# Research Papers — 2026-06 · bundle page 239 of 260 > Full-text markdown of papers #11901–11950 of 12993 published in 2026-06, newest first. AI-readable mirrors of the full 12993-paper 2026-06 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-06/page/239 [← Previous bundle](https://aipod.dev/papers/archive/2026-06/page/238.md) [Next bundle →](https://aipod.dev/papers/archive/2026-06/page/240.md) ### Seg2Track++: Probabilistic Track Validation and Data Association for Multi-Object Tracking and Segmentation Authors: Diogo Mendonça, Tiago Barros, Cristiano Premebida, Urbano J. Nunes Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03875v1 Autonomous systems require robust Multi-Object Tracking and Segmentation (MOTS) to operate reliably in dynamic environments, ensuring consistent object identities and precise mask-level delineation. Foundation models such as SAM2 have shown strong zero-shot generalization for segmentation, but their direct application to MOTS is limited by unreliable track association and false-positive propagation. This work introduces Seg2Track++, a framework that integrates instance segmentation with SAM2 and a novel track management module to perform zero-shot MOTS with enhanced temporal consistency. Tracks are associated using Mask Centroid Distance (MCD) and Confidence-Aware Cost Modulation (CCM), while Probabilistic Track Validation (PTV) employs a Bernoulli filter to validate track existence and suppress ghost tracks. Experimental results on KITTI MOTS demonstrate improved identity preservation, reduced false-positive propagation, and robust track management without fine-tuning. --- ### DyaPlex: Full-Duplex Speech-Motion Model for Dyadic Interaction Authors: Koki Nagano, Hongyu Liu, Seonwook Park, Tianye Li, Amrita Mazumdar, Christian Jacobsen, Shengze Wang, Michael Stengel, Rajarshi Roy, Ka Chun Cheung, Simon See, Shalini De Mello Published: 2026-06-02 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2606.03874v1 We present DyaPlex, a streaming, full-duplex speech-and-motion model designed for dyadic interaction. To capture the continuous and reciprocal nature of human communication, this full-duplex capability empowers the agent to simultaneously perceive and generate both speech and physical motion in a streaming fashion. At its core, our method leverages the strong priors of a foundational full-duplex speech model and integrates a novel motion pathway, thereby achieving fully synchronized multi-modal interaction. Specifically, we design a dual-tower Transformer architecture that preserves the zero-shot conversational reasoning of a frozen base speech model while constructing a deeply coupled, streaming motion pathway. By introducing a unified dyadic token interleaving mechanism and guiding cross-attention via a time-aligned speech-motion RoPE, our model effectively aligns autoregressive motions with rich latent speech features. Trained on the 4,000-hour Seamless Interaction dataset, our model effectively captures cross-speaker dependencies and establishes new state-of-the-art performance across both monadic and dyadic human interaction benchmarks. --- ### Visual Instruction Tuning Aligns Modalities through Abstraction Authors: Luis Palacios, Lorenzo Basile, Diego Doimo, Alberto Cazzaniga Published: 2026-06-02 Categories: cs.CV, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.03871v1 Visual instruction tuning effectively adapts a pre-trained Large Language Model (LLM) to process image information alongside text. Yet, it remains unclear how visual features are embedded into the layer-wise hierarchy of abstractions of the LLM backbone. Across a diverse set of vision-language architectures, we show that instruction tuning primarily serves as a bridge, embedding visual features directly into the intermediate semantic layers of the LLM, bypassing the early layers devoted to unimodal processing. With probing analyses and causal interventions, we show that these intermediate layers are the semantic core of vision-language processing and play a critical role in the performance on a broad set of multimodal benchmarks. In addition, by comparing the geometry of semantically equivalent visual and textual representations, we find that fine-tuning extends and strengthens the existing abstraction phase, aligning visual features with pre-existing textual ones. Finally, we confirm the functional role of this localized alignment by restricting fine-tuning to intermediate layers alone: this strategy preserves the performance of full fine-tuning on vision-centric benchmarks while reducing training time. Our results suggest that multimodal integration is a localized phenomenon driven by the repurposing of the internal abstraction engine of the LLM. --- ### Unified Video-Action Joint Denoising for Dexterous Action and Data Generation Authors: Dingrui Wang, YuAn Wang, Jinkun Liu, Yue Zhang, Mattia Piccinini, Yu Sun, Johannes Betz Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03868v1 Recent world action models leverage video foundation models by aligning broad visual-dynamics priors with executable robot actions. We revisit this alignment from a distributional perspective. Existing formulations typically narrow the aligned prior into an observation-conditioned policy distribution over future actions. In contrast, we keep the distribution broader by modeling the joint space of interaction videos and executable hand trajectories under multiple conditioning regimes. We propose Donk, a unified video-action denoising model for dexterous hands. With language, an initial image, and the initial hand state, Donk samples future videos and bimanual MANO trajectories as an action policy. Without the image condition, the same denoising architecture samples paired video-action rollouts from a text-conditioned distribution, turning the aligned video prior into a data engine. Across action, video, and text-only generation evaluations, Donk improves dexterous trajectory accuracy, preserves strong video fidelity, and produces smooth text-conditioned action rollouts under the same unified training recipe. --- ### A Training-Free Mixture-of-Agents Framework for Multi-Document Summarization using LLMs and Knowledge Graphs Authors: Cuong Vuong Tuan, Trang Mai Xuan, Tien-Cuong Nguyen, Vu-Duc Ngo, Thien Van Luong Published: 2026-06-02 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.03867v2 Multi-Document Summarization (MDS) plays a critical role in distilling essential information from collections of textual data. Existing approaches often struggle to capture complex inter-document relationships, rely heavily on large amounts of labeled data for supervised training, or exhibit limited generalization across domains and languages. To address these limitations, we present a training-free mixture-of-agents framework for MDS that leverages the complementary strengths of large language models (LLMs) and knowledge graphs. Our approach decomposes summarization into specialized agent tasks: extractive selection, knowledge-aware abstraction, and iterative refinement, each operating without task-specific fine-tuning. We unify their outputs using a multi-perspective consistency mechanism guided by LLMs. Experiments across four datasets in English and Vietnamese demonstrate state-of-the-art or competitive performance, validating the effectiveness and adaptability of our modular design. --- ### Taiji: Pareto Optimal Policy Optimization with Semantics-IDs Trade-off for Industrial LLM-Enhanced Recommendation Authors: Yuecheng Li, Zeyu Song, Jing Yao, Chi Lu, Peng Jiang, Kun Gai Published: 2026-06-02 Categories: cs.IR, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2606.03866v1 Scaling recommender systems via large language models (LLMs) has become a prominent trend in the industry. However, aligning the LLM's semantic space with the recommender's ID space via post-training (e.g., SFT and RL) remains challenging. Existing LLM4Rec paradigms are bottlenecked by two main issues: (1) the difficulty of measuring and improving chain-of-thought (CoT) quality in open-domain recommendation during SFT, and (2) the neglect of the trade-off between LLM semantic rewards and recommendation preference rewards during RL alignment. Inspired by these challenges, we present Taiji, a novel LLM-as-Enhancer framework designed for industrial recommender systems. To overcome the SFT bottleneck, we utilize reverse-engineered reasoning and open-ended rejection sampling to generate high-quality, domain-specific CoT data. To resolve the RL alignment issue, we propose Pareto Optimal Policy Optimization (POPO), which adaptively adjusts cross-domain reward weights. Theoretically, it achieves an optimal trade-off between the semantic world knowledge of LLMs and the collaborative ID features representing online user preferences. Extensive offline evaluations and online A/B tests validate the effectiveness of Taiji. Deployed on Kuaishou's advertising platform since May 2026, Taiji currently serves over 400 million users daily, yielding significant commercial revenue and demonstrating its robust scalability in web-scale environments. --- ### Explainable Forecasting of Scientific Breakthroughs from Concept Network Dynamics Authors: Thomas Maillart, Thibaut Chataing, Ntorina Antoni, David Dosu, Paul Bagourd, Julian Jang-Jaccard, Alain Mermoud Published: 2026-06-02 Categories: cs.SI, cs.CY, cs.DL, cs.LG, physics.soc-ph arXiv: https://arxiv.org/abs/2606.03864v1 We introduce an explainable machine-learning approach that forecasts the structural precursors of scientific breakthroughs -- the emergence and intensification of links between research concepts -- by modelling how OpenAlex concept networks evolve over time. Using 59 semantic and topological features, a two-stage LightGBM model jointly predicts the formation and the future weight of concept pairs, adding a regression stage that quantifies expected intensity to prior link-existence forecasts. Relative to the state of the art, the approach improves accuracy and explainability at once: comparative validation across four technology and biomedical domains yields ROC-AUC in (0.954, 0.967) at all horizons without re-tuning, exceeding the roughly 0.90 of prior models, while every forecast rests on structural, auditable features rather than opaque embeddings. Classification performance is high (AUC about 0.95) and regression remains stable (RMSLE 0.45 to 0.6 over one to five years). Feature attribution shows that structural factors -- particularly Adamic-Adar similarity and degree-based Hadamard measures -- consistently drive accuracy, suggesting that breakthrough-relevant recombinations emerge in tightly connected sub-networks. Two expert-anchored cases, quantum annealing and AI-enabled quantum architectures, show the model surfacing technological convergence consistent with expert expectations. We then outline a three-layer decision architecture -- detection, expert translation, institutional integration -- that turns these forecasts into evidence-based research strategy and policy, anchored in open data and explainable features. --- ### PyraMathBench: Evaluating and Improving Mathematical Capability in Large Language Models Authors: Zetian Ouyang, Linlin Wang, Gerard de Melo, Liang He Published: 2026-06-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.03858v1 Despite the pivotal role of numerical reasoning as the cornerstone of mathematical capabilities in large language models (LLMs) across applications, few benchmarks evaluate LLMs by integrating numerical processing and mathematical reasoning, hindering the interpretability of failures in math tasks. We introduce PyraMathBench, a comprehensive hierarchical benchmark with 32,505 questions derived from 7,404 math word problems, spanning 4 key cognitive aspects, 14 subcategories, and 2 modalities. Experiments reveal that LLMs' performance is severely compromised by inadequate numerical computation and weak handling of abstract numerical questions. To address this, we propose the Smart Optimization & Learning-based VErsatile module (SOLVE) and Interactive Relative Policy Optimization (IRPO), which enhance LLMs' numerical-mathematical synergy via efficient tool calls (fuzzy matching and low-quality call rejection). Comparative experiments show Qwen-2.5 achieves a 5.0 score improvement with SOLVE and IRPO training. --- ### FLARE: Fine-Grained Diagnostic Feedback for LLM Code Refinement Authors: Yinsheng Yao, Hongxiang Zhang, Weixi Tong, Tianyi Zhang Published: 2026-06-02 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2606.03852v2 Large language models often generate code with bugs. Existing methods rely on feedback signals such as test failures and self-critiques to iteratively refine the generated code. Such signals are either too coarse-grained or too high-level, which is not sufficient to inform the model where to fix the bug. In this work, we present Flare, an iterative framework with a lightweight diagnostic model that predicts line-level suspiciousness signals for bug localization and code refinement. Given the inherent uncertainty of diagnostic predictions, Flare searches over the top-k suspicious regions and selects the best candidate according to execution outcomes. Experiments on LiveCodeBench and BigCodeBench with five base LLMs show that, even without candidate search (k=1), Flare outperforms the strongest baseline with an absolute improvement from 1.72% to 7.42%. Furthermore, searching over 10 candidates yields an average improvement of 8.50% compared with no candidate search. When evaluated in isolation, our lightweight diagnostic model achieves the best performance compared with recent fault localization methods, demonstrating that it can provide reliable fine-grained guidance for code refinement. --- ### FLARE: Fine-Grained Diagnostic Feedback for LLM Code Refinement Authors: Yinsheng Yao, Hongxiang Zhang, Weixi Tong, Tianyi Zhang Published: 2026-06-02 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2606.03852v1 Large language models often generate code with bugs. Existing methods rely on feedback signals such as test failures and self-critiques to iteratively refine the generated code. Such signals are either too coarse-grained or too high-level, which is not sufficient to inform the model where to fix the bug. In this work, we present Flare, an iterative framework with a lightweight diagnostic model that predicts line-level suspiciousness signals for bug localization and code refinement. Given the inherent uncertainty of diagnostic predictions, Flare searches over the top-k suspicious regions and selects the best candidate according to execution outcomes. Experiments on LiveCodeBench and BigCodeBench with five base LLMs show that, even without candidate search (k=1), Flare outperforms the strongest baseline with an absolute improvement from 1.72% to 7.42%. Furthermore, searching over 10 candidates yields an average improvement of 8.50% compared with no candidate search. When evaluated in isolation, our lightweight diagnostic model achieves the best performance compared with recent fault localization methods, demonstrating that it can provide reliable fine-grained guidance for code refinement. --- ### Two-Action Apple Tasting with Switching Costs Authors: Tommaso Cesari, Roberto Colomboni Published: 2026-06-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.03851v1 We study the two-action apple-tasting problem with switching costs against an oblivious adversary. In an equivalent normalized formulation, at each round the learner chooses between a revealing action and a blind action: the revealing action gives reward $0$ and reveals the hidden value $x_t\in(-1,1)$ of the blind action; the blind action gives reward $x_t$ but reveals nothing. The learner pays one unit whenever they switches actions, and regret is measured against the best fixed action in hindsight. General feedback-graph algorithms with switching costs give $\widetilde O(T^{2/3})$ regret guarantees for this problem. The two-action apple-tasting graph was the natural candidate for the missing $Ω(T^{2/3})$ obstruction in the switching-cost classification: such a lower bound would have transferred to a large family of still-unclassified feedback graphs. We prove that this obstruction is not there: the oblivious minimax expected regret for this problem satisfies \( \frac{1}{2\sqrt3}\cdot\sqrt T \le R_T^\star \le 2\sqrt{3}\cdot \sqrt{T}. \) --- ### Clustered Self-Assessment: A Simple yet Effective Method for Uncertainty Quantification in Large Language Models Authors: Qi Cao, Takeshi Kojima, Andrew Gambardella, Helinyi Peng, Yutaka Matsuo, Yusuke Iwasawa Published: 2026-06-02 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2606.03846v1 Large language models (LLMs) demonstrate remarkable performance across diverse tasks, but they often generate responses that appear plausible while being factually incorrect. This problem is compounded by the lack of explicit uncertainty estimates, which makes it difficult for users to judge the reliability of model outputs. Existing uncertainty quantification methods typically rely on indirect signals, such as entropy across sampled generations. These signals can be difficult to interpret and do not fully leverage the model's ability to assess its own uncertainty. We propose a simple yet effective self-assessment method for uncertainty quantification in LLMs. Our approach groups sampled generations into semantically distinct clusters, converts them into answer options in a structured multiple-choice question, and uses the probability assigned by the LLM to each option as a confidence estimate. Experiments across multiple models and datasets show that our method consistently outperforms baseline approaches. Notably, it achieves competitive performance with as few as two additional samples, demonstrating both its effectiveness and efficiency. --- ### Re-Evaluating Continual Learning with Few-Shot Adaptation Authors: Amogh Inamdar, Matthew So, Vici Milenia, Richard Zemel Published: 2026-06-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.03843v1 Continual learning methods aim to maximize the stability and plasticity of machine learning models that are trained on a sequence of tasks. The standard measure of stability (i.e., forgetting) is the 0-shot performance of a model on previously learned tasks, and plasticity, the performance on the most recently learned task. However, 0-shot evaluation does not fully measure a model or method's ability to retain learned information or adapt quickly to new information, as it requires perfect recall across multiple tasks. In this paper, we propose few-shot evaluation as a more comprehensive assessment of the stability and plasticity of a continual learning system. We conduct a fine-grained assessment on task sequences for continual image classification and find that this paradigm produces novel insights into the performance of popular continual learning strategies. Through few-shot evaluation with a novel metric -- per-shot plasticity -- we show that adding `foresight' to continual learning methods via the meta-learning of a short sequence of future tasks induces learning-to-learn behavior over the task sequence. --- ### EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management Authors: Zherui Yang, Fan Liu, Yansong Ning, Hao Liu Published: 2026-06-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.03841v1 Recent progress in Large Language Model (LLM) agents has enabled promising advances in automated data science. However, existing approaches remain fundamentally limited by their static action sets and lack of principled long-horizon context management, hindering their ability to accumulate reusable experience across tasks and operate reliably in multi-stage, iterative data science pipelines. To address these challenges, we introduce EvoDS, a self-evolving autonomous data science agent that learns to expand its skills and adaptively managing long-term context through agentic reinforcement learning. Specifically, EvoDS introduces two key strategies: (1) Autonomous Skill Acquisition (ASA) mechanism, which enables agents to synthesize, validate, and reuse executable skills; and (2) Adaptive Context Compression (ACC) strategy, which treats context management as a learned control problem rather than passive truncation. These strategies are orchestrated within a two-stage multi-agent training scheme, enabling EvoDS to autonomously improve over time. Theoretically, we prove that EvoDS's hierarchical design reduces tool-selection error, and its optimization objective aligns with an information bottleneck principle, ensuring efficient context use. Empirically, EvoDS outperforms state-of-the-art open-source data science agents by an average of 28.9% across four diverse benchmarks while eliminating out-of-token failures. Our code and data are available at https://github.com/usail-hkust/EvoDS. --- ### Text-attributed Graph Condensation via Text Selection and Attribute Matching Authors: Haowei Han, Yuxiang Wang, Guojia Wan, Hao Wang, Shanshan Feng, Hao Huang, Jiawei Jiang, Xiao Yan Published: 2026-06-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.03839v1 Text-Attributed Graph (TAG) is an important type of graph structured data, where each node has a text description. TAG models usually train a Graph Neural Network (GNN) and language model jointly, which leads to high space and time consumption, especially on large datasets. To mitigate this, we propose TAGSAM, a condensation method that compresses TAGs while preserving training accuracy. TAGSAM comes with two key designs, i.e., subgraph text Selection and Attribute similarity Matching, which compress the text description and graph topology of TAG, respectively. For the texts, subgraph text selection selects and merges representative text chunks from multiple related text descriptions by maximizing mutual information. For the graph topology, popular condensation methods based on Matching Training Trajectories (MTT) suffer from high variance, which hinders accuracy. Our attribute similarity matching mitigates this issue by aligning stable similarity matrices. We evaluate TAGSAM against six state-of-the-art baselines, where it showcases superior performance. For the same compressed size, TAGSAM improves upon the best-performing baseline by an average of 4.9% in accuracy. Furthermore, it maintains competitive training accuracy even when the TAG is condensed to just 1% size. Our code is available at https://github.com/SundayVHan/TAGSAM --- ### Where Do We (Not) Need Temporal Context in Low-Resource Video Task Adaptation? Authors: Luc P. J. Sträter, Hazel Doughty Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03837v1 Parameter-efficient fine-tuning (PEFT) and probing enable adaptation of foundation models using only a small number of trainable parameters, making it attractive for video understanding where annotation and computation are expensive. However, video PEFT has focused on adapting image-pretrained models, while standard PEFT methods can also be applied to video representations. These settings are rarely compared and both confine temporal reasoning to a single component of the model, leaving open how temporal context should be distributed across backbone, PEFT and probe. In this work we provide a systematic study of model adaptation strategies for video understanding. We evaluate methods across appearance-focused, motion-focused and spatially dense settings, with a particular focus on scenarios with limited data where parameter-efficiency is most beneficial. Our results provide new insights into PEFT and probing across settings and demonstrate the importance of temporal context allocation for effective video adaptation --- ### Online Learning with Gradient-Variation Interval Regret Authors: Yan-Feng Xie, Shuche Wang, Peng Zhao, Zhi-Hua Zhou Published: 2026-06-02 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2606.03831v1 This paper investigates non-stationary online learning using the metric of interval regret, which requires an online algorithm to perform well over every time interval. We propose the first online learning algorithm that achieves an interval regret bound scaling with gradient variation, a fundamental measure of the cumulative change in online function gradients, which relates to various problem-dependent quantities and is closely connected to stochastic optimization and other problems. Our method employs a simple and efficient two-layer online ensemble structure that achieves strong theoretical guarantees. Specifically, it enjoys a regret bound that simultaneously adapts to various problem-dependent quantities while also preserving the minimax-optimal rate in the worst case. Moreover, recognizing the challenge of hyperparameter tuning, we introduce a Lipschitz- and smoothness-agnostic variant that automatically adapts to these potentially unknown constants. This is primarily enabled by a novel Lipschitz-adaptive meta algorithm, which may be of independent interest. Beyond interval regret, our method also yields broader implications: it provides versatile bounds for interval dynamic regret, a stronger measure that competes with changing comparators over any interval, and yields the first piecewise characterization for stochastic extended adversarial optimization. Theoretical findings are validated by experiments. --- ### BigFinanceBench: A Workflow-Grounded Benchmark for Financial-Research Agents Authors: Alex Wang, Georg Meinhardt, Jacob Katz, Joseph H. Kim, Pratyush K. Chaudhary, Chase Blagden, Eric Xu Published: 2026-06-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.03829v1 Financial-research answers are decision-relevant only when another analyst can audit how they were produced: which source was chosen, which period and accounting definition were used, which assumptions were made, and how the calculation was performed. Existing finance benchmarks largely evaluate isolated subskills or final answers, leaving the auditable derivation itself under-measured. We introduce BigFinanceBench, a 928-item expert-authored benchmark of open-ended financial-research tasks in which each item pairs a ground-truth reference answer with a point-weighted rubric that decomposes the derivation into independently checkable steps. BigFinanceBench is workflow-grounded in that it evaluates the full derivation rather than only the final output. Across 36,241 rubric points, the benchmark supports partial-credit evaluation and localization of failures across the analyst workflow. Evaluating ten current frontier and open-weight agents, we find substantial headroom: the best system reaches only 58.8% rubric score, final-answer accuracy is a useful but lossy proxy for derivation quality, and model capability varies non-uniformly across financial workflows. --- ### Conditional Latent Diffusion Model with Fourier-based Motion Modelling for Virtual Population Synthesis Authors: Shaokun Lan, Haoran Dou, Jinghan Huang, Arezoo Zakeri, Fengming Lin, Zherui Zhou, Jinming Duan, Alejandro F. Frangi Published: 2026-06-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.03827v2 In-silico trials of medical devices require the generation of virtual populations of anatomies. In cardiovascular applications, virtual anatomy is typically represented as a 3D+t mesh sampled from a generative model. However, most existing mesh generators focus on static anatomy, while sequence models often lack explicit periodicity. To this end, we propose 4D F-MeshLDM, a conditional generative framework comprising a convolutional mesh VAE to encode meshes, a structural latent space that parameterises motion using a truncated Fourier series, and a diffusion prior that learns the latent distribution over Fourier coefficient tokens. By conditioning the diffusion process on clinical covariates via affine modulation, we enable controllable synthesis. Sampling tokens and performing inverse Fourier synthesis yield cycle-consistent latent trajectories, which can be decoded into 3D+t cardiac mesh sequences. Experiments on 5,000 UK Biobank subjects demonstrate that 4D F-MeshLDM outperforms state-of-the-art baselines in anatomical fidelity and achieves near-zero cycle closure error. Furthermore, the generated cohorts accurately preserve clinical functional indices, highlighting the potential of our framework for reliable in-silico cardiac trials. --- ### Dynamic Short Convolutions Improve Transformers Authors: Oliver Sieberling, Bharat Runwal, Rameswar Panda, Yoon Kim Published: 2026-06-02 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2606.03825v1 Transformers have become the dominant architecture for large language models, largely due to the scalability and flexibility of attention, feed-forward layers, residual connections, and normalization. This paper introduces dynamic short convolutions as an additional neural network primitive for improving Transformers. Unlike static short convolutions, dynamic convolutions use input-dependent filters, which preserves the locality bias of convolution while increasing expressivity. Motivating experiments show that applying dynamic short convolutions to key, query, and value representations improves performance on challenging associative recall tasks compared with static convolutional variants. Across language-modeling experiments ranging from 150M to 2B parameters, dynamic convolutions consistently outperform standard Transformers and Transformers augmented with static short convolutions. Fitting scaling laws indicates a 1.33$\times$ compute advantage over compute-matched Transformers when dynamic convolutions are applied to the key, query, and value vectors, and a 1.60$\times$ advantage when adding dynamic convolutions after every linear layer. Dynamic convolutions also offer improvements on linear RNNs (Mamba-2/Gated DeltaNet) and mixture-of-experts architectures. We make these gains practical with custom Triton kernels that enable efficient training with a manageable end-to-end slowdown. These results suggest that dynamic short convolutions are a scalable, hardware-efficient, and expressive primitive for advancing Transformer-based language models. --- ### Calibrating Urban Traffic Simulation from Sparse Road Observations via Genetic Optimization Authors: Hunter Sawyer, Jesse Roberts, Simon Matei Published: 2026-06-02 Categories: cs.AI, cs.CY, cs.NE arXiv: https://arxiv.org/abs/2606.03823v1 Urban traffic simulation is a critical tool for infrastructure planning, including the placement of electric vehicle charging stations. However, realistic traffic simulation across many cities is hindered by two fundamental data limitations: detailed real-world traffic measurements are available for only a small fraction of road segments in most cities, and employment distribution data critical for modeling commuter traffic is rarely available at the resolution needed for simulation. This paper presents a genetic algorithm-based framework that directly addresses both limitations, calibrating urban traffic simulations from sparse road observations without requiring detailed job location data. Using the SUMO traffic simulation platform for Greensboro, North Carolina, our approach optimizes job distributions and gate-traffic parameters to align simulated traffic with a small sample of roads with known traffic-flow rates. We demonstrate that this approach produces simulated traffic that correlates well with real-world measurements, generalizes to road segments withheld from training, and produces job distributions that show promising qualitative agreement with census employment data despite never directly training on that employment data. This work demonstrates that realistic urban traffic simulation can be achieved from minimal real-world observations, offering a scalable and data-light approach to simulation calibration that reduces the barrier to deploying traffic models across diverse cities. --- ### Finding Needles in the Haystack: Transductive Active Labeling in Ecology Authors: Rupa Kurinchi-Vendhan, Sara Beery Published: 2026-06-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.03821v2 Active learning is now standard practice in labeling ecological data, enabling ecologists to quickly process large volumes of field data to understand and monitor natural environments. Current practices evaluate active learning inductively, estimating predictive performance on a held-out test set. We argue that this evaluation is misaligned with most ecological tasks, where the goal is to transductively label an entire pool of data as efficiently as possible. We demonstrate that ignoring the human-in-the-loop underestimates the importance of continuing to label, particularly for classes in the long tail which may be of disproportionate ecological importance (rare species, uncommon behaviors, etc.). Our analysis shows that, for this long tail, the transductive objective shifts importance from prediction to discovery: the true challenge becomes finding "needles in the haystack," examples of rare classes that are embedded within dense regions of abundant classes in the latent geometry, which we quantify with a novel metric of sampling difficulty. Finally, to translate these insights to practical ecological workflows, we propose a conservative hybrid stopping criterion inspired by ecological rarefaction curves, and show that combining predictive performance with discovery criteria reduces premature stopping on long-tailed pools, improving rare-class recovery when discovery, not classification, is the limiting factor. --- ### A Quantitative Approximation Framework for Flow Distillation in Diffusion Models Authors: Weiguo Gao, Ming Li, Lei Shi, Hanfei Zhou Published: 2026-06-02 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2606.03820v2 We develop a quantitative framework for diffusion distillation by viewing few step sampling as approximation through compositions of learned flow maps. For trajectory distillation of the probability flow ODE, we show that low noise multimodal regimes separate score approximability from dynamical stability: the score remains efficiently approximable, while small local errors may be strongly amplified by stiff flow dynamics. In a Gaussian mixture Ornstein--Uhlenbeck model, we prove time uniform \(L^p(p_t)\) score approximation by ReLU and ReQU networks with explicit polylogarithmic complexity, and derive a computable Lipschitz bound \(L(t)\) for the flow velocity. The stability factor \(\exp\bigl(\int_s^t L(u)\mathrm du\bigr)\) can grow exponentially as noise decreases and mixture separation increases. Comparing this certificate with a certified local Lipschitz budget for one step students identifies regimes of direct distillation difficulty, without implying an approximation lower bound. We also show that deep residual compositions control global transport error through propagated local errors, and that equalizing cumulative stability yields an optimal nonuniform segmentation. With eight segments, this grid reduces final mean relative MSE by up to \(51.9\%\) versus uniform grids. --- ### A Quantitative Approximation Framework for Flow Distillation in Diffusion Models Authors: Weiguo Gao, Ming Li, Lei Shi, Hanfei Zhou Published: 2026-06-02 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2606.03820v1 We develop a quantitative approximation framework for diffusion distillation, viewing few-step sampling as error propagation under compositions of learned flow maps. Focusing on trajectory distillation for the probability-flow ODE, we show that local approximation errors can be strongly amplified in low-noise multimodal regimes, where the underlying dynamics become stiff. In an analytically tractable Gaussian-mixture Ornstein--Uhlenbeck setting, we separate two core difficulties: approximating the time-dependent score field and controlling the dynamical amplification governed by the time-integrated Jacobian bound of the probability-flow ODE. On the approximation side, we prove constructive L^p(p_t) guarantees showing that ReLU--ReQU networks approximate the Gaussian-mixture score uniformly over time, with depth and width scaling polylogarithmically in the target accuracy and explicitly with the mixture geometry. On the stability side, we derive an explicit bound L(t) for the spatial Lipschitz constant of the probability-flow velocity and convert it into a flow map stability estimate governed by \int_s^t L(u)\,du, making late-time amplification in stiff regimes computable. Building on these estimates, we prove that deep residual compositions efficiently approximate the long-horizon transport, with global error controlled by the stability amplification factor, and identify a Lipschitz-mismatch regime in which one-step distillation is structurally unfavorable. The resulting theory yields a stability-balanced non-uniform time grid obtained by uniform partitioning in the cumulative stability coordinate. Experiments support the prediction and reduce end-to-end relative MSE by up to 51.9\% with 8 segments compared with uniform grids. --- ### TreeFlash: Parallel AR-Approximation for Faster Speculative Decoding Authors: Peer Rheinboldt, Frédéric Berdoz, Roger Wattenhofer Published: 2026-06-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.03819v1 One-shot block drafters for speculative decoding generate the full draft in a single forward pass, achieving strong throughput by eliminating sequential token generation. However, they predict each draft token conditioned only on the prefix context, with no dependence on previously drafted tokens. This non-autoregressive conditioning causes the drafter's distribution to diverge from the verifier's true autoregressive distribution as draft depth grows. This problem becomes more severe in tree-based drafting, where distinct branches are forced to share the same marginal distribution for subsequent tokens. We propose TreeFlash, which addresses this by incorporating an MLP layer conditioned on the drafter's hidden state and the previous token to approximate an autoregressive distribution. TreeFlash retains the $\mathcal{O}(1)$ decoding time complexity of one-shot drafters by employing a two-stage approximation mechanism. TreeFlash achieves state-of-the-art performance across a variety of tasks and models, improving over marginal tree drafting by $12\%$ higher block efficiency and $9\%$ higher speedup. --- ### Rethinking the Idiomaticity Decomposability Hypothesis: Evidence from Distributional Learning Authors: Maggie Mi, Golzar Atefi, Atsuki Yamaguchi, Felix Gers, Aline Villavicencio, Nafise Sadat Moosavi Published: 2026-06-02 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.03817v1 Idioms can be analysed in terms of their decomposability, the extent to which constituent meanings contribute to the figurative whole. Decomposability is thought to predict syntactic flexibility. Usage-based accounts instead attribute idiom behaviour to distributional experience, such as speaker familiarity and predictability. We examine these views using contextualised language models as controlled distributional learners. We propose a model-internal measure of decomposability and relate it to human ratings, syntactic flexibility, and predictability while tracking idiom learning during pretraining. Model-derived decomposability correlates weakly with human judgments and shows a small but consistent negative relationship with syntactic flexibility. Pretraining analyses show that stabilisation of idiom representations in models is not explained by frequency alone. Instead, surprisal, decomposability, and frequency all contribute, with decomposability showing the strongest training-dependent effect. --- ### Leveraging BART to Assess CS1 C++ Programming Assignments using Rubric-based Criteria Authors: Kelsey Rainey, Jesse Roberts Published: 2026-06-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.03814v1 This paper investigates rubric-aware, multitask fine-tuning of transformer models for automated grading of introductory C++ programming assignments, with the goal of producing grade predictions that better reflect instructor grading behavior than general-purpose LLMs. Using multi-semester CS1 data, student submissions are paired with numeric scores, letter-grade buckets, and assignment rubrics, then preprocessed into unified sequences for transformer input. A BART encoder-decoder with LoRA adaptation is trained to jointly predict numeric grades and grade buckets, augmented with a distribution-matching term to align predicted and empirical grade distributions, an evaluation dimension often overlooked in prior work. Experiments compare single-task and multitask training, hard one-hot versus fuzzy and boundary-based soft labels, and rubric versus no-rubric conditions, with additional T5 and pairwise-pretrained variants. Results show that multitask BART with boundary-based soft labels and rubric context achieves lower mean absolute error and stronger grade-distribution alignment than single-task, hard-label, or code-only baselines. Fully fine-tuned T5 further improves distributional fidelity, while pairwise pretraining reduces numeric error at the cost of minority-class sensitivity. Collectively, the findings suggest that calibration-aware, rubric-guided training produces more instructor-like grading behavior than accuracy-optimized alternatives. --- ### Enhancing Operational Safety via Agentic Dialogue Hazard Identification Analysis Authors: Sanjay Das, Ran Elgedawy, Ethan Seefried, Ryan Burchfield, Tirthankar Ghosal Published: 2026-06-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.03812v1 Operational safety in high-stakes domains such as industrial process control, autonomous, and safety-critical systems, demand reliable hazard identification. While large language models (LLMs) have shown promise in automating safety analysis tasks, single-turn, monolithic inference is brittle: it lacks the self-correction, deliberation, and contextual refinement that safety engineers apply iteratively. In this paper, we introduce HAZDIAL, a framework that investigates whether structured agentic dialogue-multi-agent, multi-turn interactions improves the quality of NLP- based hazard identification over single-pass baselines. We systematically compare two dialogue modalities: adversarial debate and constructive discussion, and propose an algorithm-based agentic interaction optimization. We evaluate all configurations against a curated golden dataset using standard classification metrics (accuracy, precision, recall, F1) and novel dialogue metrics. This work advances the intersection of dialogue systems, multi-agent reasoning, and AI safety, providing an empirical evidence for dialogue-driven hazard analysis. --- ### AI Agents Enable Adaptive Computer Worms Authors: Jonas Guan, Tom Blanchard, Hanna Foerster, Hengrui Jia, Gabriel Huang, Nicolas Papernot Published: 2026-06-02 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2606.03811v1 A computer worm is malware that spreads on a network by replicating itself from one machine to another. Traditional worms, like WannaCry, exploited predetermined vulnerabilities, and their spread can be halted by patching those vulnerabilities. Here we show that artificial intelligence (AI) agents enable a fundamentally new threat: a worm that generates tailored attack strategies to each target it encounters. The worm parasitically uses compromised machines to run open-weight large language models (LLMs) to sustain its reasoning, or extend its reach for further attacks. Deployed on a network of machines spanning Linux, Windows, and IoT (Internet of Things) devices, the worm propagated by exploiting common, real-world corporate network vulnerabilities. Since the worm is powered by stolen compute, the attacker's marginal cost per new infection is zero. This creates a destabilizing economic asymmetry between attackers and defenders. Moreover, because the worm requires no commercial AI platform, centralized safety controls, such as service refusals or rate limiting, are structurally irrelevant. Our results demonstrate that self-sustaining AI-driven cyber-threats are no longer theoretical. We must prepare for autonomous generative adversaries: malware systems that propagate without human operators and are defined not by fixed exploit code, but by the capacity to reason about targets, adapt to observations, and synthesize attack logic in real time. --- ### Consistency Training Can Entrench Misalignment Authors: David Demitri Africa, Arathi Mani Published: 2026-06-02 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.03810v2 Consistency training encourages a model to produce similar outputs across related inputs or sampling procedures. Such methods are simple, scalable, and largely label-free, but their effects on model alignment remain poorly understood. Could the self-bootstrapping nature of these methods amplify undesired behavior in models? We test seven consistency training methods on 108 model organisms: open-source models (7B--70B) fine-tuned to exhibit various forms of controlled misaligned behavior. We find that outcomes vary significantly: consistency training generally suppresses reward hacking and emergent misalignment but amplifies sycophancy. We present evidence that distribution shifts induced by the consistency labeling process, rather than variation in the selection operators, may be the primary driver of systematic alignment effects. Finally, we present a unifying theoretical framework to derive conditions under which consistency training will amplify or suppress misalignment. In total, our study establishes that consistency training is not alignment-neutral, and that its use in critical systems should be carefully audited. --- ### PURGE: Projected Unlearning via Retain-Guided Erasure Authors: Vedant Jawandhia, Daksh Ahuja, Ghufran Alam Siddiqui, Prashant Trivedi, Yash Sinha, Pratik Narang Published: 2026-06-02 Categories: cs.LG, cs.AI, cs.CR arXiv: https://arxiv.org/abs/2606.03808v1 We propose PURGE, a machine unlearning algorithm built on a simple but an under-exploited observation: continual learning (CL) and machine unlearning (MU) which are fundamentally dual problems. CL tries to learn new tasks without forgetting old ones; MU tries to erase specific data without hurting retained performance representing the same underlying tension in opposite directions. PURGE leverages this duality by adapting gradient projection from A-GEM (Chaudhry et al., 2019) so that every unlearning step is constrained to not increase the retain-set loss. On top of this, it performs multi-layer representation erasure, pushing forget-set activations in intermediate layers towards the retain distribution to remove information from hidden representations rather than just suppressing it at the output. A key design choice is the retain-confusion target: rather than pushing forget outputs toward the uniform distribution, which we found to be surprisingly easy for membership inference attacks to detect, we instead target the model's natural confusion pattern on retain data. This makes the unlearned model hard to distinguish from one retrained from scratch. Two self-regulating stopping criteria (a retain-loss budget and a forget-accuracy target) let the algorithm decide on its own when to stop, removing the need for manual epoch tuning. In experiments on five datasets (CIFAR-10, MNIST, SVHN, STL10, PathMNIST) across 22 class-level forgetting tasks, PURGE consistently keeps retain accuracy above 96% while achieving MIA AUROC close to 0.5 (the ideal), outperforming gradient ascent, KL-uniform, and several published baselines on the privacy-utility frontier. --- ### TeX-1500: A Paired Real-World LWIR Hyperspectral Dataset and Benchmark for Temperature-Emissivity-Texture Decomposition Authors: Cheng Dai, Jiale Lin, Hongyi Xu, Bingxuan Song, Ziyang Xie, Fanglin Bao Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03806v1 Temperature-emissivity-texture (TeX) decomposition seeks to recover object heat state, material spectral response, and visible-like geometric texture from long-wave infrared hyperspectral imaging (LWIR HSI). Existing TeX pipelines are mainly scene-specific inverse solvers, and the lack of paired LWIR HSI-TeX supervision has limited learning-based decomposition. To address this gap, we introduce TeX-1500, a large-scale paired LWIR HSI-TeX dataset and benchmark for supervised HSI-to-TeX decomposition. TeX-1500 contains 1,522 calibrated real-scene pairs from DARPA Invisible Headlights (DARPA IH) pushbroom imagery and our FTIR acquisitions, covering five locations, four seasons, diverse acquisition times, heterogeneous wavelength layouts, and two sensor families. Each sample stores a calibrated valid-band radiance cube, calibrated wavelength positions, and aligned temperature, emissivity, and texture supervision constructed through a consistent restoration and TeX-construction protocol. We further provide TeX-UNet, a simple wavelength-aware baseline that maps calibrated HSI bands and wavelength positions to TeX fields. Experiments on the held-out DARPA IH pushbroom scenes and zero-/few-shot transfer to FTIR scenes show that TeX-1500 provides usable paired supervision and a measurable benchmark for data-driven physical-property-centered thermal perception. --- ### Easy-to-Use Shielding for Reinforcement Learning Authors: Stefan Pranger, Bettina Könighofer Published: 2026-06-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.03804v1 Safe exploration is a key challenge in Reinforcement Learning (RL) that aims to prevent agents from making harmful decisions while exploring their environment. Safe exploration is a key challenge in Reinforcement Learning (RL) that aims to prevent agents from making harmful decisions while exploring their environment. Shielding is one such technique that assumes domain knowledge in the form of an environment model to decide upon action safety. Although well-established, shielding has seen limited adoption in RL due to the lack of accessible end-to-end infrastructure connecting formal shield synthesis with standard RL frameworks. Applying shielding typically requires expertise in formal methods and substantial engineering effort, keeping it outside the typical RL workflow. We address this by extending our shield synthesis tool Tempest into a practical backend for safe RL. Our core contribution is tempestpy, a Python library that integrates Tempest-based shield synthesis directly into the Gymnasium API, allowing shields to be synthesized and deployed within existing RL pipelines. This lowers the barrier to entry for shielding and turns formal safe-exploration methods into a usable component for RL practitioners. We also extend Tempest's algorithmic support to compute sound shields for stochastic multiplayer games, preserving formal safety guarantees. We demonstrate the resulting workflow end to end and evaluate shielded and unshielded RL across multiple environments. To facilitate modeling, we provide symbolic models for MiniGrid and introduce MiniGridSafe, a collection of playground environments designed to make shielding easily accessible and experimentally transparent. MiniGridSafe extends MiniGrid with safety-oriented scenarios featuring probabilistic transitions and additional agents, enabling the study of challenging safety aspects in a simple and intuitive setting. --- ### LiveBand: Live Accompaniment Generation in the Audio Domain Authors: Marco Pasini, Javier Nistal, Ben Hayes, Mathias Rose Bjare, Stefan Lattner, George Fazekas Published: 2026-06-02 Categories: cs.SD, cs.AI, eess.AS arXiv: https://arxiv.org/abs/2606.03803v2 We present LiveBand, a real-time system that generates high-fidelity music accompaniments to live audio input, respecting strict causal constraints. Our method trains a causal transformer generator in the continuous latent space of a pre-trained causal audio autoencoder, using adversarial sequence-level supervision from a discriminator. At each timestep, the generator receives only the causally available mix context and Gaussian noise, and predicts accompaniment latents without access to future mix frames or ground-truth target latents. Training is performed in a single parallel forward pass under causal masking, while streaming inference proceeds autoregressively with a rolling attention state. The model's training and inference computations are matched by design, eliminating teacher forcing and the associated exposure bias. On a multi-instrument music accompaniment benchmark, LiveBand improves over prior work on objective measures of audio quality, beat alignment, and mix adherence, while enabling real-time streaming generation without lookahead into the future on consumer hardware. --- ### Template Collapse and Information-Theoretic Limits in Camera rPPG Pulse Morphology Restoration Authors: Achraf Ben Ahmed Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03802v1 Objective: Consumer face camera remote photoplethysmography (rPPG) enables passive cardiovascular monitoring, but whether single-cycle waveform morphology encoding arterial stiffness biomarkers is recoverable from this measurement has not been characterised. Methods: We evaluated 16 architectures spanning six families on 153 subjects across three datasets, introducing cross-subject Pearson r to distinguish subject-specific recovery from template collapse. Results: No architecture recovered subject-specific morphology (cross-subject r range 0.773--0.9999; ground-truth ceiling 0.601). Supervised Contrastive (SupCon) converged to log N = 4.844, constituting the strongest available empirical evidence that no discriminative morphological structure is extractable from single-cycle rPPG by the encoder families tested. The VAE decoder restores population-level harmonic content absent from the rPPG input (H2/H1: 0.310 output vs. 0.275 input), generalising zero-shot to UBFC (r = +0.708); a directional hallucination gap (p = 0.150) suggests partial signal reading. Anti-collapse objectives fail when input carries no discriminative structure. Significance: Consumer cameras cannot encode individual arterial morphology; cross-subject r is a necessary collapse diagnostic for waveform reconstruction benchmarks. --- ### Trading Human Curation for Synthetic Augmentation in RLVR Authors: Akshansh, Leonardo Rosa Rodrigues, Michael Korostelev, Youssef Hassan, Mark E. Whiting Published: 2026-06-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.03800v2 The supply of high-quality training tasks is a central bottleneck for reinforcement learning from verifiable rewards (RLVR) on agentic language models. Each task requires a sandboxed setup, a prompt, and a hand-authored reward function, and only tasks that pass a quality bar produce useful training signal. Hand-curation at this quality bar does not scale economically to the task counts effective RL training requires, and the substitution rate between automatically generated task variants and human-authored ones is not yet established. We investigate using pre-specified, gate-filtered augmentations of a small hand-authored base as a substitute for additional human curation during RLVR. We formalize the cost-adjusted trade rate $ρ_{\text{cost}}$ between augmented and human-authored tasks, measure it through a controlled ablation across training corpora with varying augmentation share, and characterize the end-to-end economics of the augmentation pipeline. Substituting augmented content for additional human-authored tasks retains aggregate held-out generalization on a ten-benchmark suite spanning code, instruction following, reasoning, and multi-turn agentic function-calling. The cost-adjusted trade rate $ρ_{\text{cost}}$ between gated synthetic and human-authored RLVR tasks stays in $(1.4\times, 11.6\times)$ across the plausible $c_{\text{human}}/c_{\text{aug}}$ range. --- ### Signed Spiking Neuron Enabled by an Orthogonal-Easy-Axis Magnetic Tunnel Junction Authors: Huannan Zheng, Jingli Liu, Kezhou Yang Published: 2026-06-02 Categories: cs.NE, cs.AI arXiv: https://arxiv.org/abs/2606.03796v1 Signed spiking neurons carry richer information than standard spiking neurons. This work proposes a compact magnetic tunnel junction (MTJ)-based neuron for signed leaky integrate-and-fire (LIF) operation. With orthogonal easy axes in the free and pinned layers, the device enables bipolar spike generation and maps magnetic-moment dynamics to signed LIF membrane-potential evolution. Landau--Lifshitz--Gilbert simulations show that proper free-layer dimensions allow the device response to follow a signed LIF equation. A representative design of 10 nm x 45 nm x 50 nm corresponds to an aspect ratio of about 2:9:10. Network evaluations using the fitted device-neuron model achieve 91.06% on CIFAR-10 and 77.40% on CIFAR10-DVS, retaining most of the accuracy of ideal signed LIF neurons. --- ### Beyond Compression: Quantifying Spectral Accessibility in Vision Representations Authors: Akayou A. Kitessa, Yijun Zhao Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03795v1 Vision-language models map visual features into a shared embedding space through learned projection layers, yet it remains unclear how these transformations alter the structure of visual information. This study examines changes in representation through spatial-frequency accessibility, measured by the linear recoverability of band-limited Fourier energy from model representations. To isolate effects beyond dimensionality reduction, we introduce Residual Spectral Loss (RSL), which evaluates changes relative to a dimension-matched random projection baseline. To reduce confounding effects from optimization, the analysis uses pretrained models with all parameters frozen. The experimental results show consistent frequency-dependent changes in accessibility across CLIP and DINOv2 on ImageNet and MS-COCO datasets. Spectral accessibility follows a non-monotonic trajectory across depth, peaking at intermediate layers before decreasing toward the output representation. The final transformation differs across architectures: CLIP's learned projection is spectrally neutral, with changes explained by compression, whereas DINOv2's (CLS) pooling induces a structured loss across the spectrum. These findings identify intermediate layers and pooling mechanisms as primary drivers of spectral transformation in modern vision encoders. --- ### Limit Analysis of Graph Neural Networks with Wireless Conflict Graphs Authors: Romina Garcia Camargo, Zhiyang Wang, Alejandro Ribeiro Published: 2026-06-02 Categories: cs.LG, eess.SP arXiv: https://arxiv.org/abs/2606.03794v1 Graph Neural Networks (GNNs) have emerged as a powerful tool for wireless resource allocation that leverages the underlying graph structure of communication networks. Their transferability property enables models trained on small-scale graphs to generalize to large-scale deployments with little performance deterioration, a desirable property for currently growing networks. Wireless networks are sparse regimes, where a single node is connected to a small number of other users. This work establishes theoretical results for transferability of GNNs over graphs derived from sparse Random Geometric Graphs (RGGs). In particular, we focus on conflict graphs of RGGs used to model interference among links. Our approach considers the closeness between RGGs and Deterministic Grid Graphs (DGG) to establish bounds in the performance loss when a model is transferred across scales. We validate our theoretical findings through the problem of link scheduling, demonstrating that our learned policies consistently outperform existing benchmarks at scale. Finally, we examine the impact of our theoretical assumptions on empirical performance. --- ### Exploring Adversarial Robustness and Safety Alignment in Multilingual Multi-Modal Large Language Models Authors: Hashmat Shadab Malik, Muzammal Naseer, Salman Khan Published: 2026-06-02 Categories: cs.CL, cs.CV arXiv: https://arxiv.org/abs/2606.03793v2 Multimodal Large Language Models integrate visual perception into language reasoning, introducing a continuous attack surface susceptible to adversarial attacks. Prior work on MLLM robustness has focused largely on English-centric tasks, leaving multilingual behaviour unexplored. We address this gap through a systematic study of adversarial robustness and multimodal safety across 12 diverse languages, evaluating open-source MLLMs that acquire multilingual capability through instruction tuning. Gradient-based attacks reveal a transferable multilingual vulnerability: adversarial images optimized in one language continue to induce failure in others, demonstrating strong cross-lingual transferability. Multilingual safety further varies with how effectively a model retrieves or interprets harmful instructions. When harmful intent is issued through text, languages with stronger linguistic grounding more often elicit misuse-enabling responses, while weaker languages produce fewer unsafe outputs. When embedded in the image as typographic content, English scripts are reliably recognised and followed, whereas non-English scripts are rarely parsed by the vision encoder. Lower-resource languages may therefore appear safer, but this is an artefact of comprehension and visual-grounding failures rather than genuine alignment, a phenomenon we term safety-by-failure. In contrast, MLLMs that build multilingual capability throughout their training stages rather than only at instruction tuning, such as Qwen3-VL, exhibit genuine cross-lingual safety, maintaining active refusal across languages rather than masking comprehension failure. Shallow multilingual adaptation, such as fine-tuning on translated instruction data, may produce surface-level understanding that creates illusory safety in low-resource languages; deeper integration across training stages leads to genuine multilingual safety alignment. --- ### Exploring Adversarial Robustness and Safety Alignment in Multilingual Multi-Modal Large Language Models Authors: Hashmat Shadab Malik, Muzammal Naseer, Salman Khan Published: 2026-06-02 Categories: cs.CL, cs.CV arXiv: https://arxiv.org/abs/2606.03793v1 Multimodal Large Language Models integrate visual perception into language reasoning, introducing a continuous attack surface susceptible to adversarial attacks. Prior work on MLLM robustness has focused largely on English-centric tasks, leaving multilingual behaviour unexplored. We address this gap through a systematic study of adversarial robustness and multimodal safety across 12 diverse languages, evaluating open-source MLLMs that acquire multilingual capability through instruction tuning. Gradient-based attacks reveal a transferable multilingual vulnerability: adversarial images optimized in one language continue to induce failure in others, demonstrating strong cross-lingual transferability. Multilingual safety further varies with how effectively a model retrieves or interprets harmful instructions. When harmful intent is issued through text, languages with stronger linguistic grounding more often elicit misuse-enabling responses, while weaker languages produce fewer unsafe outputs. When embedded in the image as typographic content, English scripts are reliably recognised and followed, whereas non-English scripts are rarely parsed by the vision encoder. Lower-resource languages may therefore appear safer, but this is an artefact of comprehension and visual-grounding failures rather than genuine alignment, a phenomenon we term safety-by-failure. In contrast, MLLMs that build multilingual capability throughout their training stages rather than only at instruction tuning, such as Qwen3-VL, exhibit genuine cross-lingual safety, maintaining active refusal across languages rather than masking comprehension failure. Shallow multilingual adaptation, such as fine-tuning on translated instruction data, may produce surface-level understanding that creates illusory safety in low-resource languages; deeper integration across training stages leads to genuine multilingual safety alignment. --- ### Training-Free Multi-Concept LoRA Composition with Prompt-Aware Weighting Authors: Georgios Tsoumplekas, Stella Bounareli, Vasileios Argyriou Published: 2026-06-02 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2606.03792v1 Low-Rank Adaptation (LoRA) successfully enables personalization in text-to-image generation by adapting pre-trained diffusion models to specific visual concepts and styles. However, extending such models to multi-concept customization remains challenging. Naively combining multiple LoRA weights or their outputs often leads to interference among concepts, resulting in degraded visual quality and reduced fidelity to the reference images of individual concepts. This paper proposes a simple yet effective approach for multi-concept customization by optimally combining the outputs of multiple LoRA modules. We leverage the relative importance of each concept during generation, as inferred from its corresponding prompt tokens and introduce two methods, W-Switch and W-Composite, that employ a prompt-aware importance weighting strategy in which each LoRA is weighted according to the semantic influence of its trigger words in the target prompt. In addition, we extend existing quantitative evaluation metrics by proposing a new image-based similarity evaluation framework that assesses image fidelity and identity preservation through comparisons between real-world reference images and automatically segmented concept regions from generated images. We evaluate our approach on the ComposLoRA testbed and demonstrate consistent improvements over existing state-of-the-art methods in terms of visual quality, identity preservation and compositionality. Qualitative evaluations, including a Large Language Model (LLM) based assessment and a user study, further validate the effectiveness of the proposed methods and align with the newly introduced quantitative image-based metrics. Our code is available at https://github.com/GeorgeTsoumplekas/Prompt-Aware-Multi-LoRA-Composition. --- ### SLU-2K: A Question-Based Benchmark for Semantic Evaluation of Sign Language Translation Authors: Zeno Testa, Antonino Furnari, Lorenzo Baraldi, Natalia Díaz-Rodríguez Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03788v2 Sign Language Translation (SLT) is typically evaluated with surface-form metrics such as BLEU and ROUGE, which reward lexical overlap but do not directly measure whether a translation preserves the meaning of the source sign sequence. This is in contrast with the final objective of integrating SLT in assistive technology. In this work, we shift the focus from Sign Language Translation (SLT) to Sign Language Understanding (SLU), with particular emphasis on semantic understanding. Specifically, we evaluate systems based on their ability to correctly recover, from the input video, key semantic aspects of the original sentence, such as actions taking place and facts about people and objects. To enable this evaluation systematically, we propose SLU-2K, a dataset of 2,350 closed-ended video question-answer pairs based on the popular PHOENIX-2014T and CSL-Daily datasets. To obtain SLU-2K, we propose and extensively evaluate an automated data generation pipeline which produces questions across 7 categories, namely actions, locations, numbers, objects, people, time, and weather conditions. We show the potential of SLU-2K by evaluating popular Multimodal Large Language Models (MLLMs) and two representative state-of-the-art systems, MMSTL and SpaMo. Our results show that MLLMs reach near-random performance, highlighting the need for a more systematic integration of SLU in current AI systems. Furthermore, state-of-the-art translation systems carefully fine-tuned on in-domain data still exhibit a substantial semantic gap, with results ranging from 56.7% to 75.2%. These findings suggest that current SLT evaluation protocols overestimate true understanding and that future progress should be measured not only by fluency and n-gram overlap, but also by semantic correctness. Code, prompts, and benchmark files are available at https://github.com/ZenoTsT/SLU-2K --- ### Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs Authors: Lisa Bouger, Théo Lasnier, Philippe Loubet Moundi, Yannick Teglia, Djamé Seddah Published: 2026-06-02 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.03785v2 Backdoor attacks in Large Language Models (LLMs) are a growing security concern, where models can generate adversary-chosen content. Existing defenses target backdoors one at a time and typically require knowledge of the trigger, leaving the defender at a structural disadvantage when unknown backdoors may exist in a model. We show that backdoor neutralization through unlearning generalizes across backdoors: training a model to ignore a single trigger can also suppress other backdoors that were never explicitly targeted. We study this phenomenon across three model families, whose backdoors were injected via pretraining or continual pretraining, by analyzing the models obtained after removing one backdoor at a time. To understand why unlearning certain backdoors induces the suppression of others, we introduce the Cross Activation Shift Distance, to quantify the distance between model changes induced by different trainings. Our results open a new direction for LLM safety as defenders could deliberately inject controlled backdoors and then remove them, leveraging cross-backdoor transfer to also suppress unknown backdoors that an attacker may have previously introduced in the model. --- ### Reasoning over Grammar: Can Synthetic Linguistic Reasoning Traces Enhance Low-Resource Machine Translation? Authors: Renhao Pei, Yihong Liu, Sampo Pyysalo, Hinrich Schütze, Shaoxiong Ji Published: 2026-06-02 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.03782v2 Large language models (LLMs) offer a promising approach to machine translation (MT) for extremely low-resource languages by incorporating linguistic resources through in-context learning. However, LLMs often struggle to apply grammatical information effectively during translation. Inspired by recent progress in chain-of-thought reasoning, we investigate whether low-resource MT can benefit from structured intermediate steps of linguistic analysis and grammatical reasoning. We propose a pipeline for automatically generating step-by-step linguistic reasoning traces from Universal Dependencies treebanks, dictionaries, and grammar-rule banks. We evaluate these traces in three settings: in-context learning (ICL), supervised fine-tuning (SFT), and reinforcement fine-tuning (RFT), on Xibe and Chintang as test cases. Our results show that linguistic reasoning traces are most effective as inference-time guidance: in ICL, reliable linguistic reasoning traces substantially improve translation performance across models, languages, and metrics. In contrast, using these traces as training data yields smaller and less consistent gains, as models learn the format but often generate error-prone content. These findings suggest that LLMs can leverage grammatical information for low-resource MT when given reliable linguistic analyses, while learning to generate such analyses remains a major bottleneck. --- ### Reasoning over Grammar: Can Synthetic Linguistic Reasoning Traces Enhance Low-Resource Machine Translation? Authors: Renhao Pei, Yihong Liu, Sampo Pyysalo, Hinrich Schütze, Shaoxiong Ji Published: 2026-06-02 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.03782v1 Large language models (LLMs) offer a promising approach to machine translation (MT) for extremely low-resource languages by incorporating linguistic resources through in-context learning. However, LLMs often struggle to apply grammatical information effectively during translation. Inspired by recent progress in chain-of-thought reasoning, we investigate whether low-resource MT can benefit from structured intermediate steps of linguistic analysis and grammatical reasoning. We propose a pipeline for automatically generating step-by-step linguistic reasoning traces from Universal Dependencies treebanks, dictionaries, and grammar-rule banks. We evaluate these traces in three settings: in-context learning (ICL), supervised fine-tuning (SFT), and reinforcement fine-tuning (RFT), on Xibe and Chintang as test cases. Our results show that linguistic reasoning traces are most effective as inference-time guidance: in ICL, reliable sentence-specific traces substantially improve translation performance across most models, languages, and metrics. In contrast, using the linguistic reasoning traces as training data yields smaller and less consistent gains, as models learn the trace format but often generate erroneous content. These findings suggest that LLMs can leverage grammatical information for low-resource MT when given reliable linguistic analyses, while learning to generate such analyses remains a major bottleneck. --- ### Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models Authors: Yuetian Lu, Ali Modarressi, Yihong Liu, Hinrich Schütze Published: 2026-06-02 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.03780v2 Activation patching can identify a mixture-of-experts (MoE) block whose clean output restores a corrupted factual prediction. However, because the block output combines contributions from multiple routed experts, block-level rescue does not establish whether the recovery localizes to an individual expert or depends on the routed expert set. We study this question on single-token COUNTERFACT contrasts by corrupting subject-token embeddings, restoring clean block outputs, and then restoring clean-minus-noised expert updates under fixed routing. In Qwen3-30B-A3B-Base, a discovery sweep selects layer 44, and held-out analysis identifies L44E069 as a recurrent routed contributor with positive specificity over same-layer active experts. Its effect is fact-matched and improves true-token probability and rank, which explains part of the layer rescue. In Mixtral-8x7B-v0.1, the selected recurrent singleton is not specific; matched-size controls instead show specificity of the clean-routed top-2 set. These findings show that successful MoE-block restoration does not necessarily imply localization to a single expert. --- ### Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models Authors: Yuetian Lu, Ali Modarressi, Yihong Liu, Hinrich Schütze Published: 2026-06-02 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.03780v1 Causal tracing of factual recall has been studied predominantly in dense transformer language models, where interventions localize information flow to layers or feed-forward modules. Sparse mixture-of-experts (MoE) language models introduce a sharper question: when a factual prediction is mediated by a routed MoE block, which routed expert contributions matter? We formulate expert-aware causal tracing for sparse MoE language models. Using CounterFact facts, we first corrupt the model's factual preference by adding noise to subject-token embeddings, and then test whether clean MoE-block outputs or clean expert-level updates restore the true-vs-foil logit contrast. For Qwen3-30B-A3B-Base, a layer sweep selects and validates layer 44, and expert-level tracing identifies L44E069 as an expert repeatedly selected in the clean run whose held-out patch outperforms other active same-layer expert patches. For Mixtral-8x7B-v0.1, layer-level tracing validates a mid-layer signal, but the signal is not localized to the selected singleton expert; a coalition check instead recovers it with routed multi-expert updates. These results suggest that MoE factual tracing can be made expert-aware, while also showing that expert-level localization is model- and protocol-dependent rather than universal. --- ### From Control Boundary to Insurance Claim: Reconstructing AI-Mediated Losses Through the CER Framework Authors: Alex Leung, Rex Zhang, Kentaroh Toyoda, SiewMei Loh Published: 2026-06-02 Categories: cs.AI, cs.CR, q-fin.RM arXiv: https://arxiv.org/abs/2606.03777v1 AI losses that arise through an insured organization's generative or agentic AI system require state reconstruction, not merely event reconstruction, because the relevant state changes as the system reasons, retrieves, calls tools, and acts. The relevant question is not only what loss occurred, but what the system was allowed to do, what it actually did, and whether that reconstructed loss can support insurance claim recovery. This paper addresses losses in which the insured's AI system is in the causal chain, including externally triggered failures such as prompt injection, retrieval-augmented generation (RAG) poisoning, malicious tool output, credential misuse, and data poisoning. Specifically, this paper introduces CER, a use-case-level diagnostic for AI residual risk transfer. C (control boundary) asks whether the system had an enforceable operating envelope. E (evidence reconstruction) asks whether the system state and causal chain can be reconstructed from retained artifacts. R (insurance response) asks whether the reconstructed loss is insured: whether insurance coverage is available in the market and placed for the insured, together with the proof needed to support insurance claim recovery. The paper makes three contributions: it defines the AI-specific reconstruction problem, operationalizes that problem through CER, and specifies claim-grade evidence for AI reconstruction. Public examples include the reported PocketOS and Replit agentic database-deletion incidents and Moffatt v. Air Canada as an adjudicated output/reliance case. Keywords: AI systems; CER framework; residual risk transfer; agentic AI; generative AI; AI insurance; evidence reconstruction. --- ### AmbientEye: A Dataset for Pupil Segmentation under Natural Ambient Infrared Illumination Authors: Mingyu Han, Hyunyoung Han, Nitheekulawatn Thommakoon, Gangtae Park, Jieun Han, Xucong Zhang, Ian Oakley Published: 2026-06-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.03774v1 Eye tracking is essential for smart glasses, as it provides insight into user attention for ambient intelligence applications. However, most existing eye-tracking systems rely on active infrared (IR) illumination, creating practical barriers to all-day outdoor use due to power consumption. In this paper, we investigate whether passive IR cameras alone, without any active IR light source, can enable reliable pupil detection in unconstrained outdoor environments, where ambient sunlight serves as the sole illumination source. To support this investigation, we introduce AmbientEye, a large-scale dataset of 2,606,225 eye images collected from 35 participants from 19 countries. It is captured outdoors under natural sunlight with two off-axis camera configurations and two sun-orientation conditions. We provide high-quality pupil annotation through SAM2 automatic segmentation, followed by refinement by human annotators. We benchmark a state-of-the-art pupil segmentation algorithm on our dataset and compare its performance with that on existing datasets under controlled IR illumination. Results reveal a substantial drop in pupil segmentation performance from 0.928 on controlled IR datasets to 0.767 on AmbientEye. This performance gap highlights the challenge of the ambient-light setting. This positions AmbientEye as a first benchmark for an unexplored and highly practical eye-tracking scenario. ---