# Research Papers — 2026-08 · bundle page 115 of 240 > Full-text markdown of papers #5701–5750 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/115 [← Previous bundle](https://aipod.dev/papers/archive/2026-08/page/114.md) [Next bundle →](https://aipod.dev/papers/archive/2026-08/page/116.md) ### Retrieval-guided Twin Fusion with Similarity-aware Contrast for Molecule-Text Alignment Authors: Shunshun Gu, Shengqi Qiu, Hang Zhou, Xiao Luo Published: 2026-08-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.16005v1 This paper studies the problem of molecule-text alignment, which aims to project molecules and their textual descriptions into a joint latent space for downstream tasks including molecule search and molecular property prediction. Previous approaches typically combine graph structure mining with contrastive learning to enhance joint representation learning. However, they typically neglect fine-grained semantic relationships between substructures and texts, leading to suboptimal performance on downstream tasks. Towards this end, we propose a novel approach named Retrieval-guided Twin Fusion with Similarity-aware Contrast (RISEN) for molecule-text alignment. The core idea of RISEN is to construct a latent twin molecule for each substructure with cross-modal retrieval for semantic enhancement. In particular, for each substructure query, we retrieve relevant textual descriptions and sample several molecules that share similar descriptions of substructures. Then, we aggregate their representations via attention pooling for a twin latent representation, which would be further fused with the original substructure for representation enrichment. In addition, we measure the similarity across substructures and texts, which would further guide cross-modal contrastive learning with soft thresholding. Extensive experiments on benchmark datasets validate the superiority of the proposed RISEN in comparison with existing baselines. --- ### Prior Audit-Repair Context Shifts LLM Verifier Thresholds Toward Leniency Authors: Parsa Mazaheri, Kasra Mazaheri Published: 2026-08-17 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.16003v1 Automated checking pipelines increasingly place one language model as the checker and another (or the same one) as the fixer. We ask whether that wiring changes what the checker reports. Measuring false alarms on human-verified-correct ProcessBench traces with the present task held byte-identical, we find that a completed audit -> repair episode already in the model's context lowers false alarms in 15 of 15 model x wording combinations, by 2.8 to 11.5 percentage points against a length-matched non-audit control, a 9 to 25% reduction relative to that control. The direction contradicts what the accumulated-message literature predicts: an episode whose audit reported an error lowers false alarms further still, at all five wordings on the model where that manipulation lands cleanly, though a negativity asymmetry predicts more flagging. Decomposing the episode finds repair content and audit verdict complementary: different components carry the effect on different model families. Signal-detection analysis locates the change in the threshold rather than in discrimination -- the criterion moves in 15 of 15 combinations and survives correction in 13 while d' survives in none, though the d' test is half as sensitive by construction -- and a hand audit of 50 false alarms finds 82% simply wrong, so at this operating point the shift need not be harmful. With reasoning enabled the effect keeps its relative size on both models tested, and the threshold reading holds there too. --- ### From Sequence to Structure: Relational Uncertainty Propagation for LLM Agents Authors: Zhengzhao Ma, Boxi Cao, Yaojie Lu, Hongyu Lin, Xianpei Han, Le Sun Published: 2026-08-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.16002v2 Reliable uncertainty quantification (UQ) is essential for deploying large language model (LLM) agents in complex interactive environments. Existing UQ methods largely rely on local signals, such as token probabilities, predictive entropy, or per-step confidence, and therefore overlook the long-range dependencies through which errors accumulate across an execution trajectory. As a result, they may fail to identify agent failures whose causes originate several reasoning or interaction steps before the final answer. We propose RUPA (Relational Uncertainty Propagation for Agents), a trajectory-level UQ framework for LLM agents. RUPA represents an execution history as a directed trajectory graph in which reasoning states, tool interactions, and environment feedback are nodes connected by temporal and semantic dependency edges. It then propagates uncertainty over this graph to capture how execution risk accumulates and transfers across interaction steps. The propagated signal is combined with trajectory-level behavioral features and goal-alignment information to produce a confidence estimate for the full agent trajectory. We evaluate RUPA on representative agent benchmarks, including $τ$-2, Terminal-Bench-2, and GAIA, using 6 open-source LLMs spanning multiple model families. Experimental results show that RUPA consistently outperforms existing UQ methods by providing more accurate uncertainty estimates, enabling earlier failure detection, and improving uncertainty-guided agent execution across diverse agent tasks. These results demonstrate that explicitly modeling relational dependency is crucial to reliable UQ for long-horizon LLM agents, providing a practical foundation for trustworthy agent execution. --- ### MUPA$^{2}$E: Multimodal Unified Perception with Asymmetric Attention for Emotion Assessment Authors: Stefanos Gkikas, Eric Nichols, Christian Arzate Cruz, Randy Gomez Published: 2026-08-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.15999v1 Automatic emotion assessment can benefit from combining neural and behavioral signals, but many multimodal approaches rely on separate, modality-specific feature-extraction pipelines before fusion. This paper presents MUPA\textsuperscript{2}E, a unified perception framework that processes facial video and electroencephalography (EEG) through a single shared asymmetric-attention backbone. Facial video is represented through axis-folded frame tokens, while EEG is processed either as a raw multichannel waveform or projected into the spatial domain for multimodal fusion. The framework is evaluated on the DMER dataset under a stratified subject-independent protocol, comparing unimodal video, unimodal EEG, and fused video--EEG configurations with per-channel and merged EEG projections. Using the original recordings, with shorter trials zero-padded to match the longest duration, merged fusion at stride~$30$ achieves the highest validation performance and a test accuracy of $70.07\%$. Further analysis revealed that recording duration is unevenly distributed across the affective classes, making the padding pattern a potential classification cue. Controlling for this factor by cropping all recordings to a common duration of $20$ seconds yielded a test accuracy of $62.71\%$, providing a stricter duration-controlled assessment of the framework in which differences in recording length are removed as a potential classification cue. These findings demonstrate the feasibility of processing structurally different neural and visual signals within a compact unified architecture while highlighting the importance of controlling duration-related cues in affective datasets. --- ### Toward Optimal Second-Order Path-Length Guarantee for Adversarial Multi-Armed Bandits Authors: Mengxiao Zhang Published: 2026-08-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.15996v1 We study second-order path-length regret in adversarial $K$-armed bandits against oblivious loss sequences. Bubeck et al. (2019) designed an algorithm that achieves $\widetilde{\mathcal{O}}(K+\sqrt{KQ_{\infty,1}})$ regret, where $Q_{\infty,1}$ is the first-order path length, and left open whether $\widetilde{\mathcal{O}}(\text{poly}(K)\sqrt{1+Q_{\infty,2}})$ regret is achievable under bandit feedback, where $Q_{\infty,2}$ is the second-order path length. Somewhat surprisingly, we resolve this question positively by showing that with a more involved analysis, the exact same algorithm of Bubeck et al. (2019) achieves $\mathcal{O}\left(K\log(KT)+\sqrt{K\log(KT)\bigl(1+Q_{\infty,2}\bigr)}\right)$ expected regret when $Q_{\infty,2}$ is known, where $T$ is the horizon. This matches the $Ω(\sqrt{KQ_{\infty,2}})$ lower bound up to logarithmic factors and additive terms. We further remove the knowledge of $Q_{\infty,2}$ using an adaptive restart scheme whose path-length estimator has uniformly bounded increments. --- ### Learning Varying Physical Therapist-Patient Interactions for Robot-mediated Upper Limb Task-Specific Training Authors: Jia Quan Loh, Vincent Crocher, Marlena Klaic, Denny Oetomo, Ying Tan Published: 2026-08-17 Categories: cs.RO, cs.LG arXiv: https://arxiv.org/abs/2608.15995v1 Upper extremity motor function recovery is positively linked to Task-Specific Training (TST) and sufficient therapy dosage. Rehabilitation robots can increase TST dosage via controlled, repetitive treatment and free therapists to simultaneously manage other patients, but it has yet to demonstrate significant benefits over conventional treatment. This is potentially linked to inaccurate robotic representation of personalised physical therapist-patient interaction and lack of practice variability during TST. Hence, we advocate for robotic interventions that preserve the personalised physical therapist-patient interactions when delivering TST for patients across varying practise conditions. We propose a Learning-from-Demonstration framework using Task-Parameterised Gaussian Mixture Models (TPGMM) to learn personalised physical therapist-patient interaction in Task-Specific exercises, mapping patient joint kinematics to therapist-applied torques using few demonstrations. The model is generalised to reconstruct therapist torques in new task variations. The framework was evaluated on physical interactions from 14 mock "therapist-patient" pairs over three tasks of increasing complexity, each with six variations. A benchmark comparison against a Look-Up Table was conducted. The results show both methods reproducing interactions in unseen task variations that deviate slightly from the actual interaction, with TPGMM slightly outperforming LUT. Both methods reproduced interactions that gets increasingly closer to the actual interaction as task complexity increases. --- ### A Plug-and-Play 2D Motion Interface for Real-World Motion Language Models Authors: Kaname Yokoyama, Norimichi Ukita Published: 2026-08-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.15984v1 Motion Language Models (MoLMs) typically understand human motions by tokenizing 3D motion and processing the resulting tokens using a language model. However, obtaining accurate 3D motions from monocular videos is challenging, limiting their real-world applicability. To address this issue, we introduce a plug-and-play 2D Motion Interface that enables 3D-pretrained MoLMs to accept 2D motion inputs without modifying or fine-tuning the original models. Experiments on public datasets show that our method achieves performance comparable to 3D motion inputs across multiple MoLMs and outperforms training MoLMs from scratch on 2D motions. We further construct a monocular real-world video motion evaluation dataset and introduce a real-video adapter, demonstrating the usefulness of 2D motions over 3D motions under the evaluated monocular pose-estimation setting. These results suggest that 2D motion provides a practical interface for deploying MoLMs in real-world motion understanding settings. Code is available at https://github.com/irajisamurai/2D-Motion-Interface. --- ### Operator-Theoretic Generalization Bounds for Multitask Deep Learning Authors: Mahdi Mohammadigohari, Thomas Borsani, Giuseppe Di Fatta Published: 2026-08-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.15982v1 We develop operator-theoretic generalization bounds for deep multi-output function classes by representing network layers as Koopman composition operators on vector-valued reproducing kernel Hilbert spaces. In vector-valued Sobolev RKHSs, we derive Rademacher complexity bounds for invertible and width-expanding injective architectures. The estimates separate the output-coupling contribution, represented by the trace of the task matrix, from the layerwise operator norms, Sobolev symbol ratios, determinant factors, and restriction constants generated by the linear maps. We then analyze a distinct one-dimensional Brownian/Cameron--Martin regime. Using the exact anchored derivative-norm characterization of the vector-valued Brownian RKHS, we obtain layerwise bounds for domain-preserving scalar linear maps and anchored diffeomorphic activations; the corresponding factors scale as $|W_l|^{1/2}$ and $\|σ_l'\|_\infty^{1/2}$, respectively, and do not involve Sobolev smoothness exponents. Because the Sobolev and Brownian results concern different hypothesis spaces, neither is asserted to dominate the other uniformly. We additionally formulate shared operator learning across tasks, prove a finite-rank representer theorem, derive the exact finite-dimensional problem for squared loss, and establish a target-transfer bound when the learned operator is obtained independently of the target sample. Synthetic and MNIST studies examine stabilized Sobolev-inspired and Brownian-inspired complexity proxies; these empirical proxies are not evaluations of the proved bounds for rank-deficient architectures. --- ### Whose Gold? Annotator-Pool Disagreement Is Large at the Item Level, and Hidden by Small Leaderboards Authors: Anik Jha Published: 2026-08-17 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.15980v1 Preference benchmarks are built by hiring annotators, and the identity of those annotators is treated as an implementation detail. We measure what that detail buys. On the 2,885 MultiPref items where both pools are internally unanimous, so no tie-breaking convention is consulted at all, expert and crowd annotators assign a different majority label to 23.6% and name the opposite winner on 9.2%; on the 246 comparably unanimous MT-Bench cells, benchmark authors and recruited experts differ on 30.5% and reverse on 8.5%. Yet on both corpora the resulting model leaderboards are bit-identical: Kendall tau = 1.00 with zero of six models displaced. That invariance is far weaker evidence than it looks, and we quantify how weak. Switching pools moves a model's win rate by 1.9pp (SD), one adjacent pair in our own leaderboard sits 0.8pp apart and had a 38% chance of swapping, and an item-level bootstrap displaces at least one model in 28% of resamples. The observed zero is the common outcome, not a property of aggregation: on the same measured perturbation, a ten-model leaderboard is displaced with probability 0.86 and a twenty-model leaderboard with probability 0.9997. Reporting a six-model leaderboard is safe; the safety does not generalise, and everything that consumes labels per item is not safe at any size. We make the distinction precise, show that a widely used dataset's stated assumption of no intra-group annotator variability is false, and show that an LLM judge tracks the crowd pool over the expert pool on all three models we test, including one from a different vendor. All code, per-call outputs, and pre-registered decision rules will be released upon acceptance. --- ### ALPS: Measuring Valid Creativity in Large Language Models with Mathematical Construction Authors: Eric Xie, Wenqian Ye, Aidong Zhang Published: 2026-08-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.15979v1 Large language models produce outputs presented as discoveries - new proofs, conjectures, or molecules. Whether such an output that appears creative is truly original and effective is hard to establish: open-ended outputs require subjective judgment, the output may replicate something seen in training, or the task may be too simple to need creativity. We present ALPS (Austin-Law Proof-Synthesis), a benchmark that designs a task to measure valid creativity: producing a solution that is original and can be proven correct. Each instance is a single equational law, certified to require either the construction of an infinite mathematical structure satisfying the law, or a proof that no such structure exists. Submissions are verified by automated proof checking with no human involvement, and a public generator produces new instances without limit, so LLMs are never evaluated on problems they may have seen. A portfolio of eight configurations of leading automated provers resolves 2.2% of the 4,141-law evaluation pool, and a twentyfold budget increase adds 0.6%: the obstacle is not compute, but the absence of any method that produces the tailored structure each law requires. Under a fixed protocol, the strongest reasoning model we test succeeds in 14% of instances on the proof side, but none on the construction side. The remaining 97.2% of the pool is unresolved at every configuration and budget we test. We release ALPS in full: the corpus, the generator, and the automated judge. --- ### Fiber Fingerprints of Hidden Learning-State Dynamics Authors: Qinyou Wang Published: 2026-08-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.15976v1 A learning system can occupy execution states that are indistinguishable under every declared present-behavior readout yet respond differently to future training. We formalize this through fiber fingerprints: controlled future-learning response laws restricted to present-behavior equivalence classes. Prefix-compatible finite probes induce a predictive quotient functor, a Nerode-type minimal recursively sufficient representation, and a canonical set-level predictive fiber without assuming smoothness, reversibility, finite rank, or a manifold. Under an explicit finite-dimensional Hilbert realization, response decomposes into visible, visible-mode-reuse, and irreducible-new sectors; a history-reachability bridge retains only distinctions generated by natural training histories. Conditional mechanism results then identify a graph-Hodge chronology decomposition, a regular switching class with root-mean-square scale $\sqrt{p}η^{3/2}$ and finite-scale corrections, and an exact Adam moment section whose immediate adaptive field is constant while common future gradients can reveal hidden moment differences. Frozen Transformer--LoRA--AdamW studies with Qwen2.5-7B and Mistral-7B-v0.3 support a local action backbone, longer-horizon first-return non-closure, and fresh visible-relative completion with output-range reuse and a low-rank irreducible sector. Stronger claims remain bounded by preregistered negative or mixed results: re-anchored transport is unresolved above its measurement floor; the strict finite-grid Hodge--$3/2$ conjunction is unmet despite prospective contraction; Qwen accessibility is not established in the frozen raw moment chart; and Mistral revelation is future-context dependent rather than bank invariant. Within these support-, scale-, metric-, and context-resolved boundaries, present behavior is not a sufficient statistic for declared future learning. --- ### A Scalable Pipeline for LLM-Teacher Distillation Labeling: Work-Stealing Job Scheduling and Memory-Aware GPU Concurrency Authors: Ravi Satya Durga Prasad Yenugula Published: 2026-08-17 Categories: cs.DC, cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.15975v1 Labeling large text corpora with LLM teachers has become a practical route to training data at scale. At millions of items, hand-labeling every batch is not feasible, and two questions dominate: what label quality a teacher buys per dollar, and how to keep a fleet of GPU workers busy under skewed, failure-prone workloads. We present a simple, reproducible pipeline that addresses both. First, a work-stealing ring pool: each worker owns a queue, drains it first, and then steals from ring successors, with exactly-once task claims via atomic conditional writes and crash tolerance via stale-claim sweeping. The claim protocol requires only a compare-and-set primitive from its storage layer; we implement it on a single SQLite file, which makes the reference implementation dependency-free and the experiments reproducible on one machine. Second, a memory-aware concurrency rule that sizes per-node parallelism by how many model copies fit on the GPU, so the same code runs safely across device sizes. Third, a relabeling benchmark methodology in which the teacher relabels a public dataset that already has gold labels, so quality reduces to an agreement measurement and cost follows from measured throughput. Under skewed load the pool sustains up to 3.4 times the throughput of static sharding while matching it at zero skew, loses 0 of 2,000 tasks when half the workers are killed mid-run (static sharding loses 953), and yields measured quality and cost points for an instruction-tuned teacher on irony and sentiment tasks. All experiments run on public data and commodity hardware; code, tests, and run logs are released. --- ### Improving LLM-based Autonomous Web Agents with Filtering Authors: Zhitong Guo, Jing Yu Koh, Ruiyu Li Published: 2026-08-16 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.27770v1 Autonomous web agents, powered by Large Language Models (LLMs), have garnered significant attention for automating various web-based tasks with multi-step reasoning and decision-making capabilities. An open research question in the development of these agents lies in the format of the webpage input. Raw HTML source code, with its extensive and often irrelevant details, poses difficulties for LLMs with limited context windows. To address this challenge, we first reproduce baseline models such as GPT-3.5 and LLaMA-2-70B on the WebArena (Zhou et al., 2023) benchmark, identifying common failure modes. We then propose two retrieval strategies to filter out irrelevant context for LLM agents. We develop DeBERTa-based and T5-based models that rank HTML elements by their relevance to the task. We fine-tune them on Mind2Web trajectory data and transfer them to WebArena. Experiments show that our DeBERTa-based model improves the success rate of the LLaMA-2-70B LLM agent on WebArena from 1.97% to 2.96%. Moreover, we develop a zero-shot ColBERT-based retriever that is able to retrieve the ground-truth element with a recall of 0.52 on Mind2Web and 0.47 on WebArena. --- ### Alignment of LRMs via Counter-Aligned Few-Shot Conversation Exposure Authors: Xiangyu Zhou, Saleh Zare Zade, Dongxiao Zhu Published: 2026-08-16 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.27763v1 Large Reasoning Models (LRMs) rely on explicit chain-of-thought (CoT) reasoning and large context windows to achieve strong performance on complex tasks, but these features also introduce new attack surfaces. We show that LRMs' reasoning processes can be systematically steered by prepending counter-aligned few-shot conversations containing explicit CoT traces, leading to unsafe generations on harmful queries and unwarranted refusals on benign ones. We formalize this attack as SRCF (Steering Reasoning via Counter-Aligned Few-shot Conversations) that operates solely through a flexible conversational interface and requires no access to the model's parameters and gradients. Our key insight is that SRCF exploits an adversarial generalization issue that induces a representation drift, causing the representations of benign and harmful inputs to shift in a similar direction. This observation motivates our post-training defense, ARCF (Aligning Reasoning via Counter-Aligned Few-Shot Conversations), which exposes models to counter-aligned conversational contexts while enforcing aligned targets. ARCF is compatible with existing post-training methods and consistently improves safety and helpfulness without degrading utility. --- ### "What's That Sound?": A Versatile, Robust, and Lightweight Convolutional Transformer for Environment Sound Recognition Authors: Julia Huang Published: 2026-08-16 Categories: cs.SD, cs.LG arXiv: https://arxiv.org/abs/2609.27762v1 The conventional hearing aid is both costly and lim- ited in usage, as it is not intended to detect non-speech audio. Our objective is to develop a machine learning solution to provide a more accurate and affordable mechanism to identify surrounding sounds to improve the safety of the hearing impaired, i.e., if a car is honking behind pedestrians, or a gunshot is fired, and they need to move away from the source. By adding randomized augmentations to audio, concatenating a Mel-Frequency Cepstral Coefficients (MFCCs) diagram and a log-mel Spectrogram, and including Convolutional Neural Networks (CNNs) in a Trans- former architecture, the Randomized Audiomentational Layered Convolutional Transformers (RALCT) model efficiently extracts features from diversified audio representations. In addition, RALCT is small enough, with only approximately 310,000 parameters, to be deployed into mobile devices. Experimental results on the UrbanSound8K dataset resulted in an accuracy consistently over 93% for all variations of RALCT with the highest at 94.56%, reaching state-of-the-art levels. To leverage the capabilities of this technology, a mobile app is developed to be integrated with the model to provide real-time safety control. RALCT thus represents a robust, lightweight, affordable, and versatile deep learning tool to aid the navigation and safety of the hearing impaired. --- ### Correcting Learning-based Perception for Safety Authors: Yan Miao, Hussein Darir, Sayan Mitra Published: 2026-08-16 Categories: cs.LG, cs.CV, cs.RO arXiv: https://arxiv.org/abs/2609.22108v1 Learning-enabled perception is important in many autonomous systems. Unlike traditional sensors, the boundary where ML perception does or does not work is poorly characterized. Incorrect perception can lead to unsafe or overtly conservative downstream control actions. In this paper, we propose a two-step strategy for correcting ML-based state estimation. First, an offline computation is used to characterize the uncertainties resulting from the ML module's state estimation, using preimages of perception contracts. Second, at runtime, a risk heuristic is used to choose particular states from the uncertain estimates to drive the control decisions. We perform extensive simulation-based evaluation of this runtime perception correction strategy on different vision-based adaptive cruise controllers (ACC modules), in different weather conditions, and road scenarios. Out of 45 ACC scenarios where the original perception-based control system using Yolo and LaneNet led to safety violations, in 73% of the scenarios, our runtime perception correction preserved safety; our method wouldn't be able to recover 27% of the scenarios where the construction of the preimages of perception contracts is not fully conformant. Further, our runtime perception correction strategy is not overly conservative---on the average only a 2.8% increase in completion time is experienced in the corrected scenarios, with mild interventions. --- ### Generalized Multimodal Foundation Model Authors: Huizi Cui, Zongbo Han, Chenggong Ding, Naichuan Xiao, Jialong Yang, Jingdong Chen, Guangyu Wang, Qinghua Hu, Changqing Zhang Published: 2026-08-16 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2609.22107v1 Making prediction with multimodal data is widely used in diverse scenarios. Existing multimodal fusion models, once deployed, can only handle predefined modalities (e.g., vision, text and audio) and single tasks, making it difficult to quickly adapt to new downstream applications. Therefore, a natural yet rather aggressive question arises, whether there exists a general multimodal fusion model that can be applied to arbitrary modality combinations and arbitrary prediction tasks. We argue that a unified multimodal fusion model should not depend on specific modalities and instead encode transferable patterns of multimodal correlation. To this end, we propose a simple and effective learning paradigm based on training over the generation of large-scale synthetic multimodal datasets with diverse causal structures that formally characterize the generative processes of multimodal data in real world. Building on this framework, we propose the generalized multimodal foundation model, a unified foundation model for generalized multimodal learning. By constructing large-scale synthetic multimodal datasets with diverse correlation patterns, our model encodes transferable multimodal correlations during training and activates appropriate associations through in-context examples during inference. Extensive experiments on 18 real-world datasets spanning 12 modalities and 11 prediction tasks demonstrate that our model achieves competitive performance with specialized models without task-specific adaptation. --- ### PRQuant: Permutation Residual Quantization for Low-Overhead Inference Authors: Peiran Wang, Anqi Wang, Jiaying Zhao, Huiwen Yang, Zhenyu Ming, Yuantian Shao, Rongqian Wang, Yiwu Yao, Kun Tian, Xin Yao, Gong Zhang, Fan Yang et al. Published: 2026-08-16 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2609.22106v2 Low-bit quantization of linear layers is often dominated by a small number of outlier channels. Existing smoothing, rotation, and residual-based methods can mitigate this issue, but may shift the quantization bottleneck to weights or introduce costly online operations. To address these limitations, we propose PRQuant (Permutation Residual Quantization), a training-free framework that combines channel permutation with offline weight residual compensation. PRQuant identifies the scaled-weight columns with the largest quantization errors and permutes them into contiguous tail blocks. This structure allows the corresponding weight residuals to be precomputed entirely offline, while replacing scattered activation gathering with simple contiguous access during inference, yielding a single regular MXFP4 GEMM for compensated computation. Experiments show that PRQuant substantially reduces down-projection reconstruction error, with scaling and residual compensation providing the main numerical gains while permutation enables a hardware-friendly contiguous layout. Comprehensive experiment results on Qwen3-4B-Instruct-2507 and Qwen3-30B-A3B-Instruct-2507 illustrate that PRQuant achieves up to averagelly 2.6x and 1.8x operator speedup over BF16 respectively, while preserving near plain MXFP4 end-to-end decoding efficiency. Across five downstream benchmarks, PRQuant achieves the best average accuracy among the quantized methods, improving accuracy over MXFP4 by 1.24 and 0.55, respectively. --- ### Self-Indexing Attention for Compression-Compatible Sparse Long-Context LLM Inference Authors: Xu Yang, Jiapeng Zhang, Zhangke, Changjian Chen, Yuxin Chen, Feiqiang Sun, Chengguang Xu, Feng Jin, Zhuo Tang Published: 2026-08-16 Categories: cs.IR, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.13205v1 Sparse long-context inference requires efficient token retrieval in both prefill and decode. Existing methods often use different retrieval strategies for the two stages, preventing one retrieval representation from being reused throughout inference. We propose Self-Indexing Attention, a training-free framework built on a shared transform-domain sign-magnitude representation. The key signs provide a reusable token-level index for grouped prefill selection and decode retrieval, while the same representation remains compatible with external KV-cache compression without separate indexer metadata. This 1-bit index enables efficient retrieval through bitwise operations widely supported by modern accelerators. At 5% attention density, Self-Indexing Attention remains close to dense attention on LongBench and RULER and achieves up to 6.1x prefill and 10.3x decode attention-operator speedups. Experiments with TurboQuant and DeepSeekV4-Flash further demonstrate compatibility with low-bit KV-cache compression and pretrained sparse-attention indexers. --- ### How Language Models Choose Sides: Internal Representations of Instruction Hierarchy Authors: Enrique Balp-Straffon, Chih-Hao Hsu, Rushiraj Gadhvi, Sunishchal Dev, Callum Stuart McDougall, Anusha Mujumdar Published: 2026-08-16 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.28648v1 We study how instruction-tuned LLMs arbitrate direct conflicts between system and user instructions. We introduce a benchmark of 41 paired constraints with deterministic verifiers and evaluate eight models under matched baseline, conflict, and same-channel control conditions. Behaviourally, the models split into three regimes by System Authority Delta: hierarchy-respecting models use the system channel as an authority signal, anti-hierarchy models follow the system less often than their same-channel baseline predicts, and no-effect models show little channel sensitivity. Llama-3.1-8B is the strongest anti-hierarchy case in our suite, following the system in only 0.10 of conflict trials. We use this behavioural failure case to ask whether user-preferring arbitration reflects the absence of an internal conflictresolution signal. It does not: on Llama-3.1-8B, the conflict outcome is linearly decodable from residual-stream activations at 0.97 balanced accuracy, 17 percentage points above a metadata-only baseline, with analogous signals on Qwen2.5-7B and gpt-oss-20b. Steering with a layer-12 mean of four per-conflict logistic-regression directions raises genuine system compliance from 0.132 to 0.530, while directions selected mainly for pooled separability steer poorly. User-preferring conflict resolution can therefore coexist with a readable internal arbitration signal, and successful intervention depends on the geometry of the readout rather than probe accuracy alone --- ### Self-Specialized Teachers for Domain Post-Training Authors: Yifei Li, Rongman Xu, Lingling Zhang, Muye Huang, Zihan Ma, Jiashuai Liu, Hang Yan, Heng Wang Published: 2026-08-16 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.28647v1 Target-only post-training can improve performance in a specialized domain while degrading behaviors that a general-purpose base model acquired before adaptation. We study this problem when target-domain data are available but a representative replay corpus is not. We propose self-specialized teacher distillation (SSTD), a two-stage procedure that first trains a copy of the base model into a domain teacher, then distills its token distribution to a student on prefixes sampled from the student itself. Teacher training combines standard target supervision with base-aware key-token weighting and distribution alignment to the frozen base model; on-policy distillation then places domain feedback on states the student can encounter at inference time. On financial numerical reasoning, medical question answering, and legal holding identification, SSTD retains much of the target improvement of direct fine-tuning while improving the mean score on the evaluated general suite by 4.8--5.0 points at the reported operating point. The pattern persists across Qwen3 sizes and on Gemma backbones. SSTD requires neither an external teacher nor general replay data. --- ### Few-Shot Cross-Dataset Adaptation for Tuberculosis Detection Using DenseNet Authors: Bidhan Biswas, Shahadat Hossain Sohag, Nabil Ashab, Soumit Kumar Kundu, Saif Mahmud Parvez Published: 2026-08-16 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.21427v1 Tuberculosis (TB) is one of the most common and dangerous bacterial ailments. Every year, it causes a large number of deaths worldwide. Although many deep learning models can detect tuberculosis from chest X-rays quite accurately, severe domain shift across datasets makes the task challenging. Different imaging protocols, patient demographics, and equipment across domains make the task of generalization difficult. In real-world settings, a model may perform well on one dataset but show a noticeable drop in performance when tested on another. In this work, we address this domain adaptation challenge through a few-shot scaling study. A controlled cross-dataset evaluation is presented in this paper using TBX11K as the source domain and the Mendeley TB dataset as the target domain. It is investigated how varying the number of target samples affects model performance under three training regimes: frozen backbone adaptation, full fine-tuning of a source-pretrained DenseNet121 model, and training from scratch. The results indicate that the model can perform well even with limited data and can achieve 98.36\% accuracy with just 75 labeled samples per class. The adaptation curves demonstrate how fine-tuning effectively mitigates domain shift. These findings establish full fine-tuning of pretrained models as a highly effective and practical strategy for mitigating domain shift in low-resource clinical deployment scenarios. --- ### AI Visual Inspection for Garment Production Authors: Ray Wai Man Kong, Ding Ning, Theodore Ho Tin Kong Published: 2026-08-16 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2608.21426v1 The garment manufacturing industry is under increasing pressure to improve product quality, reduce costs, and accelerate digital transformation toward Industry 4.0. One of the most challenging quality-control activities is sewing-line inspection, where defects such as broken stitches and skipped stitches are difficult to detect consistently through manual inspection. Human-based inspection is often affected by fatigue, subjective judgement, and inconsistent performance, resulting in defect leakage, rework, and reduced production efficiency. This study presents the development and validation of an Artificial Intelligence (AI)-based visual inspection system for garment sewing-line quality control. The system utilizes Convolutional Neural Networks (CNNs) to detect sewing defects and was initially trained using black fabric and black sewing thread samples. Experimental testing was conducted on black, red, dark green, light blue, silver, and fluorescent yellow fabrics. The results demonstrated successful detection of jump sewing-line defects on black, red, and dark green materials, while performance limitations were observed for broken sewing-line defects and fabrics with significantly different visual characteristics, including light blue, silver, and fluorescent yellow colours. These findings indicate that model accuracy is strongly influenced by the diversity of training data and the ability to generalize across different fabric and thread colours. --- ### Aligning Human Sense: Calibrated Distributional Reward Learning for Video Generation Authors: Nai-Xin Zhai, Weihua Cheng, Dexu Yu, Yikai Gu, Hanwen Du, Junchen Fu, Chenxi Huang, Yingwei Song, Liyuan Lillian Ma, Yang Ran, Youhua Li, Yongxin Ni Published: 2026-08-16 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.21425v1 Video generation is central to AI-powered content creation. Aligning generated videos with human preferences is a key criterion for evaluating generation quality. Despite significant progress in visual quality, three key challenges remain. First, the reliability of reward signals is constrained by the quality of human preference data, which is often affected by subjective noise and bias. Second, standard scalar reward models collapse multi-aspect human preferences into a single value, leading to the loss of dynamic trade-offs across multiple preference dimensions. Third, in policy optimization, the widely adopted KL divergence imposes primarily local constraints and may fail to capture the global structure of human preferences. To address these challenges, we propose a unified preference-aware learning framework for video generation. First, we introduce elite-guided filtering to calibrate preference data and construct reliable supervision for reward model training. We then model video quality as a multidimensional reward distribution to capture the uncertainty inherent in human preferences, and use the Wasserstein distance to align the learned reward distribution with the empirical human preference distribution. Finally, we introduce Wasserstein-based distributional alignment into GRPO, guiding policy optimization to better match the global structure of human preferences over videos. Experiments on reward modeling and video generation demonstrate that our approach improves the reliability of reward signals and the perceptual consistency of generated videos. Our code is available at https://github.com/alignhs26/ahs. --- ### EditStream: A Unified Autoregressive Framework for Interactive Video Generation and Editing Authors: Yuqian Zhou, Zhenghong Zhou, Zongze Wu, Cameron Smith, Richard Zhang, Jiebo Luo, Eli Shechtman, Zhe Lin Published: 2026-08-16 Categories: cs.CV, cs.GR, cs.HC, cs.LG, cs.MM arXiv: https://arxiv.org/abs/2608.21424v1 Interactive video generation and editing are becoming increasingly important for creative design. In this report, we introduce EditStream: a unified framework for interactive video generation and editing. EditStream unifies multiple video creation and manipulation tasks within a single DiT-based model through flexible task-specific conditioning, and further transforms it into a fast, few-step autoregressive model for efficient streaming. It supports Text-to-Video, Image-to-Video, Video-to-Video, Editing Propagation, Reference-guided Video Editing, and Camera Pose Change, enabling flexible control over video generation, transformation, and editing within one system. To make the unified model practical for interactive use, we develop a two-stage distillation approach that combines Velocity Moment Matching (VMM) with autoregressive unrolling. VMM matches conditional velocity moments at student-reached intermediate states to preserve generation quality and motion, while unrolling exposes the student to its own autoregressive predictions to improve temporal stability. Together, they alleviate common challenges in few-step autoregressive video generation, including over-saturation, degraded motion, temporal instability, and complex training. EditStream provides a practical and scalable solution that bridges high-quality diffusion-based video models with interactive creative workflows. --- ### Agentic Security: A Systematization of Tools, Failure Modes, and Design Laws for LLM-Driven Penetration Testing Authors: Israt Moyeen Noumi, Tarannum Ahmed Nowshin, Md. Mehedi Hasan Nipu, Mohammad Sakib Mahmood, Md. Jakir Hossain, M. F. Mridha Published: 2026-08-16 Categories: cs.CL, cs.AI, cs.CR, cs.MA arXiv: https://arxiv.org/abs/2608.21423v1 Agentic security uses large-language-model (LLM) agents to plan, dispatch, and interpret security tools. As these systems move from demonstrations to deployed products, practitioners repeatedly encounter the same operational failures. We systematize these failures through a hands-on evaluation of ten widely used static, dynamic, cloud, orchestration, and AI red-teaming tools for unattended pipelines. We introduce a four-dimensional Integration Friction Index that separates one-time engineering cost from recurring organisational, legal, and maintenance cost. We then derive quantitative regularities that explain recurring failure modes. Modelling an agentic security system as stochastic LLM policies wrapped by a deterministic mediator, we show that long-lived sessions lose resident evidence with phase count, while short-lived sub-agents extend the usable horizon according to the compression ratio between raw evidence and its summary. We show that a two-stage verdict cascade multiplies scorer likelihood ratios, but provides little benefit when scorer errors correlate. We show that treating unevaluable outcomes as attack failures biases downstream measurements toward evasive and severe responses. We formulate planner-versus-worker model routing as a knapsack problem and derive a closed-form execution cap for heavy-tailed tools, eta* = alpha v/c. Finally, we show why scope and budget enforcement cannot be delegated to system prompts: prompts do not constrain what actually executes. Inspectra, our implemented platform, serves as a worked instantiation, with mechanisms labelled shipped, partial, or planned, including those that did not work. --- ### Demons on a Budget: Adaptive Measurement Placement at the Entanglement Phase Transition Authors: Rohan Pandey Published: 2026-08-16 Categories: quant-ph, cs.LG arXiv: https://arxiv.org/abs/2608.19248v1 Monitored quantum circuits exhibit a measurement-induced phase transition between volume-law and area-law entanglement as a function of the measurement rate $p$. Prior work places measurements at random locations and treats the rate as the control parameter. We instead fix the measurement budget and vary the placement process, comparing random placement against hand-designed and learned policies in brickwork random Clifford circuits at matched budget. First, placement geometry matters more than placement information. A deterministic contiguous sweep cuts the half-cut entropy by a factor of 3.4 relative to random placement, while equal-coverage unstructured placement and a greedy policy with full state access do far worse. The effect is carried by spatial order alone: measuring the $k$ least recently measured sites gives $4.14 \pm 0.06$ bits with random tie-breaking and $1.29 \pm 0.04$ bits with position-ordered tie-breaking. Second, the sweep eliminates the transition rather than shifting it. Tripartite mutual information crossings recede as $p^* \propto 1/L$, the steady-state entropy saturates at an $L$-independent ceiling near $0.46/p$, and data for $64 \le L \le 512$ collapse onto the form $S = p^{-1} f(pL)$ predicted by a ballistic regrowth argument. Third, in stabilizer dynamics every outcome is deterministic or a fair coin flip, so the record's Shannon entropy is exactly countable; the sweep dominates the entropy-versus-record-cost frontier while paying the same roughly one bit per measurement as random placement. Policies trained by cross-entropy and proximal policy optimization do not find the sweep: score-based policies parameterize which sites to measure, not the order in which degenerate scores are resolved, and the effect lives in that order. The phase diagram of monitored dynamics is a property of the placement process, not only of the measurement rate. --- ### Adversarial Review: Structured Disagreement for Grounded Agentic Code Review Authors: Eric S. Qiu, Joyce Gill Published: 2026-08-16 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2608.18167v1 Early multi-agent LLM systems often used role-separated teams, yet scaling agent count yields diminishing returns on repository-level coding tasks. Recent alternatives treat agents as passive tools (subagents), yet this removes the benefits of agent interaction entirely. We study whether a subagent paradigm can support a middle ground: minimal agentic cooperation without the overhead of large multi-agent teams. We introduce Adversarial Review (AR), a minimal cooperative code-review protocol in which a main coding agent works with a reviewer and a critic agent. The reviewer evaluates code, while the critic audits the review through structured disagreement before the main agent edits. On LiveCodeBench, AR achieves the highest pass rate among tested methods, outperforming a five-agent baseline while using only three agents. On SWE-PRBench, naive AR exposes a false-consensus failure mode, where agents converge on agreement without sufficient evidence, but a single prompt iteration that adds disagreement explicitly achieves the highest F1 among tested methods. On SWE-bench Verified, AR also shows improvements over the baselines on repository-level coding tasks. Together, AR demonstrates that cooperative code review does not require many agents or complex communication structures: it requires that disagreement be minimal, structured, and evidence-grounded. --- ### TractoGraphVLM: A Unified Vision-Language Framework for White Matter Tractography Authors: Gurucharan Marthi Krishna Kumar, Janine Dale Mendola, Amir Shmuel Published: 2026-08-16 Categories: eess.IV, cs.CV arXiv: https://arxiv.org/abs/2608.18166v1 Vision language models have transformed 2D medical imaging, yet extending them to 3D white matter tractography remains challenging due to the complex topology of fiber bundles. We introduce TractoGraphVLM, a unified framework for four tasks, bundle classification, text-to-tract retrieval, anatomical captioning, and visual question answering, built on a shared GPS architecture, training procedure, and read-out design. Fiber bundles are represented as streamline graphs whose nodes encode 3D position and tangent orientation. A General, Powerful, Scalable (GPS) graph transformer produces bundle embeddings aligned with a frozen BiomedBERT text encoder via contrastive learning, while a BioGPT decoder with visual prefix tokens generates captions and answers. A single shared encoder and decoder is trained jointly across all four tasks and evaluated from one checkpoint. Trained on HCP Young Adult subjects, TractoGraphVLM achieves 91.8% bundle classification accuracy, 84.7% retrieval R@1, BLEU-4=20.1, ROUGE-L=66.8, and 66.4% VQA accuracy on a held-out test set. The same checkpoints transfer zero-shot to HCP Aging subjects, with a modest drop on discriminative tasks and a larger drop on generative tasks, showing robustness to age and acquisition shift. Language supervision yields richer representations than label-only training, recovering structure like hemisphere and fiber family, carried by captions but never given as a label. Swapping only the visual encoder, graphs preserving fiber orientation outperform volumetric baselines, with GPS giving the best balance. Generative metrics measure consistency with a structured knowledge base rather than independent clinical text; even so, TractoGraphVLM shows that classifying, retrieving, describing, and answering questions about a white matter bundle can be served by one jointly trained model that learns transferable neuroanatomy from language alone. --- ### RDFdL: Integrating RDF with Differential Dynamic Logic Authors: Yuyang Li, Lukas Kubelka, Julia Butte, Tobias Käfer Published: 2026-08-16 Categories: cs.AI, cs.FL, cs.LO arXiv: https://arxiv.org/abs/2608.18165v1 Knowledge graphs modeled in RDF are powerful for describing static knowledge, but they cannot capture or reason about the dynamic behavior of physical systems, e.g., systems described by differential equations, which is a critical gap for AI-driven cyber-physical systems. To solve this, we propose RDFdL, a framework that integrates RDF with Differential Dynamic Logic (dL) to represent and reason about both static knowledge and the continuous dynamics of physical systems. For the dynamic part, we syntactically represent differential equations and ranges in the state space in RDF and SHACL and provide semantics using a translation to dL. Linking RDF and dL through their shared foundation in first-order logic achieves a unique integration: verification results for safety and reachability properties in the dynamic logic domain become available as entailment to SPARQL queries over RDF data. We implement the pipeline using Apache Jena for ontology-driven RDF reasoning and KeYmaera X, the theorem prover for dL, and sketch its applicability in manufacturing. --- ### MagViT: Interpretable Multi-Magnification Transformers with Patient-Level Model Selection for Breast Histopathology Authors: Nabil Ashab, Soumit Kumar Kundu, Saif Mahmud Parvez, Shahadat Hossain Sohag, Bidhan Biswas, Nazmus Subha Published: 2026-08-16 Categories: eess.IV, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2608.16959v1 Breast cancer is one of the most common types of cancer among women around the world. Rapid detection and early treatment can hinder its progress to more complex stages and can impede its spread to other parts of the body. Histopathological image classification is the most common task in cancer detection due to its robustness in analyzing cellular data. Breast histopathology classification requires handling both multi-scale tissue morphology and clinically relevant generalization beyond the source domain. This paper presents MagViT, an interpretable multi-magnification transformer framework with scale-gated fusion and patient-level model selection. The model uses four BreakHis magnifications (40X, 100X, 200X, 400X) and extracts per-scale representations with a ViT backbone, and combines them via a learnable gate that masks missing scales. Patient-level five-fold cross-validation with a fixed seed has been run and compared with three architectural branches. The most accurate branch is then selected as the final model due to the strongest patient-level accuracy while retaining the simplest fusion pathway. On BreakHis, our architecture achieves a mean image accuracy of 0.9191, a mean patient accuracy of 0.9643, and a mean macro-F1 of 0.9042. External transfer experiments provide preliminary evidence of cross-dataset generalization under controlled adaptation settings on BUSI (image accuracy 0.8306, macro-F1 0.7480, patient accuracy 0.8291) and IDC (image accuracy 0.8577, macro-F1 0.8191, patient accuracy 0.8372). Grad-CAM visualization indicates that the model focuses on diagnostically significant and meaningful regions across magnifications. Relative to prior ViT-centered BreakHis work, this study emphasizes patient-level selection and cross-dataset robustness under a reproducible protocol. --- ### ORViT-DR: Ordinally-Robust Hybrid ViT for Low-Resolution Diabetic Retinopathy Grading Authors: Soumit Kumar Kundu, Nabil Ashab, Bidhan Biswas, Shahadat Hossain Sohag, Saif Mahmud Parvez, Souvik Kumar Kundu, Zunayed Ahmed Rafi Published: 2026-08-16 Categories: eess.IV, cs.CV arXiv: https://arxiv.org/abs/2608.16958v1 Diabetic retinopathy (DR) is one of the main causes of impaired vision. A good and reliable automated grading system can make the screening process safer and more accurate. Because DR stages progress gradually, the task of grading disease severity naturally follows an ordinal structure in which neighboring classes share similar visual characteristics. In this study, ORViT-DR, a hybrid deep learning framework, is designed to improve DR grading from low-resolution retinal images. The proposed approach combines convolutional feature extraction with transformer-based global context modeling through a pre-trained ViT-Hybrid backbone, which integrates BiT-ResNetv2 with a Vision Transformer architecture. The approach is tested on the RetinaMNIST subset of the MedMNISTv2 dataset, which contains 28x28 retinal fundus images annotated with five levels of disease severity. To promote stable training and better feature learning, the training strategy applies progressive layer unfreezing, layer-wise learning rate decay, exponential moving average (EMA) parameter updates, and ensemble-based prediction during inference. Experimental results on the official RetinaMNIST test set show that the proposed method achieves 57.00% classification accuracy, along with a quadratic weighted kappa score of 0.5963 and a macro-F1 score of 0.4293. These results suggest that hybrid CNN-Transformer architectures can provide effective representations for ordinal retinal image analysis. --- ### The Price of Thinking: Reasoning Effort as a Model-Specific API Contract Authors: Yeabin Moon Published: 2026-08-16 Categories: cs.AI, cs.CL, cs.CY, cs.LG arXiv: https://arxiv.org/abs/2608.16956v1 API buyers purchase a dated contract, not a model name alone: the contract includes the requested and served model, reasoning-effort term or its omission, output rail, service product, prompt, and price schedule. We study the reasoning-effort term through a registered paired contrast of Sonnet 5 with explicit high effort against the same model with effort omitted, using 30 AIME 2026 items and five calls per item. Every paid attempt was assigned one frozen terminal category, and inference resampled items while retaining their repeated calls. Mean delivered cost was \$0.01031 per call higher under the explicit-high contract than under the omitted contract (+\$0.00204, +\$0.01974). The corresponding accuracy contrast was +0.0133 (-0.0267, +0.0467); we did not detect an accuracy difference, and the interval permits a gain of up to 4.67 percentage points that this design cannot rule out. Cost per correct answer was \$0.08665 under the high-effort contract and \$0.07662 under the omitted contract, as registered point estimates. A dated contract census, Models-API metadata, and preregistered raw-response probes further documented model-specific omission semantics, including within a provider; claims remained at documentation grade when raw structure was indeterminate. The request registry, parser, terminal taxonomy, statistical plan, and analysis pipeline were frozen before outcomes were examined; the resulting claims are bounded to the model, task, and collection date studied. --- ### WONDER: A Radio World Model-based Negotiation Framework for Multi-Agent UAV Coverage Optimization Authors: Jiahao Huang, Rongpeng Li, Zhifeng Zhao, Guoru Ding, Honggang Zhang Published: 2026-08-16 Categories: cs.MA, cs.LG arXiv: https://arxiv.org/abs/2608.16955v1 Post-disaster damage to terrestrial infrastructure can disrupt wireless coverage,while Uncrewed Aerial Vehicle (UAV) swarms provide a promising solution for rapid restoration.However, due to the limitations in local geometry observations hidden radio impact,and inter-UAV communication,there exists a significant gap between locally visible movement choices and swarm-level coverage outcomes.To combat this gap,we propose a raido World-model-based Optimized Negotiation framework for Distributed UAV covERage (WONDER).Particularly, to tackle the unavailability of the future radio field from onboard observations, WONDER uses a Joint-Embedding Predictive Architecture (JEPA)-based radio world model to learn and predict the incremental radio effect of each candidate trajectory from deployment-available information.Multi-round negotiation in WONDER then coordinates ranked proposals by committing one trajectory at a time and re-evaluating the remaining proposals under the updated context. Our theoretical analyses further validate the effectiveness of such a world model-based framework. WONDER also adopts a Proximal Policy Optimization (PPO)-style Actor and alternates between updating the world model and the actor. Furthermore,we build RadioDynamics,a comprehensive simulation environment that integrates UAV mobility,radio propagation, inter-UAV communication modeling,and digital-twin geometry with ray-traced fields in $62$ metropolitan scenes.Experiments on $11$ testing scenes in RadioDynamics show that WONDER achieves the highest balanced score among seven evaluated methods,reaching $0.870$ with a $0.162$ coverage advantage over STACCA, while maintaining $100\%$ connectivity between UAVs. --- ### CM-MAE: A Physics-Guided Cross-Modal Self-Supervised Learning Framework for Vision-Wireless Applications Authors: Yubo Zhang, Yiyao Liu Published: 2026-08-16 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.15972v1 Synchronized camera and wireless measurements observe the same scene through different physical channels. The central difficulty is that a representation learned in one deployment can fail when viewpoint, traffic, illumination, and propagation geometry change. This paper presents CM-MAE, a self-supervised vision--wireless pretraining framework for cross-scenario representation transfer. The evaluated real-data model uses only RGB frames and the measured 64-beam received-power vector available in DeepSense 6G; it does not use ray-traced paths, calibrated depth, or beam-index labels during pretraining. Its central pretraining term is a \emph{soft contrastive alignment loss}. Instead of making the synchronized image--wireless pair the only positive pair, this loss builds a target distribution from similarities between measured beam-power profiles, so nonidentical samples with similar directional responses are not forced apart as false negatives. A masked joint decoder provides the complementary local objective by reconstructing hidden visual patches and wireless angular clusters under modality dropout. After pretraining, a differential-rate fine-tuning rule lets a new fusion head adapt quickly while the encoders move slowly. Under a sequence-disjoint DeepSense 6G protocol, adding the soft alignment loss improves a matched linear-probe transfer average from 24.88\% to 29.49\%. Mild fusion fine-tuning reaches 77.38\% Top-1 accuracy on unseen Scenarios 6--8, and optional transductive normalization adaptation reaches 78.69\%. Since the fusion setting uses the contemporaneous 64-beam power vector at inference, these results should be read as representation-transfer diagnostics, not as proactive beam-prediction or reduced-sweeping claims. --- ### The Limits of Binding in Dual Encoders Authors: Kin Ian Lo Published: 2026-08-16 Categories: cs.LG, cs.CL, cs.CV arXiv: https://arxiv.org/abs/2608.15971v1 Dual-encoder models such as CLIP score an image-caption pair by a single inner product of two independently computed unit vectors, and fail at binding, often scoring near chance when asked to distinguish "a red car and a blue dog" from "a blue car and a red dog". We give a mathematical account of when this failure is necessary and when it is contingent. Working within the ideal-encoder framework proposed by Kang et al., we first show the relevant axioms are satisfiable, so every impossibility must enter through an added, checkable hypothesis. We then prove three such obstructions. Depth: for recursive role-binding codes the swap margin obeys an exact law $m(D) = 2b^{-D}$ in the nesting depth D, with a finite-dimension version holding up to one explicitly flagged concentration estimate; the resolvable depth grows only logarithmically in the dimension and is single-digit at CLIP scale, the nesting depth of ordinary language. Objective: architecture-free throttle theorems showing that the contrastive objective's entire reward for binding is bounded by the rate at which training contrasts a caption against its own swap, a rate that vanishes at web scale, and that exactly reversed binding costs only that rate times the mean binding margin; both are verified in simulation. Geometry: a tight smoothness-binding frontier: the closer the two swap-related captions must embed to a shared paraphrase anchor, the smaller the binding margin can be, with an exact constant. Measuring its text-only diagnostic across 18 deployed text encoders, every model sits at roughly 25-35% of its ceiling, and the induced per-item ceiling tracks SugarCrepe's subset difficulty at r = 0.99. Binding failure in deployed dual encoders is thus not a dimension or smoothness limit today, but an incentive and code-structure limit, with a proved depth ceiling that remains once those are fixed. --- ### BagShift: Measuring How Patch Selection Changes the Evidence Seen by Whole-Slide MIL Authors: Ruicheng Yuan, Zhenxuan Zhang, Liwei Hu, Anbang Wang, Haijie Xu, Jiawei Luo, Guang Yang Published: 2026-08-16 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.15970v1 Whole-slide multiple-instance learning (MIL) observes only the patches admitted by its selector. Deployment can alter this selector through compute limits, tissue masking, or regional workflows, even when the patch count is unchanged. We introduce BagShift, a paired protocol that changes the selector for the same case while holding its features and predictor fixed, thereby isolating selector response from case mix. With equal 128-patch budgets, sampling across the tissue or concentrating around one coordinate exposes markedly different evidence: on PANDA, the two views reduce quadratic weighted kappa by 1.57 and 17.96 points, respectively (QWK reported on the $\times100$ scale). On CAMELYON16, lesion annotations withheld from model development show that localized views retain tumor in only 10.0\% of micrometastatic observations, and matched exposure does not consistently recover the loss. The same fixed-count stressor produces a much smaller response on external lung subtyping, although differences in relative coverage make cross-task severity descriptive. When repeated localized observations are available, unioning their patches before one nonlinear MIL pass improves PANDA QWK by 7.87 points over averaging regional predictions. Patch count specifies computation, not observed evidence; deployment evaluations should report both what a selector preserves and how repeated observations are aggregated. --- ### A Banach-Space Theory of Markovian Halpern Iteration for Non-Expansive Maps Authors: Ege C. Kaya, Arda Fazla, M. Berk Sahin, Abolfazl Hashemi Published: 2026-08-16 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2608.15966v2 We study stochastic approximation of fixed points of a non-expansive operator $T$ when the oracle samples originate from a continuing Markovian trajectory. A direct block-minibatch implementation of Halpern iteration attains an expected last-iterate residual of order $O(\log N/N)$, but accrues a substantive complexity of $\tilde O(ε^{-5})$ Markovian samples. We therefore introduce a variance-reduced Markovian PAGE-Halpern method whose refresh and same-state difference blocks are analyzed through the Poisson equation. In Hilbert spaces, the cocoercivity of $I-T$ results in an $O(ε^{-3})$ sample complexity. Our main result extends this construction to a general finite-dimensional Banach space. A displacement-level Halpern bound replaces the Hilbert-space potential and yields $\tilde O(ε^{-3})$ sample complexity. We also establish a high-probability guarantee with the same leading accuracy dependence by measuring the estimator in an auxiliary smooth norm. --- ### Beat the Counter First: A Baseline for Temporal-Graph Anomaly Detectors Authors: Omair Shafi Ahmed, Zohair Shafi Published: 2026-08-16 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.15965v1 Progress in streaming, edge-level graph anomaly detection (GAD) has been marked by increasingly elaborate architectures, from count-min-sketch chi square tests to memory-augmented attention networks. Yet the empirical gains attributable to this added complexity have not been systematically evaluated. We propose SimpleCount, a reference with no parameter fitting that selects one scalar feature per dataset from a fixed pool of counts, recencies, first-occurrence indicators, and count-derived transforms. We compare SimpleCount with two temporal-graph detector models and an IsoForest control fitted to the complete feature vector across five public datasets and one synthetic dataset. SimpleCount matches or exceeds SLADE on three of six datasets and exceeds IsoForest on all six. We report paired statistical tests and five-seed SLADE evaluations. SLADE requires 23 to 133x more wall-clock time than SimpleCount. On Synth-Triangle and an additional Synth-Quad probe, pre-event structural scores recover the planted signal at AUC up to 0.955, while all evaluated detector models remain near random. The benefit of complexity is dataset-dependent, and every claimed gain should be reported against a strong one-feature reference together with its compute cost. --- ### LLMs Get Smarter from Targeted Synthetic Multilingual Data Authors: Ishika Agarwal, Arkajyoti Charaborty, Tanner Sorensen, Neha Gupta, Andreas Stolcke Published: 2026-08-16 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.15964v1 Language-specific competency (LSC) is the phenomenon of a language model performing better or worse depending on the language of the prompt. In other words, a language model outputs different (and potentially incorrect) responses to the same semantic query when prompted in different languages. Prior work attributes this to an internal misalignment of semantic representation across languages. Currently, there are two main approaches to address LSC in the literature: (1) routing all queries through English, improving performance, but limiting language expressivity to English; or (2) training on language-balanced data, equalizing model performance across languages, but reducing overall performance. In this work, we take a data centric perspective and introduce HOTFIXR: Hardness Optimized Training data For Improving X-Lingual Reasoning. It is a data generation framework that uses models to probe and learn a student model's multilingual weaknesses, and generates data to mitigate them. HOTFIXR can generate multilingual synthetic training data that can improve multilingual performance. We evaluate on three in-distribution tasks, three out-of-distribution tasks, and four out-of-distribution languages. On average, HOTFIXR (1) improves in-distribution performance by 6.2%, (2) reduces catastrophic forgetting (induced by fine-tuning) on OOD tasks by 3.7%, and (3) on OOD languages by 7.1%. Overall, as many real-world applications requires multilingual LLMs, our work contributes to the efforts of making LLMs multilingually proficient. We will release code upon acceptance. --- ### SEER: Long-Context Reasoning via Selective Visual-Text Compression Authors: Jiawei Xu, Zhilin Zhai, Jinrui Fang, Ruohan Xu, Mingfei Lu, Yi Zhang, Guanchu Wang, Tianlong Chen, Ying Ding Published: 2026-08-16 Categories: cs.CL, cs.CV arXiv: https://arxiv.org/abs/2608.15962v1 Long-context reasoning remains computationally expensive for large language models due to the quadratic complexity of attention over text tokens. Visual-text compression offers a promising alternative by rendering text into images and processing them with vision-language models, often reducing token usage. However, existing approaches apply uniform compression regardless of query relevance, potentially sacrificing precision where detailed extraction is required. We present SEER, a framework that learns to select query-relevant images through visual scanning and retrieve textual content only where needed, combining the efficiency of visual compression with the precision of text-based reasoning. Through supervised fine-tuning on tool-interaction trajectories, SEER learns adaptive tool invocation for selection and retrieval. Experiments on long-context benchmarks show that SEER improves extraction precision through selective text retrieval while retaining average prompt-token savings relative to full-text baselines. On LongBench, SEER achieves 51.11% average accuracy, outperforming the visual-text baseline Glyph-9B by 2.33 points and Qwen3-8B by 3.49 points. Code can be accessed at https://github.com/jiaweixu98/SEER --- ### Solvable Sokoban Without a Solver via Diffusion Authors: Sina Baghal Published: 2026-08-16 Categories: cs.AI, cs.GT, cs.LG arXiv: https://arxiv.org/abs/2608.15958v1 Deciding whether a Sokoban puzzle is solvable is PSPACE-complete (Culberson, 1997): solutions can be exponentially long and there is no short certificate to check. Solvability is also a fragile property, since even a single misplaced wall can silently render an entire puzzle unsolvable. In this work, we show that a transformer-based discrete diffusion model trained purely on tile completion, with no access to solvers, rewards, or solvability labels, achieves a solvability rate of 77.4%, with 94.5% of the remaining failures rendered solvable by removing a single wall. In other words, a global, search-heavy property follows from a local training objective: trained only to fill in masked cells, the model inherits solvability it was never trained on. An autoregressive model factorizes as $p(c_k \mid c_1 \dots c_{k-1})$, meaning a fixed order, always conditioned on a prefix. Masked diffusion does not: it hides a random subset of cells and learns $p(c_k \mid \text{any subset})$, so at generation time it can reveal cells in any order, each one conditioned on everything already placed, wherever it sits on the board. A puzzle's difficulty comes from exactly this kind of non-local interaction, a decision in one part of the grid constraining what will work somewhere else entirely. A generator that is not locked into a single fixed order is therefore a better structural match for the problem than one that is. The training pipeline is adapted from MD4 (Shi et al., 2024) and the dataset is DeepMind's Boxoban (Guez et al., 2019). The trained model and instructions for generating puzzles are publicly available. --- ### Navigation-Informed Embeddings: Dense-Retriever Adaptation from Agent Search Traces Authors: Shrey Shah, Levent Ozgur Published: 2026-08-16 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.15956v1 Agentic retrieval workflows produce query, retrieval, and stopping traces as a byproduct of answering questions. We study how these traces can adapt a deployed dense retriever to changing workflow distributions without new relevance labels, synthetic queries, or LLM judgments. We introduce Navigation-Informed Embeddings (NIE), a family of trace-derived objectives. NIE-Stop turns the stopping document into a soft positive; NIE-Path additionally uses preceding path documents as hard comparisons and imposes ordinal constraints with geometric decay. A BGE encoder adapted from retained source trajectories improves support Recall@20 on an independent target benchmark from 72.2 to 78.0 overall. NIE-Stop reaches 76.9 overall and 52.3 on long paths; NIE-Path raises long-path performance to 55.4, compared with 46.7 for the unadapted encoder. A shuffled-order control under the full path objective loses 3.2 points. Without public-benchmark training, the same adapter also improves nDCG@10 by 1.9 points on standard BEIR HotpotQA. NIE therefore provides a lightweight adaptation channel for settings where trajectories are already retained, with zero incremental labeling cost. --- ### Functional anatomy of Pythia-Herwig differences with Kolmogorov-Arnold networks Authors: Arghya Chattopadhyay Published: 2026-08-16 Categories: hep-ph, cs.LG arXiv: https://arxiv.org/abs/2608.15952v1 Differences between high-energy event generators can arise at several stages of the collision simulation, from the hard scattering through parton showering and hadronization to the final event. These differences are usually summarized using observable distributions or global classifier scores. While these quantify the disagreement, they do not reveal which observable-level structures carry it or whether those structures persist through different stages of event generation. In this work, we formulate this problem as a staged functional analysis of generator-model differences. Following the same hard dijet events through Pythia and Herwig at shower-only, hadronized, and full-generator levels, we use an additive Kolmogorov-Arnold network (KAN) representation of the classifier-derived log density ratio to decompose the learned discrepancy into explicit one-dimensional observable responses that can be isolated, recomposed, and transported between generator stages. Within the same eight-observable jet representation, the Pythia-Herwig difference is driven mainly by multiplicity at shower level, shifts toward jet mass and shape after hadronization, and develops a mixed shape-multiplicity driven structure in the full-generator configuration. Transporting the individual shower-level functional components downstream shows that shower-level multiplicity information can retain its reweighting power, whereas the corresponding shape responses need not do so even though shape becomes important again at later stages. The jet-mass factors, meanwhile, are limited by poor statistical support. This KAN-based framework therefore provides a functional anatomy of generator-model dependence, exposing both persistent structures and support failures that are hidden inside a single global classifier-derived reweighting function. --- ### ReliaGate: Reliability Routing for Low-Stakes Wearable Stress Prediction Authors: Jaden Moon, Yu Wu, Arvind Pillai, Andrew Campbell Published: 2026-08-16 Categories: cs.LG, cs.HC arXiv: https://arxiv.org/abs/2608.15951v1 We study when a wearable stress system should surface a prediction rather than change it. In low-stakes reflection and summary settings, aggregate accuracy is insufficient because withholding can reduce error while leaving some people with little or no information. We formulate fixed-label reliability routing: after a locked classifier emits a protocol-defined stress/non-stress label, a post-hoc gate surfaces that unchanged label or withholds it as unavailable. ReliaGate assembles established confidence, signal-quality/trust, agreement, train-standardized atypicality, and train-fitted geometry cues into a post-hoc correctness score. We evaluate four wearable datasets using subject-disjoint folds, validation-selected routing, paired held-out-subject intervals, and pooled and per-subject analyses. WESAD point estimates favored ReliaGate, UBFC-Phys primary coverage/risk intervals favored ReliaGate, and E4 checks were mixed. ReliaGate provides an operational framework for studying surfaced-label error, output availability, and accepted-output distribution across subjects, without revising labels or providing clinical or finite-sample risk guarantees. --- ### Ask to Be Sure: Informative Interactions for Confident Multi-Turn LLM Recommendation Authors: Cedar Site Bai, Zhenyu Liao, Duanshun Li, Sheikh Sarwar, Huiyuan Chen, Yuan Chen, Changhe Yuan, Haiyang Zhang, Qilin Qi Published: 2026-08-16 Categories: cs.IR, cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.15949v2 Recent advances in large language models (LLMs) have enabled their use as conversational recommender systems (CRS), demonstrating strong recommendation accuracy and natural dialogue. However, guiding multi-turn interactions to elicit user preferences effectively remains challenging. Existing approaches either use separate reinforcement learning agents with templated interactions or optimize for interactivity judged by another LLM, without measuring how much useful information is actually gained. We propose a new approach that quantifies the effectiveness of each interaction by the reduction in the assistant's uncertainty, measured via entropy over recommendations. We apply this entropy reduction as a reward---without relying on ground-truth recommendations, which are often unavailable in real-world scenarios---to fine-tune the LLM, enabling strategic interaction generation. Empirical results with supervised fine-tuning (SFT) and direct preference optimization (DPO) on the INSPIRED and ReDial datasets show that our method improves both recommendation quality and conversational efficiency. --- ### The Null Token Knows: Reducing Message-Free Hallucination in ASR and NMT Authors: Kirill Borodin, Vasiliy Kudryavtsev, Ivan Viakhirev, Grach Mkrtchian Published: 2026-08-16 Categories: cs.CL, cs.LG, cs.SD arXiv: https://arxiv.org/abs/2608.15940v2 Modern encoder-decoder systems can produce fluent text even when their input contains no recoverable message. We study this failure in ASR and NMT through the models' reserved null tokens, asking whether the score for ending generation already carries a usable abstention signal. Across speech recognizers and translation models, we audit native null-token scores and scalar logit shifts. In Whisper, we additionally probe decoder states and compare supervised row edits with conventional external gates. The evaluated models often expose a useful abstention signal, but stock decoding does not reliably act on it. Raising the null-token score can sharply suppress fabrication, but aggressive intervention also deletes valid speech or shortens legitimate translations. These findings turn the null token into a diagnostic lens on hallucination and motivate evaluating abstention methods by both suppression and deletion costs, rather than by hallucination reduction alone. --- ### Aborted but Not Forgotten: KV-Cache Retention Breaks Rollback Consistency in Language Agents Authors: Guijia Zhang, Harry Yang Published: 2026-08-16 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.15939v1 Stateful language agents assume a rejected branch can be taken back by clearing it from the application transcript. We show this breaks when the serving session retains key/value (KV) state across the logical abort: the model can continue attending to content the application believes it discarded. We formalize the missing guarantee as rollback consistency: a complete abort must restore the state the model attends, not just the transcript. The key failure is cross-layer: a correct logical rollback need not compose with retained inference state, and the gap can remain invisible to the application. To isolate cache effects from text effects, we introduce a same-token/different-cache audit that holds decision-step tokens identical while varying only whether the cached prefix is stale or rebuilt from committed state. Across seven open-weight families (3.8B-36B), retained KV alone flips a typed protected effect in 25 of 63 audited cells, while attacker tokens are absent from the served request in all 63; rebuilding the cache closes every cell. The channel reproduces in an end-to-end session application, on the default Hugging Face Transformers cache-reuse path, and under LangGraph time-travel, where verified logical rollback can still leave attended KV stale. Susceptibility varies across models, but the underlying attended-state integrity violation is structural. We rule out position and length confounds, generalize across protected effects, policy structures, and a cache-isolated Mixture-of-Experts model, and show that transaction-local cache restoration closes the channel without requiring a global cache flush. All headline results are deterministic and reproducible from released artifacts. --- ### Token Distribution versus Data Volume: Domain Balancing in Multi-Domain Meeting Summarisation Authors: Ashima Sood, Bryan Gardiner, Joan Condell Published: 2026-08-16 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.15935v2 Jointly fine-tuning an LLM on meeting-summarisation corpora of widely varying size raises a question that prior work leaves confounded: when a domain-balanced training mixture helps, is the gain due to the distribution of tokens across domains, or merely to the volume of data seen? We disentangle these factors by constructing balanced and natural (native-proportional) token mixtures at matched token budgets (2-32M) over five English meeting corpora, fine-tuning Mistral-7B with QLoRA, and evaluating per domain. Balancing redistributes quality, improving the data-scarce minority domains at a low cost to the data-rich ones. The trade favours balancing whenever the minority domains matter: their share under proportional allocation is fixed at 1-2% regardless of budget, so matching balanced quality on those domains requires far more total data. We further find that pruning low-value transcript lines removes ~15% of tokens from the conversational corpora at no measurable cost, and that balancing by tokens is not the same as balancing by examples. Fine-tuning one model per domain is competitive only on the data-rich domains and falls below the zero-shot model on the data-scarce ones. A two-annotator study of 741 judge-labelled facts validates our fact-level evaluation. Together these results give practitioners a basis for deciding when to balance an imbalanced multi-domain mixture, and on what unit. --- ### Differentiable Voxelization of Surface Representations Authors: Tobias Djuren, Ugo Finnendahl, Markus Worchel, Hendrik Meyer, Marc Alexa Published: 2026-08-16 Categories: cs.GR, cs.CV arXiv: https://arxiv.org/abs/2608.15934v1 Different shape representations facilitate different computations. Surface representations, in particular meshes, are often used for modeling, whereas volume representations are useful for spatial queries such as intersection or containment. Optimizing a surface representation based on a volumetric properties by gradient descent requires the derivatives of the volume relative to its bounding surface. We derive this gradient for winding numbers and show that it can be efficiently computed for volumetric values sampled on a regular grid (voxel representation) and surface parameters based on vertex sets (triangle meshes). This enables an efficient solution for a variety of optimization problems. We demonstrate the practical use of this approach at the examples of deforming meshes to resolve intersections, being manufacturable by cutting with a bandsaw from three directions, and creating shapes that are close to tiling 3D space. ---