# Research Papers — 2026-08 · bundle page 66 of 240 > Full-text markdown of papers #3251–3300 of 11995 published in 2026-08, newest first. AI-readable mirrors of the full 11995-paper 2026-08 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-08/page/66 [← Previous bundle](https://aipod.dev/papers/archive/2026-08/page/65.md) [Next bundle →](https://aipod.dev/papers/archive/2026-08/page/67.md) ### PatchWrite: One Line, Not One Section -- Compile-Gated, Validity-Preserving Editing for AI-Drafted Manuscripts Authors: Weiwei Yang Published: 2026-08-24 Categories: cs.AI, cs.CL, cs.SE arXiv: https://arxiv.org/abs/2608.23001v1 Automated manuscript pipelines often regenerate an entire section to repair a local defect, allowing unrelated metrics and citations to change even when the resulting PDF still builds. PatchWrite instead constrains how candidate edits become committed manuscript states: it reuses bounded EDIT N M editing and rollback, but tightens compilation acceptance with fatal-log checks and adds evidence locks that require every cited key and experimental numeric token to be attested by a reference registry or experimental log. Candidates that fail either check are rejected and the previous HEAD is retained. On a 24-manuscript x 8-fault oracle stress test (768 jobs, evenly split between compile-breaking and content-only faults), whole-slot rewriting mutated an unrelated "12-layer" line in every case (0/192 preserved; numeric Jaccard 0.6667), whereas PatchWrite preserved it in 192/192 cases. Removing the compile gate reduced acceptance to 0, while removing the evidence gate allowed a hallucinated citation to pass. The same pattern held across all eight faults. To test the protocol with generation rather than oracle edits, we reran the 192 jobs with the writer model proposing the edits. The model's candidates were accepted in 75% of cases; nearly all rejections came from one reproducible failure mode in which the model attempted to delete a line using an empty replacement unsupported by the current grammar. Every accepted candidate passed both gates, and 93.75% fixed the injected fault; the remaining cases involved a technically valid but sentence-inappropriate citation and one markup-changing near-miss. In a blind evaluation of sixteen PDF pairs, both raters preferred PatchWrite for preserving lab-grounded facts (C1 Likert 5.0 vs. 2.0), while rating prose quality nearly identically. Logs from 193 in-product drafting tasks show the same classes of failures occurring in practice. --- ### ENCORE: Entropy-Guided Cropping and Attention Regularization for Robust Vision--Language Understanding Authors: Yuanhao Sun, Huawei Ji, Jiaxin Ding, Luoyi Fu, Xinbing Wang Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22996v1 Vision-Language Models (VLMs) perform well on diverse vision-language tasks, but transformer-based visual encoders split images into fixed-resolution sub-images, compromising object integrity in lightweight VLMs. Existing methods only focus on the visual modality and fail to dynamically preserve the integrity of prompt-relevant regions, limiting performance. In this work, we observe that the early-layer image-text entropy of cross-modal attention strongly correlates with answer grounding quality and task accuracy. Building on this finding, we propose \textbf{ENCORE}, an entropy-guided framework with two components: At inference, an \textbf{Entropy-based Cropping Strategy} (ECS) evaluates a small set of candidate crops and selects the one with minimal entropy, preserving contiguous regions relevant to the prompt. At training, \textbf{Entropy Regularization Training} (ERT) augments next-token prediction with an entropy term that sharpens attention on key visual tokens while down-weighting irrelevant ones. Experiments on ten VQA benchmarks show that ENCORE, fine-tuning only 0.14\% of parameters, achieves an average 1.43\% accuracy gain and state-of-the-art performance among recent 2B-parameter VLMs. Our code is released in https://github.com/baokou-fw2/ENCORE. --- ### A Physical Response-and-Memory Model for Muon Optimization Authors: Yinze Hu, Hongjun Xiang, Xingao Gong, Hongyu Yu Published: 2026-08-24 Categories: cs.LG, cond-mat.dis-nn, cond-mat.stat-mech, cs.AI, physics.comp-ph arXiv: https://arxiv.org/abs/2608.22994v1 Training large language models is costly. How low a loss the same compute can ultimately reach depends on how each step's gradient is converted into a weight update; the rule that performs this conversion is the optimizer. From SGD and AdamW to the recent Muon, effective update rules have mostly been shaped by engineering intuition and then selected on benchmarks. Muon semi-orthogonalizes the momentum matrix before applying the update and has kept breaking records on public training benchmarks; yet why the semi-orthogonalized direction works, and over how long a history the momentum should average, are two questions at present answered mainly by experience. Here we treat the weight matrix during training as a responsive medium with memory and build a physical model for it, in which both questions find answers: the semi-orthogonalized direction is the maximally dissipative response under an output-side safety budget, which explains why it works; momentum is the internal stress accumulated by the medium; how long it should average is set by the relaxation of this stress, and a real medium relaxes on more than one timescale, the simplest form being one fast and one slow. On this basis we propose the Bi-Maxwell optimizer. The framework further yields a testable consequence: gradient directions change fast early in training and more slowly later, so the optimal memory length should grow with training stage; step-by-step measurements of a proxy for it by a read-only probe across 8 independent training trajectories are consistent with this consequence. Replacing the memory kernel alone, from a single timescale to two, brings training to the target loss in noticeably fewer steps on a public large-language-model optimizer benchmark. --- ### LLM Pedagogical Behavior in AI Tutoring Interactions Authors: Suhyeon Lee, Juneha Baek, Jaehyeong Park, Donghyuk Shin Published: 2026-08-24 Categories: cs.CL, cs.HC arXiv: https://arxiv.org/abs/2608.22993v1 Students increasingly use LLMs as tutors for coursework and problem solving. Little is known about the level of assistance LLMs provide when students use them as tutors in authentic learning interactions. This matters because tutoring responses can differ substantially in how directly they help students complete a task. We operationalize this dimension as scaffolding level and develop a five-level scale, validated against human annotations, that characterizes responses according to the degree of direct assistance they provide. We apply the scale to 14,637 LLM responses from 203 students in a university AI course. Responses are overwhelmingly concentrated at high levels of assistance, with more than 95% classified as either Explaining or Solving. Scaffolding level is systematically associated with students' subsequent conversational behavior, but provides little additional predictive information about performance on three subsequent exams beyond prior achievement and dialogue behavior. These findings provide an empirical baseline for LLM assistance in tutoring interactions and a measurement framework for evaluating how alternative tutoring designs change that assistance. --- ### Channel-Token Attention for Reliable Dynamic Spectrum Access under Bursty Primary-User Traffic Authors: Krishna Acharya, Dinanath Padhya, Utsab Dahal, Ashish Kandel, Binod Sapkota Published: 2026-08-24 Categories: cs.NI, cs.LG, eess.SP arXiv: https://arxiv.org/abs/2608.22992v1 Dynamic spectrum access must coordinate secondary users under bursty primary-user activity while preserving packet reliability and delay. We present TACAN, a centralized policy that represents each channel as a token containing occupancy history and automatic-modulation-classification entropy; a context token supplies queue class, delay and user identity. A Transformer encoder is warm-started from an occupancy-greedy policy and refined with proximal policy optimization. The frozen policies were trained to maintain a channel assignment in every slot, including when queues were empty. We therefore replay them on held-out trajectories and distinguish standby assignment success from packet-present access and packet delivery. In a 20-channel network with 60 primary devices and 4 secondary users, TACAN achieves 92.53% +/- 0.47 packet-present access success, compared with 89.94% for Greedy and 83.53% for PPO+MLP. Its paired gain over Greedy is 2.59 points (parametric 95% CI 1.89-3.29), with wins in all five seeds; the exact two-sided sign-test value is 0.0625. The gain rises from 0.57 points at normal primary-user load to 7.67 points at extreme load. TACAN also reduces mean delivery delay from 1.208 to 1.123 slots and the conditional user-reliability gap from 9.69 to 3.15 points. Delivered packets per SU-slot remain arrival-limited (30.12% versus 30.11%), so no packet-throughput gain is claimed. --- ### What Does Activation Steering Control? Attribution Across Answer Encodings and Output-Sensitive Subspaces Authors: Zhiwei Gao, Shaowen Peng, Shoko Wakamiya, Eiji Aramaki Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.22985v1 Activation steering is often evaluated under the answer encoding used to construct the direction. A reported gain may reflect the intended judgment or compatibility with answer identifiers seen during construction. We introduce Cross-Encoding Steering Evaluation, which freezes an intervention while re-encoding answers to the same held-out items. On NormBank, after A/B/C identifiers are reassigned, contrastive activation addition (CAA) induces larger target-versus-source score changes for the extraction indices than for the semantic labels under the new mapping. We call this extraction-index following. Varying identifier vocabulary (A/B/C, X/Y/Z, or 1/2/3) and row order shows that the effect tracks extraction index rather than row position. After matching direction norms across layers, extraction-index following emerges mainly at later depths. A low-rank output-sensitive component containing 15.4% of the direction's squared norm retains 96.3% of this effect. An Inference-Time Intervention (ITI)-style method also favors extraction-index over semantic-label following on NormBank in three models. In aggregate, MNLI favors extraction-index following, whereas Social Chemistry 101 (SC101) favors semantic-label following. Multiple-choice and open-ended evaluations can yield different behavioral conclusions. Thus, a steering gain under one answer encoding does not by itself identify what the intervention controls. --- ### Hierarchy-Aware Semantic Losses for Knowledge Graph Link Prediction Authors: Filip Kronström, Ross D. King Published: 2026-08-24 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.22981v1 Knowledge graphs are often accompanied by ontological class hierarchies that encode valuable semantic information, yet many link prediction methods either ignore such hierarchies or incorporate them indirectly through additional graph edges. Recent work introduced hierarchy-aware graph neural networks (GNNs), which use semantic losses derived from box embeddings to encourage satisfaction of subclass relationships during GNN-based representation learning. While this approach has shown promise for biological regression tasks, its effectiveness for knowledge graph link prediction has not been investigated. In this paper we evaluate hierarchy-aware semantic losses on link prediction across three benchmark datasets: AIFB, CoDEx, and BioKG. We combine graph neural network encoders with box-embedding-based semantic losses that encourage learned representations to better satisfy ontology-derived class hierarchies, and compare this approach to both standard link prediction models and models incorporating subclass relations as graph edges. Across all datasets, hierarchy-aware semantic losses significantly improve mean reciprocal rank (MRR) and consistently outperform models that incorporate hierarchy information through additional subclass edges. Relative to the baseline GNN models, MRR improved by 7.6%, 2.4%, and 15.5% on AIFB, CoDEx, and BioKG, respectively. Furthermore, semantic losses consistently outperform the alternative of augmenting the graph with subclass edges. These results are consistent with ontology-derived class hierarchies providing complementary information to graph structure, and suggest that encouraging hierarchical consistency through semantic losses is an effective and comparatively parameter-efficient mechanism for improving knowledge graph link prediction. --- ### Hypergraph Embedding Indexing for Efficient Dense Vector Retrieval Authors: Kishore Konda Published: 2026-08-24 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2608.22980v1 Dense vector retrieval has become the foundation of modern semantic search, yet existing approximate nearest neighbor (ANN) indexes treat an embedding as an indivisible point in a high-dimensional space. In this work, we propose the Hypergraph Embedding Index (HEI), a framework that instead organizes documents according to combinations of highly activated latent embedding dimensions. This formulation enables inverted-index style candidate generation while preserving the semantic ranking capabilities of dense embeddings. We further demonstrate that constructing multiple complementary hypergraphs substantially improves retrieval coverage without the combinatorial growth associated with increasing the dimensionality of a single hypergraph. Finally, we establish that the statistical properties of embedding activations strongly influence coordinate-inverted indexing efficiency, introducing \emph{activation diversity} as a diagnostic metric governing embedding indexability in coordinate-inverted frameworks. --- ### SA-RSQ: A Versatile Sparse Representation Framework for Multi-modal Recommender Systems Authors: Xiang Wang, Shigang Quan, Tingzhen Chang, Kang Yang, Sitong Chen, Yabo Fan, Xingxing Wang, Zhaodian He Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.22979v2 Deploying high-dimensional multimodal features in industrial recommender systems incurs substantial storage and latency overhead. Hard quantization is compact but introduces boundary distortion, whereas dense soft quantization couples representation quality to the limited storage budget. We propose Sparse Activation-based Residual Soft Quantization (SA-RSQ), which uses Top-K sparse routing and softmax weights to store compact (Index, Probability) tuples. The stored tuples decouple per-item storage from codebook dimensionality; for a fixed selected support, gradients propagate through the routing weights and weighted reconstruction without relying on a straight-through estimator. Experiments on a proprietary food-delivery advertising dataset show favorable reconstruction-performance and CTR trade-offs across storage budgets of 8-48 bytes per item. A preliminary Next-Distribution Prediction study and a one-week online A/B test further demonstrate the practical potential of SA-RSQ, with relative lifts of +2.51% in CTR and +3.66% in CPM. --- ### Budget-Constrained Embodied Perception: Four Resource Walls and a Pre-Registered Evaluation of Access-Structured Perception on Open Models at less than 31B Authors: Defu Lin, Wenhui Chen, Ziyao Lin, Jianlin Chen, Peiji Long, Chi Man Vong Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.22975v1 Embodied multimodal agents must answer from growing observation streams under a fixed per-decision token budget. We formalize this constraint through four resource walls: a perceptual Shannon wall for bounded state, a horizon wall for query-independent frame selection, a round wall for non-adaptive retrieval, and a conditional composition wall for fixed-depth inference. We introduce ASP, a training-free wrapper for frozen multimodal models that combines a capped structured state, a verbatim episodic index, and query-conditioned budget allocation with iterative access. Following a pre-registered protocol, we evaluate seven open-weight models from 3B to 31B on SEW-Bench, a license-free synthetic long-horizon walkthrough benchmark constructed to instantiate these walls. The registered natural-video benchmarks were not run because their frames require dataset agreements; our evidence therefore concerns access mechanisms, not natural-scene perception. Under a 4,096-token decision budget, ASP reaches 75 to 94% episodic retrieval accuracy, compared with 3 to 19% for equal-budget query-independent sampling, and budget reallocation outperforms quadrupling the sampling budget on every backbone. However, the full three-component architecture does not validate channel duality: removing the compressive state raises the flagship mean from 35.4 to 58.0, ASP does not outperform the verbatim-only baseline on any backbone, and two of four pre-registered falsification criteria fire. These results show that query-conditioned access, rather than parameter count or context growth alone, is decisive under a fixed budget, while prompted online compression does not earn its cost in this setting. --- ### Toward Effective and Reliable LLM Agents via Dynamic Ontology Authors: Xiaohui Zhang, Zequn Sun, Chengyuan Yang, Yuanning Cui, Lingbing Guo, Wei Hu Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.22974v1 Large language model (LLM) agents rely heavily on knowledge encoded in model parameters or presented as unstructured context. In domain-specific tasks, this leaves important semantic connections implicit. This often results in incomplete evidence use and brittle multi-step decisions. Ontologies offer a way to externalize domain concepts and relations as machine-interpretable structures, but constructing task-usable ontologies traditionally requires substantial effort from domain experts and is difficult to scale. Automatic construction is also challenging: an ontology that appears semantically plausible may not contain the relational structures needed for actual decision making. We present OaK, an ontology-as-a-kernel framework that dynamically constructs and refines task-oriented ontologies for LLM agents. Given task requirements and training data, OaK constructs an ontology and its knowledge graph, generates task-adaptation functions for graph reasoning, and uses judge feedback to iteratively refine both. By making relevant concepts and relations explicit, the ontology grounds knowledge retrieval and multi-step decision making. We evaluate OaK on TravelPlanner, CRMArenaPro, and ToolQA. Results show that OaK improves standard LLM agents, strengthens evidence grounding, and boosts the reliability of multi-step reasoning. --- ### Optimize Surgical Triplet Recognition: A Knowledge-Driven Mixture-of-Experts Solution Authors: Yiyi Zhang, Yuchen Yuan, Ying Zheng, Jialun Pei, Jinpeng Li, Zheng Li, Pheng-Ann Heng Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22972v1 Surgical action triplet recognition constitutes a critical task in context-aware robot-assisted surgery, facilitating automatic surgical action perception by identifying instrument, verb, target, and their association. However, existing works struggle to analyze such complex surgical scenes due to three main issues: (1) component-level optimization conflicts caused by entangled feature spaces, (2) category-level optimization conflicts arising from severe data imbalance, and (3) lack of domain knowledge guidance that limits model interpretability and robustness. To address these challenges, we propose a Mixture-of-Experts-guided Co-Optimization (\textit{MoeCo}) framework powered by knowledge-driven learning. Within the co-optimization pipeline, to first mitigate component-level conflicts, we introduce a component-tailored adapter that disentangles task-specific features across spatial-temporal regimes, facilitating effective component specialization. Next, we develop a coordinated gradient learning strategy to handle category-level conflicts, which adaptively rebalances positive-negative gradients to enhance the perception of rare categories. Notably, inspired by surgical domain expertise, we introduce a knowledge-driven mixture-of-experts mechanism that dynamically integrates multimodal large language model-guided knowledge via activated experts, thereby enriching the co-optimization pipeline with more expressive and robust representations. Extensive experiments on the public CholecT45 and CholecT50 datasets confirm the effectiveness of the proposed co-optimization pipeline and the superiority of dynamic priors integration via the knowledge-driven mixture-of-experts mechanism. --- ### ParallelWorld: Test-Time Scaling for Embodied Reasoning Authors: Min Chen, Shengjun Zhang, Yuxin Li, Zhang Zhang, Xin Fei, Chong Xia, Yueqi Duan Published: 2026-08-24 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2608.22971v1 Embodied Reasoning constitutes a fundamental capability of embodied intelligence, serving as the basis for autonomous perception, reasoning, and interaction within physical environments. Recent studies have shifted the paradigm of embodied reasoning from static perception toward dynamic exploration, where agents acquire task-relevant information through interactions with the environment. However, existing active reasoning approaches generally generate exploration trajectories incrementally without long-horizon planning. Even recently emerged test-time scaling frameworks often resort to myopic, single-step lookaheads, which struggle to resolve the delayed feedback inherent in complex, occluded spatial environments. To address this limitation, we propose ParallelWorld, a multi-horizon test-time scaling framework for embodied reasoning. Instead of greedy, single-step trials, ParallelWorld empowers agents to simulate and evaluate multi-step future trajectories in parallel before committing to an action. Specifically, we introduce a verifier-guided tree-search paradigm. Starting from the current state, ParallelWorld branches into multiple parallel trajectories and rolls them out continuously across a multi-step horizon. At each simulation step, a verifier agent evaluates the intermediate state transitions, dynamically pruning unpromising branches and prioritizing paths with the highest information gain. Once the multi-step prospective simulation is complete, the agent synthesizes the long-horizon outcomes to commit to the optimal action sequence. Finally, an answer agent performs reasoning over the selected trajectory to produce the final reasoning. Extensive experiments on ESI-Bench demonstrate that ParallelWorld consistently improves active perception and reasoning performance. --- ### Do Time-Series Foundation Models Pay Off for Industrial Monitoring? A Cost-Aware Empirical Study Authors: Guan-Hua Wen, Kuan-Yu Chen Published: 2026-08-24 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.22968v1 Industrial monitoring models must detect operationally relevant deviations while satisfying target-specific data, calibration, and resource constraints. Time-series foundation models (TSFMs) promise reusable representations and zero-shot forecasts, yet evidence for their deployment value remains mixed when task definitions are heterogeneous and lightweight baselines are competitive. This work presents a protocol-aware empirical assessment across three settings: a C-MAPSS degradation-risk proxy, normal-only training for anomalous-sound detection on MIMII, and BDG2 forecasting-residual diagnostics with synthetic target perturbations. We assess classical one-class methods, compact neural autoencoders, residual forecasters, MOMENT-small, Chronos-T5, and TimesFM 2.5 in terms of anomaly-ranking performance, risk-horizon sensitivity, residual forecasting and perturbation sensitivity, and local implementation cost. Across 100 C-MAPSS engines evaluated out of fold, TCN-AE reaches fold-weighted AUROC/AUPRC 0.9570/0.8960, compared with 0.7310/0.3080 for MOMENT reconstruction; paired engine-cluster bootstrap confidence intervals exclude zero for both differences. Across five matched MIMII pump evaluations, OCSVM also exceeds MOMENT reconstruction in AUROC and AUPRC. On a fixed 12-meter BDG2 panel, TimesFM 2.5 has the lowest aligned forecast error and the highest synthetic AUROC point estimate, although synthetic AUPRC is similar across TSFM and fitted residual models. Same-device measurements show that MOMENT incurs higher latency, peak allocated VRAM, and serialized state-dictionary size than TCN-AE. Under the evaluated frozen and zero-shot settings, TSFMs are task-dependent deployment options rather than default replacements for fitted lightweight models. --- ### Closed-Loop Bayesian Molecular Inverse Design with Semantic LLM Surrogates Authors: Yaoyao Xu, Xinjian Zhao, Xiaozhuang Song, Lei Bai, Tianshu Yu Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.22967v2 Practical molecular inverse design is rarely a one-shot generation problem; it often takes the form of closed-loop candidate-pool enrichment, where under a limited oracle budget the goal is to increase the fraction of generated molecules that match a desired property profile. Bayesian optimization (BO) offers a natural framework for this setting, yet standard Gaussian-process surrogates typically operate in compressed continuous embeddings, which discard the substructural and reference-similarity signals that chemists naturally use to decide where to look next. We propose BoMolLLM, a closed-loop framework in which the surrogate, rather than the generator, is treated as the locus of design choice, and instantiate it with a frozen large language model that reasons directly over the task instruction, SMILES-level optimization history, and oracle feedback in their native textual form. At each iteration, the surrogate returns a structured decision signal that selects informative reference molecules under an exploration and exploitation principle, optionally with a concise guidance sentence. This signal is converted into next-round conditioning text for a frozen molecular generator, yielding an inspectable optimization trace in natural language. Experiments on MolQA drug and material design tasks show that BoMolLLM improves over one-shot prompting, is competitive with or stronger than GP-based BO baselines, and reveals a domain-dependent interface: reference-only transfer works best for binary drug targets, while adding a concise surrogate summary is more beneficial for continuous material --- ### Simplified Cross-Modal Calibration for Heterogeneous Event-RGB Stereo Systems Authors: Nico Hessenthaler, Adam T. Müller, Nicolaj C. Stache Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22965v1 Accurate extrinsic calibration between event-based and frame-based cameras remains a practical bottleneck for heterogeneous stereo systems. Existing approaches often require sensor or target motion, precise synchronization, or computationally expensive event-to-image reconstruction. We propose a simple, motion-free cross-modal calibration framework that uses a temporally modulated, blended ChArUco target presented on standard consumer displays. By alternating between the original pattern and a partially blended version, the target reliably triggers events while remaining continuously observable to a frame-based camera, avoiding blank frames and reducing synchronization constraints to a coarse, trigger-based alignment. We discretize events into frames coarsely aligned with the RGB images, apply lightweight denoising, and perform ChArUco-based intrinsic and stereo extrinsic calibration. Extensive experiments assess robustness to blending opacity, display brightness, external illumination, viewing angle, and handheld acquisition. Compared to the strongest motion-based reference (E2Calib + Kalibr) and a non-motion-based reference (Plasberg et al.), our approach reduces the mean reprojection error by $44\%$ and $6\%$, respectively, while substantially simplifying the calibration procedure. Finally, we demonstrate practical utility in a robotic eye-to-hand calibration case study, showing consistent transformations and stable downstream geometric measurements even under partial occlusions. Code is publicly available at https://github.com/nhessenthaler/simple-evrgb-cal. --- ### Buried in Textual Debt: Context Pruning with Visual Evidence Preservation for MLLM Agents Authors: Yuchen Huang, Sijia Li, Jun Zhang, Yi R. Fung Published: 2026-08-24 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.22963v2 Multimodal Large Language Models (MLLMs) are increasingly deployed as multi-step agents, where explicit reasoning supports task decomposition and tool coordination but also accumulates self-generated text. Over long trajectories, this text can dominate the context and suppress visual evidence, creating textual debt. We observe that reasoning becomes redundant once task-relevant visual evidence is grounded, while stale hypotheses can misguide later inference when grounding remains uncertain. Pruning must therefore remove redundant text without discarding visual evidence. We propose SPARE, a Kullback-Leibler (KL)-guided framework for pruning accumulated reasoning in multimodal tool-use agents. SPARE uses a compact task-state summary as privileged diagnostic context. For each candidate segment, it replays the same model under the original and summary-conditioned contexts. Reverse-KL divergence from on-policy self-distillation (OPSD) then tests whether the summary sufficiently covers the segment without disrupting future reasoning. We further fine-tune the summarizer with supervised fine-tuning (SFT), enabling more compact summaries, broader coverage, and more aggressive pruning. Across multi-step visual tool-use benchmarks, SPARE achieves the highest average accuracy among pruning methods while removing 37.89-64.58\% of reasoning tokens. This favorable accuracy-context trade-off shows that reducing textual dominance restores reliance on visual evidence and mitigates over-conditioning on self-generated language. --- ### What Process Evaluation of Coding Agents Actually Measures: Action, Task, and Step Are Three Different Levels Authors: Jiawei He, Mengyu Shi, Jie jia, Xikai Yang, Dong Sun Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.22960v1 Coding agents are increasingly evaluated not only by whether they solve a task, but also by how they execute it. However, existing process-level evaluations often treat action prediction, task uncertainty, and step attribution as if they were the same problem, which makes it unclear what such evaluations actually measure. In this paper, we introduce a measurement framework for process evaluation in coding agents and instantiate step-level causal attribution with SCAE, a replay-based estimator derived from a structural causal model of agent execution. Our framework combines prefix-conditioned identification, replay/intervention-based estimation, and controlled judge-information manipulation to study process evaluation at the action, task, and step levels. Experiments on 499 file-localization episodes from 12 repositories show that next actions are driven primarily by execution provenance rather than code-graph transitions, execution uncertainty is structured at the task rather than step level, and full-trace judges exhibit systematic collider bias, suggesting that current process evaluation often measures semantic relevance rather than certified causal contribution. --- ### WildHandBench: A Benchmark for Handwritten Text Understanding that Challenges MLLMs and Humans Authors: Jun Zhang, Qiao Zhao, Cheng Cui, Jianying Qu, Zhongkai Sun, Jianwen Yang, Changda Zhou, ZhuoXin Liu, Shubin Han Published: 2026-08-24 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.22959v1 While the top model on OmniDocBench now reaches 96.34% overall on printed-document parsing, the ability of current models to handle challenging handwritten documents remains largely uncharacterized. Existing benchmarks focus on isolated text or formulas, overlook handwritten tables and real-world degradation, and report aggregate accuracy without explaining why models fail. We present WildHandBench, a benchmark containing 500 handwritten documents across three structures (free text, tables, formulas), four languages, and nine real-world scenarios. We introduce a Prior-Driven Error (PDE) metric that quantifies whether errors originate from language priors rather than visual evidence. Evaluating 18 state-of-the-art models together with calibrated human baselines, we find: (1) the best model achieves only 71.85% overall; (2) humans outperform all models yet the gap is narrow (77.09% vs. 71.85%); and (3) model errors are qualitatively different from human errors -- 63-91% of model errors are prior-driven versus only 49% for humans, exposing systematic reliance on language priors that conventional accuracy metrics cannot capture. --- ### The Illusion of Control: Why Bare Classifier Inversion Silently Fails in Concept-Bottleneck Text Generation Authors: Qi Bing, Xiaowei Shao Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.22956v1 Concept-bottleneck controllable generation routes multi-attribute control through a low-dimensional concept code that, at deployment, must be synthesised from a target attribute configuration. We study this problem in concept-bottleneck text generation under multi-axis compositional generalisation, comparing three ways to obtain the inference-time code: classifier inversion against the encoder heads, reference-text encoding, and a post-hoc label-conditioned prior. Since a concept code admits no direct LM-fluency term, regularising inversion must instead constrain the code toward the encoder's training distribution. We therefore test bare inversion and three regularised variants: label-agnostic and label-conditioned Mahalanobis penalties, and a conditional normalising-flow density baseline. Every inversion variant we test underperforms a simple post-hoc prior fitted to per-combination encoder means on the same checkpoints, across three backbone families spanning $124$M to $8$B parameters. The bare form of classifier inversion also silently collapses to chance, traceable to a directly measured off-manifold code. We validate this diagnosis on real-world benchmarks and under external evaluators, enabling fair comparison with published baselines. --- ### Stochastic gradient descent with initial regularization Authors: Nabil Kahalé Published: 2026-08-24 Categories: cs.LG, math.OC, stat.ML arXiv: https://arxiv.org/abs/2608.22953v1 We analyze a variant of stochastic gradient descent with initial regularization (SGDIR) and derive dimension-free upper bounds on its expected excess risk for the squared loss. In the noiseless case, we obtain new bounds for both averaged and non-averaged SGDIR under moment, source, and capacity assumptions. For a particular value of the source parameter, these bounds are of order $m^{-2}\log^{2}m$, where the number of training samples is of order $m$. For another value of the source parameter, we obtain, for any $ε>0$, bounds of order $m^{-3+ε}$, provided that the capacity parameter exceeds $ε^{-1}$. We also establish a lower bound that matches our upper bounds in certain regimes up to a polylogarithmic factor. In the noisy case, we provide an instance-based comparison between SGDIR and ridge regression. Under general assumptions and a mild lower bound on the regularization parameter, we show that the expected excess risk of SGDIR is no larger than that of ridge regression, up to a polylogarithmic factor. Numerical experiments on synthetic and real data are consistent with our theoretical findings. --- ### WADE: A Reasoning-Annotated Benchmark for Multi-Instance Floating-Waste Grounding with Compact Vision-Language Models Authors: Md. Asaduzzaman Shuvo, Ahsan Farabi, Md. Abdul Ahad Minhaz, Mahedi Hasan, Israt Khandaker, Ibrahim Khalil Shanto, Muhammad Nomani Kabir Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22950v1 Floating waste in inland waterways threatens aquatic ecosystems and requires timely monitoring under cluttered, multi-object conditions. Existing aquatic-waste datasets provide limited geographic coverage, sparse multi-instance annotations, and little supervision beyond boxes and labels. Compact vision-language models (VLMs) therefore remain insufficiently evaluated for jointly localizing, classifying, counting, and explaining floating waste. We introduce WADE, a reasoning-annotated benchmark containing 2,167 images from rural Bangladesh, 13,608 bounding boxes, and ten waste categories. Each annotation is associated with class-level recognition rules covering visual cues, likely confusions, and discriminative features. We evaluate six VLMs under zero-shot, two-shot, reasoning-guided, and fine-tuned settings using detection, counting, and hallucination metrics. For resource-efficient adaptation, we jointly fine-tune Qwen3-VL-2B on boxes, labels, and reasoning chains using QLoRA. Fine-tuning increases recall from 0.0248 to 0.2339 and F1 from 0.0257 to 0.2163, while reducing image-level hallucination from 0.6836 to 0.0883. However, over three-quarters of instances remain undetected, establishing WADE as a challenging benchmark for dense floating-waste grounding with compact VLMs. --- ### What Proves You Wrong: Benchmarking Language Models on Falsifiable Research Ideation Authors: Ziyue Wang, Aomufei Yuan, Yiran Yao, Linli Yao, Hongyao Zuo, Ziwen Gong, Yuanxin Liu, Shicheng Li, Yishuo Cai, Tong Yang, Xu Sun, Xiaohui Li et al. Published: 2026-08-24 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.22948v1 Large language models are increasingly used to propose research ideas, yet the prevailing ways of judging such ideas supply no shared decision rule: free-form judging sways with style and position, and scoring against a later paper rewards recovery of one realized trajectory. We introduce a benchmark that carries a proposal from Literature to Test: the Lit2Test benchmark centers on a six-field contract organized around a falsifying outcome, so that every proposal precommits the observation that would prove it wrong, making its quality decidable in the first place rather than merely arguable. Built prospectively from 200 real-paper neighborhoods, Lit2Test elicits proposals from four frontier models and compares them through 1,200 pairwise comparisons judged blind in both presentation orders. The protocol audits its own reliability through diagnostic controls and bounded human calibration, with three annotators corroborating the conclusions within explicitly stated reliability bounds. Lit2Test recovers a strict ranking of the four models in all 10,000 bootstrap replicates, and the separation comes from the quality of the proposed tests and metrics rather than from surface fluency. We release the benchmark, construction pipeline, and audit artifacts for public use. --- ### A Momentum-Based Variance-Reduced Algorithm for Federated Multiobjective Optimization Authors: Yong Zhao, Chunlin You, Minh N. Dao, Zai-Yun Peng Published: 2026-08-24 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2608.22945v1 Federated learning has traditionally been formulated as a single-objective optimization problem, primarily focused on maximizing model utility. In real-world applications, however, machine learning models often need to optimize multiple and potentially conflicting objectives simultaneously. This motivates federated multiobjective optimization (FMOO), which provides a natural framework for jointly handling multiple task-specific objectives in federated learning. In this paper, we propose a momentum-based variance-reduced algorithm for federated multiobjective optimization. The method incorporates a momentum-driven gradient estimator into the local updates to reduce the variance of stochastic updates, leading to an improved convergence rate. We establish theoretical guarantees showing that the expected Pareto stationarity measure of a randomly selected output iterate decays at a rate of $\mathcal{O}(T^{-2/3})$, improving upon the $\mathcal{O}(T^{-1/2})$ rates established for existing methods such as FSMGDA and FedCMOO. Numerical experiments on federated multiobjective optimization benchmarks demonstrate the effectiveness and competitive performance of the proposed algorithm. --- ### Quality Inspection of Printed Circuit Board Pin Insertion via Semantic Segmentation and Board-Level Feature Extraction Authors: Nils Rabeneck, André Kiunke, Nicole Hoess, Wolfgang Mauerer Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22937v1 Quality control during printed circuit board (PCB) assembly is a critical step in ensuring reliable electronic products. Detecting misaligned pins during or after pin insertion remains a particularly challenging inspection task. This paper presents an automated defect detection method for identifying incorrectly inserted pins on PCBs. The proposed pipeline combines semantic segmentation using a U-Net architecture with contour-based feature extraction and logistic regression for board-level pass/fail classification. Segmentation masks are used to derive contour representations of individual pins, from which board-level features -such as average contour size- are extracted and used to train a logistic regression classifier. We evaluate the method on two datasets: an industrial collection of real-world PCB images, and a publicly available PCB pin-inspection dataset with substantially different visual characteristics. To assess the effectiveness of the proposed approach, a comparison against PatchCore, an anomaly detection technique new to be applied to pin inspection, as well as instance segmentation-based pin detection is made. The developed method achieved Area Under the Receiver Operating Characteristic Curve (ROC-AUC) values of 0.990 on a random test set split from the industrial data and 1.000 on the public dataset indicating strong separation between pass and fail boards. The results indicate that the proposed approach is a promising candidate for automated pin inspection in industrial environments and achieves strong performance on datasets with substantially different visual characteristics after dataset-specific training. --- ### Concepts for Securing Agentic AI Coding and the Terok Environment Authors: Jiří Vyskočil, Franz Pöschel, Andreas Knüpfer Published: 2026-08-24 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2608.22930v1 Agentic AI is a fascinating new tool for software development. It is a huge step forward compared to "conventional" AI assisted coding, which in turn was a considerable breakthrough earlier. AI support through LLMs is a young and very fast-moving field. The "conventional" (non-agentic) flavor became useful and productive in early 2025 (around 18 months ago) and the agentic flavor followed in fall 2025 (approximately 9 months ago). Besides all its benefits and potential, it also carries some fundamental risks for IT security. And the agentic approach added very severe risks while making others much more dangerous. With all the motivation to explore this fascinating new tool we should not ignore the risks but actively address them. We present (I) an assessment of the IT security risks, (II) a concept for mitigating them without breaking its benefits, and (III) an overview about an implementation of our concept. In this very dynamic field this is likely not the final and once-and-for-all answer to the identified issues but still a substantial step forward in responsible usage of Agentic AI for software development. It should also be a contribution to the community to allow early and eager evaluation of the potential of agentic AI for software development without actually suffering from its implied IT security risks. --- ### Motion-Based Tokenization for Cross-Dataset Egocentric Gaze Modeling Authors: Virmarie Maquiling, Zhuojiang Cai, Enkelejda Kasneci Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22926v1 Gaze is increasingly used as an input signal for vision and multimodal models, yet no consensus exists on how to represent it across datasets. Raw traces preserve detail but are noisy and device-dependent, while coarse event labels are easy to model but can discard local motion structure. We formulate event-aligned, fixed-horizon angular displacement as an interpretable, event-conditioned motion vocabulary and compare it with event-only, spatial, absolute-angle, learned vector-quantized, and continuous representations. To assess transfer alongside target predictability and token collapse, our evaluation combines next-token prediction with target-domain regret, low-order target references, paired bootstrap, order sensitivity, motif overlap, and frozen structural probes. In an event-aligned headset benchmark, angular-motion tokens have lower target-domain regret than frozen-codebook VQ tokens in one transfer direction, while the reverse direction is inconclusive. The probes reveal complementary representation properties, and event-only tokens show that low perplexity can retain little motion information. On a third egocentric dataset, a matched comparison of I-VT, native, and frame-span interfaces shows that event construction materially changes transfer: native events have the lowest regret into EGTEA, while frame-span events have zero motif overlap and fail severely as a source. Motion-based tokenization therefore provides a compact representation for event-aligned egocentric gaze streams, while the evaluation identifies how target predictability and event construction shape cross-dataset conclusions. --- ### Deep Learning-Based Multi-User Communication Design for Dense IoT Networks: Interference-Aware Finite-Blocklength Communication and Preliminary MIMO Extensions Authors: Arkadeep Sinha, Shubham Paul, R. Manivasakan Published: 2026-08-24 Categories: cs.IT, cs.AI arXiv: https://arxiv.org/abs/2608.22923v2 Dense IoT networks require reliable communication despite limited spectrum and substantial multi-user interference while maintaining manageable receiver complexity. This work introduces a deep-learning-based end-to-end multi-user communication design for interference-limited finite-blocklength IoT scenarios, focusing on short and medium blocklengths. We extend a prior 2-user SiameseNet transceiver framework to accommodate 2, 4, and 8 users, leveraging learned redundancy for interference suppression and noise robustness. Compared to conventional non-orthogonal access baselines, our method demonstrates strong Block Error Rate (BLER) performance across various scenarios without resorting to joint detection; the per-user decoder scales roughly linearly with the number of users. Further, we examine the robustness under interference mismatch and unequal interference strengths, critical for practical deployments with heterogeneous devices. The Latent-space analysis reveals that the learned codeword distance increases as the effective per-user rate decreases, corroborating with the observed BLER improvements. In addition, we also present preliminary results for a 2X2 MIMO setup under fixed-channel CSIT and CSIR, indicating potential for extending the framework to IoT gateways with multiple antennas. --- ### HelaBERT: Enhancing Sinhala Language Understanding with Dual Pooling Classification Head Authors: Thisen Ekanayake, Nisansa de Silva Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.22922v1 We present HelaBERT, a family of two BERT-based masked language models pre-trained from scratch on approximately 1 billion tokens of Sinhala text sourced from MADLAD-400, CulturaX, and a custom corpus comprising news articles, Sinhala Wikipedia, and web crawl data. HelaBERT-Small (~23.3M parameters, 6 layers) and HelaBERT-Large (~110M parameters, 12 layers) both use a SentencePiece Unigram tokenizer (vocabulary size 32,000) tailored to Sinhala's agglutinative morphology and complex script. We evaluate both models on four downstream Sinhala text classification tasks: news category classification, news source classification, sentiment analysis, and writing style classification, using 5 independent seed runs with stratified 80/20 train/test splits. We additionally propose a dual pooling classification head and evaluate it systematically across all four tasks, finding consistent improvements on sentiment analysis and a moderate gain on news category classification for HelaBERT-Small, while the standard (CLS)-linear head remains competitive on news source classification, a headline-level task with short average input length. We release both models to support further research in Sinhala NLP. --- ### Beyond Observed Auxiliary Relations: Environment-Conditioned Modeling for Multi-Behavior Recommendation Authors: Seunghan Lee, Hyunsik Yoo, Jian Kang, Susik Yoon, SeongKu Kang Published: 2026-08-24 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.22920v1 Multi-behavior recommendation (MBR) leverages auxiliary behavioral signals, such as clicks and add-to-cart, to enhance target behavior prediction like purchases. While recent graph neural network-based approaches have achieved strong performance by systematically propagating auxiliary behavior signals, they still suffer from two fundamental challenges inherent to auxiliary behaviors: (1) missing auxiliary signals, which hinder generalization to items without auxiliary observations, and (2) unreliable auxiliary signals, which amplify noise misaligned with the target behavior. To address these challenges in a unified manner, we propose BOAR, an environment-conditioned MBR framework that addresses missing and unreliable auxiliary signals through two complementary modules conditioned on auxiliary observability. Extensive experiments demonstrate that BOAR consistently outperforms state-of-the-art baselines, achieving up to 7.82% gains in HR@10 overall and up to 44.2% gains for target items without auxiliary observations, highlighting its ability to capture hidden preferences beyond observed auxiliary relations. Our code is available at: https://github.com/LSH0411/BOAR. --- ### TSWAP: A Multilingual Retrieval-Augmented Thai Wellness Advisor Authors: Pornthep Ukosaramig, Kobkrit Viriyayudhakorn Published: 2026-08-24 Categories: cs.CL, cs.IR arXiv: https://arxiv.org/abs/2608.22917v1 We present TSWAP, a deployed eight-language conversational wellness advisor grounded, via retrieval-augmented generation, in a verified knowledge base of Thai traditional medicine and certified wellness providers. An unmodified open-weight LLM (Qwen3.6-35B-A3B on vLLM) is grounded on a ~30.6K-chunk Thai index by a hybrid dense-sparse retriever with cross-encoder reranking; a first-turn query classifier forces tool-based retrieval for entity lookups; a rule-based safety layer enforces medical scope and Thai emergency routing; and all eight languages are served zero-shot with translate-then-retrieve. We release the first Thai traditional-medicine/wellness retrieval benchmark (50 questions with gold document IDs; Recall@5 = 0.88), production QA logs (91.1% test-retest pass over 259 cases), and a 71-question frontier no-retrieval probe showing what each grounding pillar contributes: without the safety prompt the backend model family produced a full drug-dosing schedule and complied with out-of-scope requests, and without the knowledge base it produced zero verifiable provider recommendations. We further report two transferable deployment findings: English-calibrated 4-bit AWQ quantization corrupts Thai tone marks, and forced-retrieval routing is necessary for reliable grounding. --- ### Knowing Isn't Always Saying: When Do Spatial Encodings Reach Answers in Vision-Language Models? Authors: Zeyu Wang, Xinming Xu Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.22916v1 Vision-language models are known to encode spatial information in their hidden states, yet often fail to use it when answering. However, it remains unclear when and where this encoded information reaches the answer. We address this with direction patching, a class-conditioned causal intervention applied across layers, token positions, and prompt formats. Using spatial-ID directions constructed following prior encoding evidence, we find that causal influence on answer logits emerges only at mid-to-deep depths. Text chain-of-thought suppresses immediate object-word argmax-level transport in most models, while visually grounded prompts keep it open. Positive target-logit gain can remain below the argmax threshold, and transport can re-emerge at the final prefix token or at the answer step in deeper layers. Across the ten VLMs we study, these local effects form descriptive transport patterns. Complementary experiments characterize how these patterns shift across datasets, attributes, and encoding amplitudes. Together, these results reframe the encoding-grounding gap as a problem of conditional transport in VLMs. --- ### Safety Hacking in Constrained Best-of-$N$ Inference-time Scaling Authors: Akifumi Wachi, Takumi Tanabe, Youhei Akimoto Published: 2026-08-24 Categories: cs.LG, cs.AI, cs.CL, cs.CR arXiv: https://arxiv.org/abs/2608.22915v1 Inference-time pipelines often sample multiple outputs, filter them with a learned safety model, and return the proxy-feasible output with the highest learned reward. We show that this composition creates a two-stage failure: an imperfect safety proxy first contaminates the feasible set with unsafe outputs, and reward maximization can then amplify this residual contamination. We define \emph{safety hacking} as selecting an output that passes the learned constraint but violates the true safety criterion. For constrained Best-of-$N$ sampling, we derive finite-$N$ bounds governed by the joint upper reward tails of safe and unsafe outputs within the proxy-feasible set. If unsafe-but-feasible outputs have the heavier tail, safety hacking becomes asymptotically certain as $N$ grows, even when false-positive mass and average safety- and reward-proxy errors are arbitrarily small. We also show that policies within a bounded $χ^2$ divergence from the proxy-feasible reference distribution admit an $N$-independent safety-hacking bound, and instantiate this general coverage-control principle with constrained pessimistic sampling. Coverage control limits amplification but cannot repair a contaminated feasible set: admitted unsafe outputs may still be favored, and regularized selection is not necessarily safer than constrained Best-of-$N$ for every reward proxy. Toy and language-model experiments characterize both contamination and its reward-tail amplification, which exposes an inherent difficulty in inference-time scaling with learned safety models. --- ### Results of the 1st Asynchronous CASTLE Challenge at the Joint Egocentric Vision Workshop in Conjunction with CVPR 2026 Authors: Luca Rossetto, Werner Bailer, Cathal Gurrin, Graham Healy, Omar Shahbaz Khan, Stevan Rudinac, Klaus Schöffmann, Allie Tran Published: 2026-08-24 Categories: cs.CV, cs.MM arXiv: https://arxiv.org/abs/2608.22914v1 This report summarizes the contributions and results of the 1st Asynchronous CASTLE Challenge at the Joint Egocentric Vision Workshop in conjunction with CVPR 2026. --- ### A Commutator Framework for Selective Spectral Alignment in Deep Neural Networks Authors: Kaj Nyström Published: 2026-08-24 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2608.22910v1 We develop a finite-width geometric framework describing how learned feature geometries are organized, transported, and selectively aligned in deep neural networks. Incompatibility among weight-generated covariance, gates, and backward sensitivities is quantified through three families of commutators: between gates and covariance, between sensitivities and covariance, and between average gradient outer products (AGOPs) and neural feature matrices (NFMs). An exact layerwise identity decomposes the sensitivity-covariance commutator into four sources: downstream transport, adjacent-layer imbalance, pointwise sensitivity fluctuations, and nonlinear gate-covariance interactions. The AGOP-NFM commutator is a singular-value-weighted transport of the internal commutator, explaining why observed feature-side alignment alone does not determine the internal geometry from which it emerges. Buffered localized energies resolve mixing between separated covariance subspaces. We establish spectral-gap, projector-evolution, and stabilization estimates, and formulate conditional Lyapunov principles that yield decay under explicit geometric error-bound or intrinsic-damping assumptions. These criteria do not follow from gradient flow alone and clarify why risk reduction need not imply commutator collapse. Analytic examples and numerical experiments exhibit factorization of spectral and activation geometry, transient growth, and cancellation among nonzero sources. In tested finite-time regimes, cancellation dominated by a negative transport-imbalance interaction persists across depths, widths, and two regression benchmarks. Spectral alignment therefore appears as a layer- and scale-dependent compatibility phenomenon governed by transport, interaction, cancellation, and possible damping, rather than a universal consequence of training. --- ### Exploring Dowker Homology for Sentence Similarity Authors: Marius Huber, Juri Opitz Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.22909v1 Dowker homology is a topological tool that may be used to analyze the relative position of two point clouds living in a common space. We investigate whether Dowker homology captures sentence similarity information by treating the embeddings of the tokens that constitute a sentence pair as a pair of point clouds in the latent space of a transformer model, using both models that have and have not been fine-tuned for sentence similarity. We find that Dowker homology captures sentence similarity information, as measured by regressing Dowker homology features onto ground-truth similarity scores, and that it can be used for visual inspection of similarity data and models. In an attempt to make Dowker homology readily applicable, we derive from it single-number summaries that we expect to capture sentence similarity directly. These turn out to work reasonably well, but without outperforming standard sentence similarity measures based on established pooling methods. --- ### Do Spoken Language Models Hear Speech as They Read Text? Bridging Structural Gaps Between Speech and Text Authors: Hyeonyu Kim, Hwayeon Kim, Youngwon Choi, Myeongkyun Cho, Huu-Kim Nguyen Published: 2026-08-24 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.22908v1 Spoken Language Models (SLMs) generate textual responses directly from speech, offering an alternative to cascaded systems. Despite recent advances, existing SLMs still exhibit weaker instruction-following behavior and limited generalization across diverse tasks compared to text-based language models. Our analysis shows that speech and text representations in current SLMs remain weakly aligned despite strong downstream performance, indicating that structural differences between continuous, temporally varying speech and discrete text remain insufficiently addressed. To address this, we propose a simple framework that decouples length mismatch from semantic alignment and encourages closer correspondence between speech and text representations. Experiments across multiple benchmarks demonstrate competitive performance against strong baselines, underscoring the importance of explicitly addressing structural differences between speech and text in SLM training. Our code is publicly available at https://github.com/jaykim9870/Do_SLMs_Hear_Speech_as_They_Read_Text. --- ### AquaFlow: A Monocular Gaussian Splatting SLAM for Underwater Streaming Reconstruction Authors: Yingxiang Xu, Kerui Ren, Wenqi Guo, Changjian Jiang, Tao Lu, Linning Xu, Mulin Yu Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22906v1 Recent monocular 3D Gaussian Splatting (3DGS) streaming reconstruction methods have achieved impressive performance by balancing reconstruction quality and efficiency. However, extending these frameworks to underwater scenes remains challenging due to severe visual degradation, such as light attenuation and scattering, which degrades camera pose tracking and distorts scene geometry. To address these challenges, we propose AquaFlow, a monocular Gaussian Splatting streaming reconstruction framework for efficient and high-fidelity underwater reconstruction. Specifically, AquaFlow fine-tunes a 3D vision foundation model on large-scale underwater data for robust pose and pointmap estimation, and introduces a medium-guided incremental Gaussian initialization strategy for streaming mapping. Furthermore, we develop a streaming-compatible hybrid scene representation that integrates structured, distance-conditioned neural Gaussians with a physics-inspired optical model to compensate for underwater image formation effects, enabling accurate scene reconstruction. We evaluate AquaFlow on a comprehensive dataset of 62 diverse underwater trajectories, collected from both public benchmarks and in-the-wild web videos across various scales. Extensive experiments demonstrate that AquaFlow achieves state-of-the-art tracking and rendering performance, reducing average localization error by 13.2% and improving PSNR by 4.74 dB compared to WaterSplat-SLAM. --- ### CDEG: Learning Decision-Critical Evidence for Long-Horizon Diagnostic Agents Authors: Xiwei Dai, Zijie Meng, Zhiting Fan, Yixuan Tang, Guanyu Jiang, Ziru Niu, Zuozhu Liu Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.22899v2 Unlike static medical question answering, long-horizon diagnosis captures the sequential nature of clinical practice: evidence is progressively acquired, integrated, and evaluated over multiple rounds of interaction before reaching a final diagnosis. However, existing doctor agents often fail when critical evidence is either not acquired or not adequately incorporated into diagnostic reasoning. Recent agentic approaches attempt to address these failures by reusing historical trajectories or distilled memories. But their diagnostic gains remain constrained because such experience may contain noisy or incidental information and is typically reused without validating which evidence actually drives diagnostic decisions. To address this limitation, we introduce CDEG, a graph-based framework that learns reusable decision-critical evidence from historical diagnostic trajectories. CDEG contrasts successful and failed trajectories from the same case to identify candidate evidence, validates their diagnostic impact through controlled counterfactual interventions, and organizes the resulting diagnosis--evidence--action relations into a structured graph. During inference, CDEG tracks the evolving patient evidence state to retrieve relevant diagnostic relations and selectively guide missing evidence acquisition or overlooked evidence reappraisal. Across in-domain and out-of-distribution benchmarks with multiple doctor agent backbones, CDEG consistently improves diagnostic performance, achieving up to an 11.5% accuracy gain over vanilla agents. These results demonstrate that reliable long-horizon diagnosis requires moving beyond trajectory-level experience reuse toward evidence-level learning of the factors that truly shape clinical decisions. --- ### SelFusion: Self-distillation for Diffusion Language Models Authors: Hyeongsoo Lim, Jinyoung Kim, Eunseo Seo, Minho Jang, Jiwon Yoon Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.22898v1 Diffusion language models (DLMs) alleviate the inherent latency bottleneck of autoregressive (AR) large language models (LLMs), but their degraded generation quality limits practical applicability. Although knowledge distillation (KD) can be a promising direction for improving performance, we empirically find that naively applying conventional KD yields only marginal gains, or even degrades generation quality. Based on these observations, we propose a novel self-distillation framework for DLMs, namely SelFusion. To enable effective KD without an external teacher model, SelFusion performs two forward passes with different masking levels, defining the hard mode with a larger masking probability and the easy mode with a smaller masking probability. However, the easy mode is not always more accurate than the hard mode and can be overconfident on incorrect tokens. Thus, we introduce bidirectional KD between the two modes, which can dynamically determine the distillation direction based on token-level correctness. Experimental results on instruction-following tasks show that the proposed self-distillation substantially outperforms other KD methods with external LLM and DLM teachers. In many configurations, the student trained with SelFusion even surpasses the performance of the LLM teacher, providing a practical path toward improving DLM generation quality. Source code can be found at https://github.com/scai-research/SelFusion_official --- ### AraDetox: A Multi-Dialect Arabic Detoxification Dataset Authors: Mo El-Haj Published: 2026-08-24 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.22894v1 Arabic harmful-language detection has received considerable attention, yet Arabic text detoxification remains underexplored. We introduce AraDetox, a multi-dialect Arabic detoxification dataset comprising 10,500 harmful social-media posts and 84,000 detoxified rewrites generated using GPT-5 and Gemini 2.5 Flash across Modern Standard Arabic, Gulf, Levantine, and Egyptian Arabic. The generated outputs were assessed through human evaluation and automatic analyses of lexical change, semantic preservation, sentiment, and dialectal style. Results show that detoxification is primarily a meaning-preserving rewriting task: substantial lexical and structural reformulation is accompanied by consistently high semantic similarity. Human evaluation confirms successful harmful-language removal while largely preserving the original meaning. Dialectal analyses further indicate that the generated variants exhibit measurable stylistic alignment with reference Arabic dialect corpora. Comparison with existing resources highlights two complementary approaches to detoxification: minimal-edit lexical substitution and meaning-preserving reformulation. Our findings demonstrate that large-scale Arabic detoxification resources can be constructed through LLM-assisted generation and human verification. The dataset is publicly available at https://github.com/ArabicNLP-UK/AraDetox to support future research on Arabic detoxification, safe text generation, and multi-dialect Arabic NLP. --- ### NemoSplat: Feed-Forward 4D Gaussian Splatting for Media-Aware Underwater Reconstruction Authors: Xiaopeng Guo, Wai Chung Tse, Yipeng Zhu, Hanwen Zhang, Huajian Huang, Sai-Kit Yeung Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22888v2 Reconstructing photorealistic scenes in unconstrained underwater environments remains challenging due to severe media-induced light scattering and unpredictable dynamic objects. Recent feed-forward visual foundation models have demonstrated remarkable capabilities in generalized novel view synthesis and tracking. However, when directly applied to aquatic videos, optical attenuation and motion interference fatally corrupt their feature aggregation, leading to severe tracking and reconstruction failures. To overcome these limitations, we present NemoSplat, the first feed-forward 4D Gaussian Splatting framework tailored for media-aware dynamic reconstruction directly from uncalibrated marine videos. Beyond providing robust estimations of camera poses and dense scene depth, we devise a Promptable Dynamic Disentangler that utilizes a confidence-aware fusion strategy of learned dynamic probabilities and optional semantic text priors, effectively isolating massive transient entities. Furthermore, to counteract visual degradation, a Media-Aware Gaussian Predictor is formulated to jointly estimate intrinsic 3D Gaussian attributes alongside physical media parameters, rendering pristine scene appearance in a single forward pass. Additionally, we introduce a large-scale underwater dataset with massive dynamic elements to facilitate training and evaluation. Extensive experiments on our dataset demonstrate that NemoSplat achieves state-of-the-art tracking accuracy and high-fidelity rendering. Homepage: https://nemosplat.hkustvgd.com --- ### Proxy reliance in large language model decisions is uncalibrated to predictive evidence Authors: Zengqing Wu, Chuan Xiao Published: 2026-08-24 Categories: cs.AI, cs.CL, cs.CY arXiv: https://arxiv.org/abs/2608.22887v1 Large language models (LLMs) are entering decisions in triage and lending, where task-relevant inference must be distinguished from impermissible proxy use. Current audits ask whether decisions change when demographics change. But attributes correlated with a protected group carry predictive value, so a changed decision can be discrimination or sound inference. We measure causal proxy effects in four LLMs on a clinical-ranking task with known ground truth, where the reliance the evidence warrants can be computed exactly and used as the reference. One audit signal yields three verdicts: over-reliance, warranted and under-reliance. Under neutral labels every model relies on proxies with no information. Informative proxies draw all three. Social field names push reliance down, below the reference in one model. Two findings explain this. Reliance severely undertracks the evidence, and social-label suppression is fragile, since in-context examples raise it above zero in every model. Accuracy-based evaluation detects none of this. --- ### DRAgent: Discriminative Reasoning Agent for Referring Expression Segmentation Authors: Yujie Qi, Luyan Zhang Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22885v1 Referring Expression Segmentation (RES) aims to generate a pixel-level mask for the object specified by a language expression. Recent methods based on multimodal large language models (MLLMs) often rely on one-pass coordinate prediction for visual localization, which serializes continuous spatial locations as discrete text tokens and may lead to localization bias and alignment errors. To address these issues, we propose DRAgent, an MLLM-driven discriminative reasoning (DR) framework for RES. Instead of requiring the MLLM to generate localization coordinates, DRAgent first constructs a detector-generated candidate space and then uses the MLLM as a visual-semantic target discriminator. Specifically, the MLLM performs reliable target selection among potential distractors through a two-stage DR mechanism, which first screens high-recall candidates and then performs instance-wise verification. The selected target box is subsequently used as a spatial prompt for a foundation segmentation model to produce the final pixel-level mask. Furthermore, we construct a self-consistency-filtered reasoning-chain data pipeline for LoRA-based fine-tuning, providing more reliable supervision for enhancing the MLLM's discriminative reasoning capability. Experiments demonstrate that DRAgent achieves competitive performance on RefCOCO, RefCOCO+, and RefCOCOg. --- ### FOVEA: Focused On-Demand Visual Evidence Adaptation for Cache-Friendly Multimodal Speculative Decoding Authors: Hengjie Zhu, Dayan Wu, Zihao Zhang, Xinze Liu, Jingxuan Yu, Peng Fu, Zheng Lin, Weiping Wang, Ding Wang Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22883v1 Multimodal speculative decoding accelerates vision-language models by allowing a lightweight draft model to propose candidate tokens for parallel verification by a larger target model. Existing methods typically condition the drafter on a fixed visual interface, such as a predefined visual-token budget or a static compressed representation. However, our controlled visual-budget analysis shows that visual demand varies substantially across tasks and decoding stages, which means more visual input is not always beneficial. Actually, insufficient evidence may weaken visual grounding, while excessive context adds overhead and may disrupt drafting. We propose FOVEA (Focused On-demand Visual Evidence Adaptation), a cache-friendly approach that builds a reusable visual memory and dynamically retrieves a bounded subset for a draft state. A cumulative-mass rule determines both how many and which entries are selected. The selected entries are aggregated into a visual readout and fused with the current draft hidden state through a lightweight gated residual correction. Rather than inserting visual tokens into the autoregressive context, the correction modifies only the representation passed to the language-model head. Experiments across multiple vision-language backbones and multimodal benchmarks show that FOVEA improves draft acceptance and end-to-end decoding speed, achieving up to $2.13\times$ speedup over autoregressive decoding. These results demonstrate that state-conditioned evidence retrieval is an effective alternative to reusing a fixed visual representation throughout multimodal generation. --- ### Large-Small Model Collaboration for Zero-Shot Surgical Phase Recognition Authors: Yiyi Zhang, Ying Zheng, Wenxin Fan, Yu Zhu, Yuchen Yuan, Litao Zhao, Zheng Li, Pheng-Ann Heng Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.22879v1 Task-specific lightweight models for surgical phase recognition excel at capturing temporal dynamics but generalize poorly under domain shift. Conversely, surgical foundation models (FMs) offer superior transferability via large-scale pretraining, yet their lack of explicit temporal modeling often yields temporally inconsistent predictions, leading to degraded performance. To exploit the complementary strengths of both paradigms, we propose \textbf{La}rge-\textbf{S}mall \textbf{T}emporal adaptation (\textbf{LaST}), a novel large-small collaborative framework that enables zero-shot adaptation to unseen clinical domains. In LaST, the FM initiates the pipeline by generating frame-level phase priors that serve as initial weak supervision. To effectively utilize these noisy phase priors, we introduce an iterative temporal refinement scheme that integrates dynamic quality control to filter reliable predictions and dual-model cross-learning to mitigate confirmation bias. Simultaneously, the lightweight model leverages its intrinsic temporal modeling ability to progressively correct inconsistent predictions and enhance overall accuracy across iterations. At the end, a cycle replay strategy is employed to close the loop: the refined, more accurate predictions are utilized as upgraded supervision signals for the subsequent iterations, fostering a self-reinforcing evolution of both label quality and model capability. Extensive experiments demonstrate that LaST achieves robust adaptation to unseen domains for zero-shot surgical phase recognition, outperforming the baseline (PeskaVLP) by 24.85\%-43.17\% in accuracy and even surpassing fully supervised linear probing and several state-of-the-art few-shot approaches. Codes will be released at https://github.com/YIYIZH/LaST. --- ### The Mask Is Not the Model: Auditing Prefix Invariance in Attention, State-Space, and Hybrid Sequence Models Authors: Taebong Kim, Youngsik Hong, Minsik Kim, Sunyoung Choi, Jaewon Jang, Minseo Kim Published: 2026-08-24 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.22876v3 Hybrid sequence models must satisfy prefix invariance: representations at position t must not depend on future inputs, yet this is rarely verified. We formalize prefix invariance and give a lightweight audit, two forward passes, no training or gradients, yielding a per-layer score localizing where causality breaks. Attention-mask inspection, the field's default check, is incomplete: causality is a graph-level property, and leaks can occur via scans, aggregations, or normalization despite correct masks. Across 192 injected-fault trials on eight checkpoints, mask inspection detected none, while our audit localized all 192/192 to the exact layer. Static/dynamic analysis of chunked-scan code in transformers found the same defect in Zamba2 and Nemotron-H, an inter-chunk axis error fixed via the reference implementation. The method fits on one page and runs in seconds. --- ### Stochastic Separability of Embedding Manifolds Authors: Liqing Zhang Published: 2026-08-24 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2608.22874v1 Neurobiological studies and representation learning have observed that representations of objects belonging to the same category in high-dimensional neural spaces exhibit low-dimensional object manifold characteristics, and different object manifolds are linearly separable in these neural spaces. However, these experimentally observed phenomena lack rigorous theoretical validation to date. This paper proposes a new stochastic separability theorem for embedding manifolds of two different object categories. First, we establish a projection measure concentration theorem for embedding manifolds under general conditions. We develop a new two-layer measure concentration analysis technique, which unifies two estimation bounds via the law of total expectation to derive measure concentration inequalities. Based on the measure concentration theorem, we further prove a stochastic separability theorem for embedding manifolds of two different object categories. If two datasets have distinct means and bounded total variances, their samples become linearly separable with high probability, provided that the projection direction satisfies a non-singularity condition. The main contributions of this paper are twofold: 1. We prove the projection concentration properties of embedding manifolds in high-dimensional spaces by using two-lawyer tail-bound inequalities. 2. We identify a non-singularity condition for the stochastic separability between embedding manifolds, and rigorously prove the stochastic projection separability theorem. The theorem not only uncovers geometric and statistical properties of the object embedding manifolds, but also provides a novel mechanism for representation learning in deep networks. --- ### Better Retrieval, Worse Robustness: How Multi-hop RAG Amplifies Upstream ASR Errors Authors: Zhenghua Bao Published: 2026-08-24 Categories: cs.CL, cs.IR, eess.AS arXiv: https://arxiv.org/abs/2608.22872v2 Speech-based applications pass spoken queries through automatic speech recognition (ASR) before any retrieval module, so ASR errors enter the pipeline as a fixed upstream constraint. We empirically test whether two extensions to standard retrieval-augmented generation (RAG), entity-graph linking and iterative reformulation, absorb or amplify these errors. Using four English accents synthesized through neural TTS, we evaluate four RAG configurations on three multi-hop QA benchmarks (HotpotQA, 2WikiMultiHopQA and MuSiQue) against a clean-text oracle. Although the structurally richer configurations generally retain higher absolute F1 under ASR input, both extensions amplify the error: the F1 gap from clean text to the highest-WER accent is 36-67% larger under their combination than under naive dense retrieval, on all three benchmarks. The dominant failure mode is corruption of one or more query entities, accounting for 87-96% of degradation cases on 2WikiMultiHopQA across all four methods. Two lightweight surface-form mitigations leave most of the gap intact, indicating that downstream retrieval structure amplifies remaining entity errors. We release code and data at https://github.com/Continuum-AI-Corp/spoken-multihop-rag . --- ### UniMem: Unifying Multimodal Memory and Control for Vision-Language-Action Models Authors: Lars Osterberg, Maggie Wang, Mac Schwager Published: 2026-08-24 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2608.22869v1 While Vision-Language-Action (VLA) models have leveraged internet-scale pretraining and task-focused finetuning to achieve strong performance on long-horizon tasks, they often struggle with non-Markovian tasks that require memory. Existing approaches to memory typically involve additional Vision-Language-Models (VLMs) for long-term memory management, introducing a memory bottleneck and a fractured training pipeline. Conditioning on multiple historical frames can provide the VLA with access to more descriptive features of past scenes, but can degrade performance if frames are chosen at arbitrary, fixed intervals. To address these limitations, we present UniMem, a framework that unifies high-level, multimodal memory and low-level control under one backbone. UniMem employs an event classifier for memory updates, a keyframe encoder for dense spatial memory, and a keyframe caching technique to minimize overhead during policy rollouts. We evaluate UniMem across five simulation and four hardware tasks targeting sequential and spatial memory, demonstrating that our unified, single-model system outperforms fixed-interval image sampling baselines (93.4% vs. 68.2%) in simulation and hierarchical baselines (80.0% vs. 43.5%) in hardware, while offering faster inference and a simple training pipeline for easy adoption. Project website: https://losterberg3.github.io/unimem-vla/ ---