs for Efficient Memory Compaction in LLM Agents Authors: Jingbo Yang, Kwei-Herng Lai, Xiaowen Wang, Zhaoxuan Tan, Pei Zhou, Mengting Wan, Yaar Harari, Evgeniy Gabrilovich, Shiyu Chang Published: 2026-09-30 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.40118v1 As LLM capabilities advance, agents are tackling increasingly complex tasks over longer horizons. Their growing interaction histories make memory compaction essential for staying within context windows and reducing prefill cost. Existing methods summarize the history or compress its KV cache, often adding model computation to preserve information for future requests. A new user request can change which history matters, but reassessing that history with the model requires re-encoding it if the KV cache has expired. Past attention provides signals of historical importance and dependencies between messages, while relevance to the current task must be assessed using the new user request. We introduce ReCAP, a memory compaction method that stores attention-derived importance scores and dependency links in a lightweight, persistent context graph. For each new request, ReCAP combines stored importance with relevance cues from the request and follows dependency links to select messages and their supporting context, without additional model calls for selection. Compared with Codex's default summarization-based compaction, ReCAP reduces estimated latency for compaction and cold restoration by approximately 95% on both Qwen3-Coder and gpt-oss. It also roughly halves the historical context per call on SWE-Together at comparable task quality and improves accuracy on the code tasks of Lost-in-Conversation over full history by 19.8 and 41.2 points. --- ### Beyond Model Ranking: Regime Diagnosis for Distributional-Statistical Misspecification in Industrial Time-Series Forecasting Authors: Pengyu Nie, Chenglang Xu, Yaoshi Chen, Chaogan Ren, Wei Hu, Chao Yang, Jiangong Zhang Published: 2026-09-30 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.40117v1 Time-series forecasting models achieve strong benchmark performance but exhibit severe systematic bias in industrial deployments. This train--deploy gap is conventionally attributed to temporal-structural errors or distribution shifts. We characterize a complementary source that these explanations overlook: canonical losses embed fixed statistical priors, while industrial demand mixes benign and pathological regimes---zero-inflation, skewness, high variability---in which these priors are systematically violated. The induced bias persists even under perfect temporal modeling, remains in a distributional-shape component that normalization cannot remove, and creates an aggregation trade-off invisible to aggregate metrics. We turn these observations into an evaluation toolkit centered on the Regime-wise Relative Bias Vector (RBV): a metric-agnostic, regime-decomposed diagnostic that audits how pooled training allocates systematic mismatch across pathological subpopulations. A controlled attribution analysis decomposes RBV into a model-independent intrinsic floor, set by each loss's estimand, and an excess component attributable to training, tracing observed bias to the loss rather than the model. A large-scale study---13 loss objectives, 3 seeds, 60,000+ series spanning RetailShiftBench and M5, with random-split controls---shows that regime-aware diagnosis separates optimization-type from bias-type failure, and that regime-aware training resolves the pooling-induced bias that capacity scaling cannot, for mean-type losses. A formal structural observation, that risk under evaluation-distribution contamination is affine in the pathology mixture weight, grounds these findings. Our work complements model ranking with mechanism-grounded, regime-oriented evaluation. --- ### Unlearnable, or Unmeasured? On the Reliability of Difficulty Labels in RLVR Authors: Chandak Chakma, Syed Nazmus Sakib, Nafiul Haque, Shifat E. Arman Published: 2026-09-30 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.40115v1 Reinforcement learning with verifiable rewards (RLVR) has become an important approach for improving reasoning during post-training. Recent work suggests that some difficult prompts remain resistant to learning even when they occasionally produce correct solutions. We revisit this unlearnability phenomenon and find that the affected prompts do improve, at roughly one third of the learnable rate, while the difficulty-defined set used to study them is much less reproducible than expected. These difficulty labels are estimated from a limited number of sampled responses. Combining them across seeds can further change which prompts are selected instead of simply reducing measurement noise. We develop a sampling-based framework for quantifying this instability and determining how much evaluation is required for difficulty assignments to reproduce reliably. We also revisit the gradient-similarity evidence proposed to explain unlearnability and show that part of the observed separation arises because difficult prompts provide fewer correct rollouts from which their gradients can be estimated. Matching this sample count weakens the gradient difference but does not remove it. Overall, the slow-learning phenomenon survives our reanalysis, while both the prompts used to define it and the evidence used to explain it require more careful measurement. --- ### Agent Error Dataset: Scaling 50,000 Error--Diagnosis Pairs for Failure Analysis and Error-Aware Post-Training Authors: Kunlun Zhu, Xuyan Ye, Yibo Li, Cheng Qian, Beibin Li, Heng Ji Published: 2026-09-30 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.40111v1 An unsuccessful LLM agent rollout contains more information than its final reward: the observations available to the agent, the actions it chose, and the environment's responses. Reusing this experience for learning requires identifying a decision to revise and testing a concrete alternative. We introduce the Agent Error Dataset (AED), comprising 50,228 error-diagnosis pairs from 9,961 source tasks across 33 environments, 19 harness families, and 23 policy models in text-based agent systems. We retain source traces and execution metadata to support cross-setting failure analysis and re-diagnosis without repeating the original rollout. Our five-stage Agentic Error-to-Training (AET) pipeline collects natural failures, generates diagnoses and proposed corrections, and checks them against recorded evidence. Where replay is supported, we compare corrections with original-action retries from the same checkpoint under matched execution settings. We then construct separate training views for diagnosis and actor recovery. Across 3,062 matched replay pairs, first-proposal corrections raise verifier pass rates from 18.4% to 51.1%, a gain of 32.7 percentage points. Using a separately frozen diagnosis release, full-diagnosis fine-tuning on 1,656 source tasks raises Qwen3-8B's exact-step agreement with internal teacher labels from 47.2% to 63.6%, averaged over three seeds on a 943-case holdout. The strongest prompted reference in this comparison scores 54.7%, and mean agreement improves at each of four increasing training-set sizes. In a single-seed comparison of actor-training recipes, action-only repair training scores 6.67 percentage points higher on WebShop-lite than success-only training. --- ### OverdoseMoE: A Multi-Expert Framework for Opioid Overdose Risk Prediction Authors: Mingchen Li, Rohan Pandey, Junhui Qian, Feiyun Ouyang, Sunjae Kwon, Avijit Mitra, Zonghai Yao, Hong Yu Published: 2026-09-30 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.40108v1 Opioid overdose remains a major clinical and public health burden, highlighting the need for scalable approaches to identify patients at high risk. Here, we investigate diagnosis-specific adaptation for 180-day opioid overdose risk prediction from patients' preceding one-year longitudinal ICD histories. We develop OODMAMBA and OODQWEN through continued pretraining on longitudinal diagnostic sequences followed by task-specific fine-tuning. Building on the stronger Qwen-based predictors, we further propose OVERDOSEMOE, a multi-expert framework that integrates models of different scales using complementary expert-weighting strategies. Diagnosis-specific adaptation consistently improved predictive performance over general-purpose language-model baselines, with OODQWEN achieving an AUPRC of 24.47 and an AUROC of 68.56. OVERDOSEMOE further improved discrimination and precision, achieving an AUPRC of 25.17 and an AUROC of 69.49 while outperforming the strongest single-model baselines. Among patients ranked in the top 5% of predicted risk, OVERDOSEMOE identified substantially enriched overdose risk, achieving a PPV of 25.38% while retaining meaningful recall. Evaluation on an independent MIMIC-IV cohort further demonstrated cross-cohort robustness, with complementary weighting strategies showing advantages across different performance measures. These findings demonstrate that diagnosis-specific language-model adaptation combined with multi-expert integration can improve opioid overdose risk stratification and support more robust prediction across heterogeneous electronic health record populations. --- ### JuryFlow: Disagreement-Guided Human-in-the-Loop Multi-Agent Evaluation Authors: Mufeng Yang, Junwei Yu, Yepeng Ding Published: 2026-09-30 Categories: cs.CL, cs.HC arXiv: https://arxiv.org/abs/2609.40103v1 Large language models (LLMs) are increasingly deployed as automated judges for AI-generated content, yet a single judge is unreliable and even a panel of judges leaves a hard residue: when judges disagree, majority voting discards the conflict instead of resolving it. We present JuryFlow, a disagreement-guided, human-in-the-loop multi-agent evaluation framework that treats inter-judge disagreement not as noise to be averaged away, but as a precise, claim-level signal indicating where an evaluation is uncertain. JuryFlow decomposes each candidate response into atomic claims, has a panel of heterogeneous judges assign per-claim verdicts, and builds a disagreement graph whose nodes are scored by verdict entropy and whose edges encode structural similarity between claims. A human acts as a structural guide, selecting which disagreement to resolve through a single, minimal intervention rather than re-labeling the response, after which the focal claim is re-evaluated, the correction propagates along graph edges and to historically similar cases, and is crystallized into reusable rubric entries that all judges inherit, making the evaluator progressively self-refining. To enable large-scale, reproducible benchmarking without human studies, we evaluate JuryFlow in an automatic configuration in which focal selection is made by entropy ranking. On MT-Bench and LLMBar, JuryFlow improves agreement with gold labels over single-judge and majority-vote panel baselines, and ablations isolate the contributions of disagreement-targeted re-evaluation, propagation, and rubric induction. We contribute (1) a human-in-the-loop paradigm that recasts the human from labeler to structural guide, (2) the JuryFlow framework operationalizing it through a disagreement graph, focal re-evaluation, and closed-loop rubric induction, and (3) an evaluation protocol with ablations that isolate where the gains originate. --- ### AutoDataBench: A Data-centric Testbed for Accelerating Auto Research Authors: Ruifeng Yuan, Yizhi Li, Yaxin Du, Fengyu Cai, Yiqi Liu, Hou Pong Chan, Chenghua Lin, Yun Chen, Jian Yang, Bryan Dai, Pinyan Lu, Chenghao Xiao Published: 2026-09-30 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.40097v1 Existing auto-research benchmarks often entangle multiple sources of improvement, including training frameworks, hyperparameters, compute budgets, and data, making it difficult to attribute why one frontier agent outperforms another to specific research capabilities. In this work, we isolate and systematically evaluate Data Intelligence: an agent's ability to understand, manipulate, and improve the data that shapes model capabilities. We introduce AutoDataBench, a controlled testbed built on a conceptual framework of data intelligence spanning data diagnosis, data organization, and data construction, instantiated through three highly curated optimization tasks while holding non-data factors fixed. Across tool use, retrieval, and knowledge injection, we evaluate frontier LLMs' ability to improve training data through iterative experimentation under task-specific resource budgets. Beyond optimization performance, we ask: do LLMs understand what their data interventions do? We compare predictions made before training with observed outcomes to seek evidence of data-effect reasoning beyond trial and error, and explore whether iterative feedback helps LLMs better understand how changes to training data affect model performance. Finally, we show that reusing AutoDataBench trajectories for mid-training improves downstream coding performance, highlighting its value in both evaluating data intelligence and generating high-quality training data. Code and resources are available at https://github.com/AutoDataBench/AutoDataBench. --- ### Efficient Expert-Parallel Communication on PCIe-Connected Consumer GPUs Authors: Jaehwan Lee, Sangmin Lee, Chaewon Kim, Junsik Shin, Jaejin Lee Published: 2026-09-30 Categories: cs.DC, cs.LG arXiv: https://arxiv.org/abs/2609.40093v1 Expert parallelism (EP) enables inference of large Mixture-of-Experts (MoE) models by placing their experts across multiple GPUs, but requires substantial communication between GPUs at every MoE layer. As contemporary MoE models activate more experts per token, this communication accounts for a growing fraction of inference time. The cost becomes particularly pronounced on PCIe-based consumer GPU systems, where all inter-GPU transfers traverse CPU memory. However, existing MoE-specialized EP communication libraries assume that direct GPU-to-GPU access is available, largely overlooking consumer GPUs. Therefore, most LLM frameworks instead rely on NCCL, whose CPU-staged communication incurs redundant PCIe transfers and competes with expert computation for GPU resources, limiting their overlap. We present ThunderEP, a novel communication design for such systems that removes the relay hops of traditional ring algorithm, moves data through DMA engines to avoid compute resource contention, and minimizes synchronization latency by reducing the polling overhead of completion flags in CPU memory. We integrate the proposed design into vLLM and evaluate it on three widely used MoE models. Experiments on two PCIe systems equipped with RTX 4090 and RTX 5090 GPUs show that ThunderEP achieves average speedups of 2.00$\times$ and 1.53$\times$ over NCCL for dispatch and combine, respectively, and up to 1.66$\times$ end-to-end speedup over state-of-the-art MoE inference frameworks. --- ### GateSPINE: Gated Cross-View Fusion for Lumbar Spine MRI Report Generation Authors: Hoang Nguyen Van, Cuong Vuong Tuan, Trang Mai Xuan, Bien Tran Van, Nam Tran Van, Thien Van Luong Published: 2026-09-30 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.40091v1 Automated report generation can ease the burden radiolo gists face when interpreting multi-sequence MRI studies. Unlike CT, MRI examinations comprise multiple sequences and imaging planes, each con tributing complementary diagnostic information. Existing methods en code a study as a single volume and combine multiple acquisitions by fixed rules. Findings visible in only one plane are thus diluted and of ten missed, lowering recall on clinical efficacy metrics, where a missed abnormality is most costly. We propose GateSPINE, a vision-language framework that fuses sagittal T1 and T2 volumes with a training-free operator, encodes the fused sagittal and axial volumes with two parallel 3D encoders, and decodes their combined representation into a report. Its core mechanism is a gated cross view fusion module that predicts, per feature channel and token, how much of each view to admit, so the more informative view dominates at each spatial location. We evaluate GateSPINE on three lumbar MRI datasets, comprising two public bench marks and a private cohort collected from Phenikaa University Hospital, using both natural language generation (NLG) and clinical efficacy (CE) metrics. GateSPINE achieves the highest CE F1 through improved re call on all three datasets; on SPIDER, which lacks an axial sequence, this reflects the sagittal fusion component rather than the gated cross-view mechanism, which is validated on the two cohorts with both imaging planes. GateSPINE also remains competitive on standard NLG metrics. --- ### PTNO: Training Neural Operators with Noisy Monte Carlo Estimates for Particle Transport Problems Authors: Yubo Cao, Xi Deng, Mengqi Xia, Vignesh Gopakumar, Ander Gray, Anima Anandkumar Published: 2026-09-30 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.40090v1 Particle transport under multiple scattering is central to radiative transfer and plasma physics, yet high-fidelity Monte Carlo (MC) simulations must trace prohibitively many particles. Learning-based surrogates can amortize this cost, but typically train on expensive, well-converged MC solutions. We propose the Particle Transport Neural Operator (PTNO), a neural operator that learns particle transport surrogates directly from noisy, low-cost MC labels. Such labels pose two challenges: (1) high variance, which destabilizes standard supervised learning, and (2) a high dynamic range (HDR) spanning many orders of magnitude. For the first, we learn the solution operator from noisy labels of many configurations, amortizing MC cost and generalizing to unseen configurations. Because MC labels are unbiased, we show that the squared loss on them shares its minimizer with the loss on converged solutions, and our budget-allocation study over training scenes $M$, MC samples per render $N$, and independent renders per scene $K$ shows that many noisy scenes beat fewer converged ones. For the second, a nonlinear transform such as the logarithm biases noisy supervision. Instead, PTNO keeps labels in physical space and enforces positivity with a softplus output layer that represents small values effectively. We further train with a pointwise relative $L_2$ loss (PRelL2), the stop-gradient relative loss of HDR denoising and neural rendering, which normalizes each residual by the stop-gradient prediction instead of the noisy label. We demonstrate PTNO on neutron transport in fusion reactors and radiative transfer in participating media. On the two neutronics tasks, PTNO is $10^4$-$10^5\times$ faster than converged MC on the same CPU and $10^3$-$10^5\times$ cheaper than MC at matched accuracy; on the two radiative-transfer tasks, MC at matched accuracy costs $0.8$-$11\times$ as much as PTNO. --- ### Replay on Demand: An Emergent Curriculum for Balancing Adaptation and Forgetting in Continued Pretraining Authors: Lukas Thede, Shengzhuang Chen, Stefan Winzeck, Matthias Bethge, Zeynep Akata, Jonathan Richard Schwarz Published: 2026-09-30 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.40089v1 Continued pretraining enables language models to adapt to new domains and knowledge, but often at the cost of forgetting previously acquired capabilities. Replay can mitigate this trade-off, but fixed replay mixtures allocate training independently of the model's actual retention needs. We introduce Replay on Demand (RoD), which instead derives the replay allocation from the model's learning dynamics. RoD jointly prioritizes adaptation samples by their remaining learning potential and replay samples by their observed forgetting. Their competition for a shared training budget yields an online curriculum that determines what to train on at each step. Across models, scales, and adaptation domains, RoD reaches or improves upon the adaptation-forgetting frontier of tuned fixed-replay baselines and model merging without prescribing a replay allocation in advance. Replay concentrates on sources that are more vulnerable to forgetting and dynamically increases and redistributes as forgetting emerges during training. Together, our results show that replay can be allocated online from the model's evolving state, targeting what is needed, when it is needed. --- ### MeanVoiceFlow2: Joint Optimization of Mean Flow and Content Encoder for Fast One-Step Zero-Shot Voice Conversion Authors: Takuhiro Kaneko, Hirokazu Kameoka, Kou Tanaka, Yuto Kondo Published: 2026-09-30 Categories: cs.SD, cs.AI, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2609.40087v1 Flow-matching approaches to voice conversion (VC) have gained attention owing to their high speech quality and strong speaker similarity. Among them, one-step models such as MeanVoiceFlow are particularly attractive because they enable efficient inference; however, their reliance on a computationally intensive content encoder remains a bottleneck. We therefore propose MeanVoiceFlow2, a framework that jointly optimizes a flow-based conversion module and a computationally efficient content encoder. The model is trained through conversion distillation using MeanVoiceFlow and the reconstruction of real data. We further incorporate diffusion-GAN training with sample mixing and teacher-guided conditioning augmentation to enhance realism and disentanglement. Experiments on zero-shot VC showed that MeanVoiceFlow2 achieved higher perceptual quality and approximately $9\times$ faster inference than MeanVoiceFlow while maintaining comparable speaker similarity. Audio samples are available at https://www.kecl.ntt.co.jp/people/kaneko.takuhiro/projects/meanvoiceflow2/. --- ### BatSLAM 2.0: Sequence-Verified Sonar Place Recognition in a Robust Pose Graph Authors: Jan Steckel Published: 2026-09-30 Categories: cs.RO, cs.AI, cs.LG, eess.SY arXiv: https://arxiv.org/abs/2609.40085v1 Echolocating bats can navigate dark and cluttered spaces using echolocation. Over a decade ago, BatSLAM showed that a robot with a biomimetic binaural sonar can build a topological map of the environment, by recognizing places from the received acoustic signals. Sonar place recognition, however, is ambiguous by nature: corridors produce nearly identical echo trains, and wrong loop closure can collapse the topological map. In this paper, we introduce BatSLAM 2.0, a novel sonar-only SLAM system built from three elements: an updated acoustic front-end, a sequence verifier that tracks and verifies loop closure candidates and a pose graph implemented on a high performance factor graph framework. The system was thoroughly evaluated both in simulated as well as real world recordings. In both cases, the BatSLAM2.0 algorithm shows the capability of robust topological map creation, countering map collapse, and robust scaling of map size. --- ### Tissue Detection Determines False Positives in Diffusion-Based Histopathology Artifact Detection Authors: Konstantinos Moutselos, Ilias Maglogiannis Published: 2026-09-30 Categories: eess.IV, cs.CV arXiv: https://arxiv.org/abs/2609.40083v1 One-class artifact detectors for whole-slide images learn normal tissue from a clean training pool and flag departures from it. The pool is built by a preprocessing pipeline whose tissue-detection step is usually treated as neutral. We tested whether it is. On 16 annotated TCGA slides, we rebuilt the clean pool of a diffusion-based detector with different tissue detection methods and compared the resulting models in a four-fold cross-validation. Per-slide saturation-Otsu detection excluded normal tissue, chiefly tissue with large clear spaces such as adipose tissue and alveolar parenchyma, and on slides with thick marker ink kept the ink while excluding ordinary tissue. Replacing it with entropy-based detection reduced the false-positive fraction on held-out clean slides from 0.102 to 0.016, in every fold and with a second training seed, without loss of sensitivity; the gain came from the composition of the pool, not its size. Across three tissue detection methods, false positives followed the fraction of such clear-space tissue in the pool, a statistic that needs no labels or training (0.103, 0.016 and 0.009). The effect did not carry over at the same size to a nearest-neighbour detector on foundation-model features. On an external cohort, the curated pool lowered clean-control false positives by about 20%, far less than within TCGA, and the remaining cross-center loss was not explained by stain differences. For one-class quality control, tissue detection decides what the model learns as normal and should be chosen and reported accordingly. --- ### LongEmo: Towards Emotion Understanding and Reasoning in Long Videos Authors: Shuo Zhang, Yifan Zhou, Han Wang, Jinsong Zhang, Jingyu Li, Hongbing Li, Zhejun Zhang, Chengyi Zhao, Yuquan Hao, Yitong Liu, Jiyin Li, Ruiqi Tang et al. Published: 2026-09-30 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.40079v1 While recent Multimodal Large Language Models (MLLMs) have shown promise in affective computing, their reasoning capabilities are largely confined to short video clips with limited interactions. However, real-world emotions are not merely isolated instantaneous reactions but dynamic and cumulative processes deeply shaped by past experiences and ongoing events. To bridge this gap, we introduce LongEmoBench, a benchmark dedicated to emotion understanding and reasoning in long videos. It assesses progressive capabilities scaling from continuous scene interactions to complex episodic developments. Furthermore, we propose LongEmo, a novel memory-augmented agentic framework designed to tackle the immense challenges of long-range affective reasoning. LongEmo processes continuous video streams to construct an Event Memory Graph, explicitly modeling long-range dependencies and capturing emotional dynamics across discrete events. Given a question, the agent retrieves a query-relevant event stream from the graph, iteratively integrating multimodal memories and relational dependencies to deduce the final answer. Extensive evaluations of 17 representative methods reveal that they struggle significantly with emotion understanding and reasoning in long videos. In contrast, LongEmo achieves state-of-the-art performance, demonstrating the efficacy of its event-centric memory architecture. --- ### Robust and Learned Online Matching in Growing Trees Authors: Marek Gałązka, Hanna Wdowicka Published: 2026-09-30 Categories: cs.DS, cs.LG, math.PR arXiv: https://arxiv.org/abs/2609.40077v1 We study irrevocable maximum-cardinality matching in trees revealed by successive leaf attachments, with a known horizon and an exogenous growth law that is misspecified or unknown. For deterministic affine attachment forecasts with nonnegative degree reinforcement, the optimal threshold policy loses at most twice the cumulative expected conditional total-variation error relative to an online oracle knowing the actual growth law. This follows from a unit-span property of the Bellman continuation score and has no additional horizon factor. A four-vertex example attains the coefficient two for the specified deterministic policy, and a two-model argument gives a lower bound linear in the model-error budget for arbitrary policies under general misspecification. For uniform-preferential attachment, the local error has an exact expression through the leaf count. When its constant mixture parameter is unknown, we estimate it from the same growing tree and update the threshold policy at geometric times. A parameter-sensitivity bound for individual Bellman prices and uniform degree-moment estimates yield expected regret $O(\sqrt{n}\log^2 n)$, using $O(n^2\log n)$ arithmetic operations and $O(n)$ stored entries. The exact minimax rate remains open. --- ### Accelerated Algorithm for Sparse Regularized Partial Optimal Transport Authors: Khoa Nguyen, Dung T. Nguyen, Thong Huynh, Hoang-Hiep Nguyen-Mau, Anh Nguyen, Minh Ngoc Dinh, Juho Kannala Published: 2026-09-30 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.40075v1 Partial Optimal Transport (POT) extends the classical optimal transport problem by relaxing the strict mass conservation constraint, enabling its use in a wide range of real-world applications. In many of these settings, sparse transport plans are preferred for their interpretability and computational benefits. While smooth and strongly convex regularizers - such as quadratic or elastic net - have been vastly used in various machine learning applications to induce sparsity and accelerate computation, they have received less algorithmic attention compared to entropic approaches for computational POT. In this paper, we propose a new optimization framework that leverages these regularizers through a penalty-based reformulation, enabling efficient gradient-based updates while preserving the structure of the original problem. Our method accommodates a broad class of regularizers that promote structured and sparse transport plans. Building on this formulation, we design an accelerated first-order algorithm that alternates between smooth updates and simple projection steps. Through empirical benchmarks on color transfer, domain adaptation, and point cloud registration, our approach consistently outperforms established baselines - achieving lower transport cost, higher sparsity, and faster convergence - making it a practical and scalable solution for modern transport problems. --- ### Grounding Time-Series Foundation Models in Digital Twin Topology for Predictive Maintenance Authors: Sizhe Ma, Katherine A. Flanigan, Mario Bergés Published: 2026-09-30 Categories: eess.SP, cs.AI arXiv: https://arxiv.org/abs/2609.40071v1 Digital twins increasingly support downstream analytical tasks that depend on time-series data, motivating interest in time-series foundation models (TSFMs) as scalable backbones. However, TSFMs are primarily pretrained for temporal continuation and often underperform on unseen tasks such as regression, and systematic empirical comparisons against state-of-the-art dedicated models in digital twin contexts remain limited. This paper makes three contributions. First, we benchmark five well-known TSFMs with frozen backbones on remaining useful life (RUL) prediction using the C-MAPSS dataset, finding that multivariate architectures substantially outperform univariate ones, particularly under varying operating conditions. This raises a deeper question: when cross-channel dependencies can be modeled through pretrained weights, target-task adaptation, and digital twin-derived representations, how much does each contribute, and are they complementary? Second, we propose a topology-informed fusion approach in which topological constraints, derived from the asset structure the digital twin stores among its information models, explicitly shape cross-attention, so that fused representations respect the physical system's local connectivity rather than relying on unconstrained all-to-all interactions. Third, we conduct an ablation study across C-MAPSS subsets of varying operational complexity that isolates the three sources and their interactions. The sources prove complementary rather than redundant, and topology-constrained attention outperforms unconstrained fusion, though by a small margin, enabling a frozen TSFM informed by digital twin representations to remain competitive or in some cases exceed state-of-the-art performance on this regression task. --- ### Inference Auctions Authors: Keegan Harris, Siddharth Prasad, Asher Trockman, Nika Haghtalab, Michael I. Jordan Published: 2026-09-30 Categories: cs.LG, cs.AI, cs.GT arXiv: https://arxiv.org/abs/2609.40070v1 When inference demand exceeds available compute capacity, model providers must decide which requests should be served first. Users have different tolerances for delay from an LLM API, but current priority pricing schemes compress these differences into coarse fixed-price service tiers. We design an inference auction that allows users to bid for faster service. Our auction allocates priority in an economically efficient way without sacrificing latency, and we develop fast algorithms for implementing prices that incentivize truthful bidding. We also design an autobidding agent for our inference auction, where users specify an inference budget and the autobidder dynamically adjusts its bids over time to maximize user utility subject to the budget constraint. Experiments validate the practicality of our auction: it increases system welfare while maintaining the cache utilization and latency advantages of SGLang, a state-of-the-art inference serving framework. --- ### Community-Driven API and AI Writer Design for Openly Scaling Community Notes Authors: Brad Miller, Jay Baxter, Jiansong Chao, Keith Coleman, Sophie Hilgard, Daniel Ortiz Published: 2026-09-30 Categories: cs.SI, cs.AI arXiv: https://arxiv.org/abs/2609.40067v1 Community Notes is a crowd-sourced approach for adding context to posts on X. Contributors propose and rate notes, forming the inputs to an open-source, open-data algorithm that determines which notes show broadly to users. Since September 2025, Community Notes' AI Note Writer API has provided an open, public interface for using AI to propose notes, while adhering to the founding principle that users, not the platform or an AI, control which notes show on X. Explicit note requests and user posts on X determine the AI API post feeds, ensuring that AI note writing responds to demand from X users. We present the design, operation and impact of the AI API, including analysis of the interaction between AI and human generated notes across topics. Unless otherwise stated, measurements and system description reflect June 2-29, 2026. The Community Writer is the largest AI API client and contributes the bulk of AI API output, generating 52% of notes selected as Helpful and shown broadly on X. The writer is guided by community input during both training and operation to prioritize, draft, evaluate and delete proposed notes. Beyond scale, the writer also offers speed, submitting the first proposed, non-deleted note on 60% of posts when compared to other writers. AI note writing is additive on top of human note writers, extending coverage of Community Notes on X. Among posts that have Helpful notes, 42% have only AI notes, indicating human raters did not feel motivated to propose an alternative. In contrast, 30% have only human notes, reflecting contribution beyond the scope of AI writing. The Community Writer is open-source software released under the Apache 2.0 license. --- ### From Tweets to Trades: Analyzing the Influence of Public Mood over Stock Market Performance in Turkiye Authors: Ece Elif Adak, Bertaç Şakir Şahin, Şaziye Betül Özateş Published: 2026-09-30 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.40064v1 Purpose: This study examines whether domain-specific public mood is associated with stock-market dynamics and whether these relationships vary across communication domains and market conditions. It distinguishes public mood from investor sentiment and investigates whether heterogeneous sources of public communication exhibit different relationships with market behaviour. Design: The study analyses 610,422 posts published by 176 curated X accounts between January 2022 and December 2023, covering Politics and Government, Economy and Finance, and Media and Society. Posts are classified using fine-tuned Turkish transformer models under three domain-specific and one pooled regime. Public mood measures are constructed at daily, weekly, and monthly frequencies and examined alongside BIST100 and BIST30 market measures using correlation, Granger causality, vector autoregression, and impulse response analyses across the full period and selected market conditions. Findings: Public mood is not associated with the direction of stock-market returns but is associated with the magnitude of price movements, particularly for Media and Society and pooled communication. These relationships become stronger at longer aggregation frequencies. Predictive relationships are concentrated in Economy and Finance communication, while their magnitude and direction vary across market conditions, particularly during the 2023 election period. The pooled measure largely reflects the most active communication domain. Originality: The study contributes to behavioral-finance research by incorporating communication - domain heterogeneity into the analysis of public mood and market dynamics. It also demonstrates how aggregating heterogeneous sources can obscure domain-specific relationships between public communication and financial markets. --- ### LARC: Low-Rank Adaptive Residual Connections for Learning in Frozen Models Authors: Junyi Zou, Avrova Donz Published: 2026-09-30 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2609.40063v1 Low-Rank Adaptive Residual Connections (LARC) give a frozen model a compact numerical state that can learn from feedback. The map $h+BAh$ adds a low-rank correction to a hidden representation. A slow state $ρ$ learns starting factors across tasks; a private fast state $Φ$ copies them, changes with feedback, and resets to the trained initialization. This report specifies an input-side realization of the numerical policy carrier in Memory-Mediated Learning Architecture and examines its factor-space dynamics and learning lifetime. We study a rank-4 input residual with 12,288 trainable parameters on a frozen MiniCPM5-1B-SFT substrate. In a four-candidate program-selection task, two feedback-gradient steps reduce expected query execution error by 24.65 and 36.65 percentage points relative to resetting to the respective trained static and post-adaptation initializations. These development results cover 16 parameter groups and three paired training seeds. A direct support-loss selection rule is much more accurate, reaching 0.78125% error. In a repository-balanced chronological replay of public continuous-integration jobs, retaining online updates raises half-Brier loss from 0.1274 to 0.1808. A fixed follow-up intervention records same-batch non-descent and inconsistent future benefit from shrinking updates. Together, the algebra and measurements distinguish residual capacity, adaptation relative to a starting point, and usefulness on later decisions. --- ### Less Data, Better Timing: Student-Curriculum Coupling for VLM On-Policy Distillation in Temporal Video Grounding Authors: Jiacheng Qiu, Yunsoo Kim, Ruichen Xu, Jian Luo, Petar M. Djurić, Sima Mofakham Published: 2026-09-30 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.40055v1 On-policy distillation (OPD) provides dense supervision directly on student-generated trajectories, making it an effective post-training strategy for vision-language models in temporal video grounding (TVG). However, existing pipelines typically construct the training curriculum from a fixed teacher and the initial student state, implicitly assuming that selected examples retain positive supervision value throughout optimization. We show that supervision trustworthiness and supervision necessity are distinct yet coupled: the former concerns target credibility, while the latter varies with the student's current task competence; together, they shape supervision value. Building on this coupled view, we introduce Student-Curriculum Coupling (SCC), a closed-loop framework in which a compact Anchor-Frontier curriculum defines the candidate supervision space and the evolving student dynamically determines its active subset. Supervision can therefore be activated, suspended, or reactivated as competence changes, concentrating teacher computation and optimization on current task-level deficits. Across three TVG benchmarks, SCC achieves a 5.1% relative improvement in mean recall over Video-OPD on its original curriculum, while using 60.0% fewer training examples and reducing training time by 50.4%. Ablations support the complementary roles of capability-structured curriculum design and student-dependent supervision in achieving these gains. Together, these results establish SCC as a data- and compute-efficient framework for TVG post-training, delivering stronger temporal grounding by aligning trustworthy supervision with the student's evolving learning needs. --- ### Proximal Balancing for Causal Effect Estimation under Unmeasured Confounding Authors: Yonghan Jung Published: 2026-09-30 Categories: stat.ML, cs.LG, stat.ME arXiv: https://arxiv.org/abs/2609.40051v1 Estimating causal effects from observational data is central to science and policy, but the effects are not identified when confounders are unmeasured. Proximal causal inference addresses this problem with proxies of the unmeasured confounders. However, existing proxy-based approaches either designate proxy roles and solve an inverse problem, which is ill-posed and hard to estimate with high-dimensional proxies, or use a latent-variable model, which assumes that the learned latent variable matches the hidden confounder and leaves bias when it does not. To address these challenges, we introduce proximal balancing. It carries the classical idea of covariate balancing to confounders that are observed only through proxies: it learns a low-dimensional summary of the covariates and proxies that makes the treatment groups comparable, and then adjusts for this summary. It needs no designated proxy roles, inverse problem, or latent model. We give identification theory, finite-sample guarantees, and a practical algorithm, PROBE. We demonstrate the method on low-dimensional, high-dimensional, and image proxies and on real-world data. --- ### CoEvoWhen: Policy-Tool Coevolution for Ultra-Long Video Temporal Grounding Authors: Yiduo Jia, Muzhi Zhu, Jinchuan Shi, Hao Zhong, Yuling Xi, Ke Liu, Hao Chen Published: 2026-09-30 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.40048v1 Ultra-long video temporal grounding requires balancing long-range evidence search with fine-grained event understanding under a limited visual budget, yet existing agentic methods still rely largely on predefined policies and tool capabilities. Motivated by this, we propose a novel policy-tool coevolution framework that jointly evolves high-level policies and executable media tools from the agentic reasoning trajectories of a VLM, forming a reusable skill without updating model parameters. During evolution, an external skill updater distills transferable task experience in long-video temporal grounding, accordingly refining the orchestration of long-range image-based and fine-grained video-based observations. Alongside these policy updates, the updater employs its coding capabilities to upgrade existing tools or create new ones, adapting the tools to long-video evidence acquisition. Equipped with the evolved skill, the VLM autonomously orchestrates tools under the guidance of the evolved policy, coordinating image and video observations for agentic inference without relying on a separate, stronger planning model. Extensive experiments spanning five benchmarks and three VLMs show that policy-tool coevolution consistently improves temporal grounding accuracy in ultra-long videos while reducing visual token cost at inference, and that the evolved skill yields substantial performance gains on general long-video QA without additional task-specific evolution, demonstrating the effectiveness and generalizability of our framework for long-video understanding. --- ### Gromov-Wasserstein Distillation for Inductive Multi-View Embedding Authors: Rafael Pereira Eufrazio, Eduardo Fernandes Montesuma, Charles Casimiro Cavalcante Published: 2026-09-30 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2609.40047v1 Gromov-Wasserstein multidimensional scaling (GW-MDS) learns low-dimensional representations from relational data but remains transductive, providing no explicit mapping for unseen samples. We introduce an inductive framework based on barycentric distillation. A GW-MDS teacher learns a latent support and an optimal transport plan from the training data, and barycentric projection converts the resulting coupling into sample-aligned targets. A neural student then learns an explicit out-of-sample mapping, avoiding additional relational-matrix construction and GW optimization at inference. We formulate the approach for single-view data and extend it to Mean-GWMDS and Multi-GWMDS teachers through consensus and selected-projection targets learned by a multi-view student with view-specific encoders. We also investigate a direct neural baseline trained solely with a GW objective. Experiments on synthetic and real-world data using Euclidean, geodesic, and cosine relations show that the distilled models preserve the teacher geometry on unseen samples and consistently outperform direct neural GW training in sample-indexed relational preservation. These results establish barycentric projection as an effective bridge between transductive GW embeddings and inductive neural mappings. --- ### MAGiDiff: Sampling the Photospheric Vector Field from UV/EUV Filtergrams Authors: Ruoyu Wang, David Fouhey Published: 2026-09-30 Categories: astro-ph.SR, astro-ph.IM, cs.CV arXiv: https://arxiv.org/abs/2609.40043v1 Photospheric vector magnetic fields are foundational to modeling, understanding, and forecasting solar activity. These data are usually produced by inverting and disambiguating the full Stokes vector at multiple passbands, which is demanding. Here, we investigate how well we can estimate photospheric vector magnetograms from UV/EUV filtergrams. This problem is challenging and intrinsically ambiguous without polarization information, as the mapping from UV/EUV intensity to the magnetic field is indirect and ill-posed. We introduce MAGiDiff, a machine-learning-based method that uses denoising diffusion models to estimate vector magnetograms from UV/EUV filtergrams. As input, MAGiDiff takes a stack of filtergrams from the Solar Dynamics Observatory (SDO) / Atmospheric Imaging Assembly (AIA); as output, it is trained to estimate the disambiguated vector magnetogram as seen by Hinode / Solar Optical Telescope-Spectro-Polarimeter (SOT-SP). We show that MAGiDiff can accurately mimic the Hinode ground-truth. Additionally, we probe MAGiDiff's understanding of the physical structure and magnetic connectivity. On full-disk, we show that it produces plausible structures for active regions. MAGiDiff generalizes across solar cycles despite hemispheric polarity reversal, and can be fine-tuned to other EUV instruments including STEREO/EUVI and GOES-R/SUVI. While clearly not a substitute for a dedicated instrument, MAGiDiff opens the door to new capabilities. --- ### MGhana-ST: A Low-Resource Speech Translation Dataset for Ghanaian Languages and an Analysis of Multilingual Training Trade-offs Authors: Frank Lawrence Nii Adoquaye Acquaye, Eric George Parakal, Jesse Johnson, Kishankumar Bhimani, Jochebed Afua Basil Published: 2026-09-30 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.40041v1 We present MGhana-ST, a speech translation dataset for four low-resource Ghanaian language varieties: Ga, Twi (Akuapem and Asante), Ewe, and Fante. MGhana-ST is an ongoing annotation effort; the experiments here use a fixed subset of about 16.1 hours of paired speech and English translations. The audio is curated from two existing Ghanaian speech resources. Unlike in those resources, the English translations are produced directly from audio by 37 native-speaker annotators and include verbal and non-verbal event annotations. Using Whisper-small, we compare monolingual and multilingual training under severe data scarcity, reporting means over three seeds. Flat multilingual training benefits no variety in this regime. Ga and Twi are unchanged within seed variance (+0.51 and +0.06 BLEU against monolingual standard deviations of 1.63 and 2.20), while Ewe declines by 6.99 BLEU and Fante by 5.11. The degrading varieties are Ewe, which is linguistically distinct and drawn from a different source corpus, and Fante, the least-resourced. Comparing empirical cross-lingual transfer with typology-based similarity, we find that transfer BLEU identifies closely interacting language pairs better than URIEL similarity, though neither predicts which varieties benefit from joint training. We also report a methodological finding. An earlier single-run analysis found positive transfer for three of four varieties; this did not survive replication across seeds. For Ga and Twi, monolingual baselines trained on 1.6 to 6.2 hours of audio have seed standard deviations roughly five and thirty times those of the multilingual models (0.35 and 0.07 BLEU). When the monolingual condition is noisier, a single-run comparison can show apparent transfer of this size from seed variation alone. We release MGhana-ST to support research on African language speech technology and low-resource speech translation. --- ### Enhancing Autoregressive Video Generation via Representation Adversarial Distillation Authors: Fangyu Lin, Xingtong Ge, Lunjie Zhu, Yi Zhang, Zhening Liu, Tianhang Wang, Mengfei Li, Yumeng Zhang, Guanglu Song, Yu Liu, Jun Zhang Published: 2026-09-30 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.40037v1 Few-step autoregressive video generation enables efficient streaming synthesis, but errors introduced in early temporal blocks are reused as context and can propagate through subsequent rollouts, leading to detail degradation, structural drift, and unstable motion. Existing distribution matching distillation (DMD) primarily aligns student and teacher distributions in diffusion latent space, but provides no direct supervision over the perceptual quality of decoded videos. We introduce Radian, a representation-space adversarial distillation framework that complements on-policy DMD with real-data adversarial supervision in the feature space defined by a frozen visual foundation model (VFM). During training, Radian sparsely decodes frames from autoregressive student rollouts, extracts multi-level visual representations, and applies lightweight discriminator heads to distinguish generated outputs from real video frames. The DMD objective anchors the student to the pretrained teacher, while the representation-space adversarial objective supplies complementary perceptual and semantic gradients that promote high-quality modes. These additional components are discarded after training, leaving the generator architecture and inference-time denoising budget unchanged. Experiments on Wan2.1-1.3B cover four-step chunk-wise, one-step frame-wise, and minute-long autoregressive generation. Our method achieves a VBench Total of 0.8444 and a VideoAlign Total of 0.8033 under four-step generation, and improves VBench-Long from 0.7805 to 0.8041 over Rolling Forcing while using fewer denoising steps. Controlled comparisons across image, video, and diffusion representations further indicate that the choice of representation spaces induces distinct adversarial signals, and external VFM gradients complement DMD more effectively than adversarial supervision derived from diffusion-internal features. --- ### OPTS-TTPO: Enhancing Finite-Sample Policy-Gradient Learning with Tree Search Authors: Junyu Lu, Shichao Weng, Zhiqiang Wang, Haojie Luo, Jingfan Zhang, Yuhua Zhou, Cheng Du, Yuzhuo Zhang, Xi Li, Jinwei Du, Tiancheng Feng, Chuan Xiao et al. Published: 2026-09-30 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.40035v1 The policy-gradient theorem gives the exact gradient under the current policy, but finite on-policy samples may miss rare high-return trajectories. We study whether tree search improves their coverage within a fixed budget while controlling gradient bias. We introduce On-Policy Parallel Tree Search (OPTS) and Tree Trajectory Policy Optimization (TTPO) using on-policy tree trajectories, which sample new suffixes from the current policy at visited states. This needs no action-distribution correction, although branching changes state visitation. Our Branch Aggregation Lemma shows that branch-weighted tree statistics recover chain expectations when branch choices and weights are fixed before outgoing transitions are sampled. OPTS selects expansion states using estimated performance differences. Under deterministic dynamics, exact values, and max-backup advantages, the induced search policy's expected return improves monotonically with the budget. We bound the gradient bias from adaptive expansion and show that max backup assigns prefix credit to actions leading to better discovered suffixes. Against a finite chain reference, TTPG's measured bias stays near its no-branching level, while NaivePG's bias grows from 0.1251 to 0.4884. At matched budgets, reward- and value-guided OPTS improve correct-answer coverage and majority-vote accuracy over independent sampling. At matched branch counts, OPTS + TTPG gains coverage with a modest bias increase relative to Fixed-branch + TTPG. Under matched interaction or rollout budgets, OPTS-TTPO improves MuJoCo tail returns over PPO by up to 28.6%, achieves a 34-22-1 win-loss-tie record against PPO on Atari-57 under the last-100-log mean-return metric, and improves micro-averaged avg@32 and pass@32 over PPO across all four Qwen3 models. ---