# Research Papers — 2026-06 · bundle page 75 of 260 > Full-text markdown of papers #3701–3750 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/75 [← Previous bundle](https://aipod.dev/papers/archive/2026-06/page/74.md) [Next bundle →](https://aipod.dev/papers/archive/2026-06/page/76.md) ### Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies Authors: Ruixiao Lin, Xinhao Deng, Qingming Li, Jianan Ma, Yunhao Feng, Yuqi Qing, Zhenyuan Li, Yechao Zhang, Shiwen Cui, Changhua Meng, Tianwei Zhang, Xingjun Ma et al. Published: 2026-06-22 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2606.23075v1 Self-evolving LLM agent systems, which autonomously update their model parameters, memory, tools, and architectures, introduce a qualitatively new threat landscape in which adversarial influences become permanently encoded, self-amplify across generations, and propagate through populations without sustained attacker access. We present a systematic security and privacy analysis organized around the Module-Lifecycle Attack Surface (MLAS) matrix, which decomposes the attack surface into five functional modules (Brain, Cognitive Resource, Execution, Self-Design, Collective) $\times$ five lifecycle stages (Bootstrap, Propose, Evaluate, Commit, Serve). Analysis of the resulting 25 cells reveals that 17 face critical threats for which no effective partial mitigation. We identify seven cross-cutting amplification effects that interact synergistically and cannot be addressed by securing individual modules in isolation. Comparative case studies of two open-source frameworks demonstrate that evolution-native design activates $3.5\times$ more attack surface cells and achieves a 100% attack persistence rate (40/40 payloads across all CIA+Privacy categories), while co-located security scanners block only 2.5% of attacks. Our findings establish that self-evolution converts every known attack category from session-bounded to lineage-persistent, gives rise to entirely new attack classes, and renders static defenses structurally inadequate, motivating evolution-aware security frameworks and formal verification for self-modifying systems. --- ### Rethinking Prototype-based Similarity Learning for Few-Shot Object Detection Authors: KunHo Heo, Seungjae Kim, Wongyu Lee, SuYeon Kim, MyeongAh Cho Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23069v3 Few-shot object detection aims to detect novel object categories from only a few labeled examples, avoiding costly large-scale annotation. Recent prototype-based similarity learning approaches enable training-free adaptation by matching query features with class prototypes. However, they suffer from two fundamental limitations: (i) class confusion arising from inter-class similarity margin collapse, and (ii) insufficient visual cues for precise localization, as similarity scores capture only class-level semantic affinity while providing limited spatial information. To address these issues, we introduce two complementary components. Text-Anchored Semantic Mask (TSMa) leverages class-level text features as semantic anchors to identify semantically aligned channels through channel-wise interaction between visual and text features. By suppressing style-induced spurious responses and emphasizing class-intrinsic signals, TSMa enlarges inter-class similarity margins and mitigates class confusion. We further propose Stage-Aligned Hierarchical Autoregressive Regression (SHARe), which reformulates localization as a hierarchical autoregressive process that progressively refines bounding boxes across multiple stages. SHARe leverages the layer-wise characteristics of ViT representations by aligning feature abstraction levels with regression stages: deeper layers guide early coarse localization, while shallower layers rich in edge and texture cues refine spatial details in later stages. Experiments on COCO demonstrate a new state of the art, outperforming the previous best by +10.1 nAP, with extensive analysis validating each component. The code is available at https://github.com/VisualScienceLab-KHU/ReSet. --- ### Attention-Spectrum Regularization for Replay-Free Continual Multimodal LLMs Authors: Chuangxin Zhao, Canran Xiao, Siyuan Ma, Mengyao Lyu, Yanbiao Ma, Jun Xia, Guiguang Ding, Yang Liu Published: 2026-06-22 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.23063v1 Multimodal large language models (MLLMs) are increasingly required to adapt to non-stationary streams of visual domains, question types, and user instructions, yet continual fine-tuning often causes severe forgetting of previously acquired multimodal skills. Existing continual vision-language methods mainly preserve outputs, replay data or pseudo-data, regularize embedding geometry, or allocate task-specific parameters, but they provide limited control over how internal cross-modal attention patterns supporting old skills drift during adaptation. We propose Attention-Spectrum Regularization (ASR), a replay-free continual learning framework that preserves skill-conditioned structures of cross-modal attention. ASR treats cross-attention maps as two-dimensional signals, summarizes their scale and directional properties into compact spectral statistics, and stores only skill-wise prototype distributions instead of replaying past image-question pairs, generated pseudo-examples, or old-stage teacher snapshots. In later stages, a phase-invariant spectral regularizer constrains harmful drift of these prototypes while allowing instance-level attention to adapt to new tasks. We provide theoretical analysis showing that skill-conditioned spectral drift controls forgetting under a spectral sufficiency assumption, and that Fourier power spectra are stable to spatial translations and bounded perturbations. Experiments on continual VQA and multimodal instruction-tuning benchmarks, including VQA v2, VQACL, CLT-VQA, CoIN, and UCIT, show that ASR consistently improves final performance and reduces forgetting over strong replay-, regularization-, and adapter-based baselines. Preserving skill-level attention structure is an effective and lightweight mechanism for continual MLLMs. Code is available at https://github.com/Creative-zcx/attention-spectrum-replay --- ### VolHuMe: a High-Resolution Large Scale Dataset of Volumetric Human Meshes Authors: Giulia Martinelli, Niccolò Bisagno, Nicola Garau, Esa Rahtu, Nicola Conci Published: 2026-06-22 Categories: cs.GR, cs.CV arXiv: https://arxiv.org/abs/2606.23062v1 We introduce VolHuMe, a dataset of high-quality 4D human scans captured with a state-of-the-art volumetric studio using 64 RGB and 32 depth cameras. VolHuMe contains individual captures of 104 subjects and provides extensive ground truth, including SMPL-X, high-resolution meshes, multi-view RGB/depth images, rigged meshes, point clouds, garment segmentation, and detailed hand and facial geometry. Unlike prior datasets that primarily rely on full-body imagery, VolHuMe uses a close-range, high-resolution capture setup that preserves fine-grained body-part details, improving geometric fidelity and texture resolution. We benchmark VolHuMe on state-of-the-art methods across 3D and 4D human reconstruction tasks, showcasing the dataset's quality and exposing the limitations of current evaluation testbeds. --- ### MotionHalluc: Diagnosing Kinematic Hallucinations in Fine-Grained Motion Reasoning Authors: Weile Guo, Shenghong He, Danying Mo, Chengdong Xu, Xuexun Liu, Chao Yu Published: 2026-06-22 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.23061v1 Motion instruction generation in cross-video comparison aims to produce corrective feedback that describes the differences between a query and a reference motion. However, existing models often generate instructions that exhibit motion hallucinations, failing to reflect actual kinematic differences between paired videos. To systematically investigate these hallucinations, we introduce MotionHalluc, a dedicated benchmark for evaluating motion hallucinations in paired-video comparison. MotionHalluc comprises 1540 fine-grained questions over 553 video pairs, evaluating hallucinations along three core dimensions: (1)directional hallucination, (2)attributional hallucination, and (3)temporal hallucination. Extensive evaluations of state-of-the-art large multimodal models demonstrate high susceptibility to these hallucinations. Furthermore, we provide Perceive-Parse-Verify (PPV) as a training-free measurements extraction and verification baseline that converts candidate instructions into executable measurement queries and supplies kinematic measurements at inference time. Our results show that this simple measurements injection yields an average 10.6% performance gain across models, suggesting that motion reasoning with explicit quantitative measurements is a key factor in reducing hallucinations in cross-video comparison. Our code and dataset will be made publicly available upon acceptance. --- ### From Text Metrics to Model Internals: A Study of Whisper ASR Hallucination Detection Authors: Jan Jasiński, Mateusz Barański, Julitta Bartolewska, Marcin Witkowski, Konrad Kowalczyk Published: 2026-06-22 Categories: cs.SD, cs.AI, eess.AS arXiv: https://arxiv.org/abs/2606.23060v1 Hallucinations of ASR models - fluent transcriptions with no basis in audio - degrade system performance and pose risks in downstream applications. Robust detection of such errors remains a challenge. This paper studies Whisper large v3 hallucination detection on real-speech human-annotated data across three paradigms: text-based, LLM-based, and internal decoder state probing. Text classifiers utilizing metrics for text evaluation achieve high recall but degrade without reference transcripts. LLM-based detection improves precision with domain-specific prompt conditioning, yet remains less competitive than the lightweight text-based methods. Probing Whisper's decoder representations, without a ground-truth reference, yields the strongest performance, revealing that hallucination traits are encoded across intermediate decoding layers. A late-fusion meta-classifier combining text and internal-state outputs achieves the best overall detection performance. --- ### Three-Step Hierarchical Transformer for Multi-Pedestrian Trajectory Prediction Authors: Raphaël Delécluse, Hazem Wannous, Laurent Grisoni, Laurent Guimas Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23058v1 Pedestrian trajectory prediction requires modeling temporal dynamics, multimodal cues, and social interactions in crowded environments. Existing methods often address these factors separately or entangle them in costly attention blocks, limiting scalability, flexibility, and interpretability. We propose a three-step hierarchical Transformer that explicitly separates temporal encoding, multimodal fusion, and scene-level interaction reasoning. Lightweight GRU summaries enable efficient cross-modal attention, while social attention over time--agent tokens captures inter-pedestrian influences at manageable cost. Experiments on JTA, JRDB, and the Pedestrians and Cyclists in Road Traffic dataset show state-of-the-art performance on real-world datasets (JRDB, Urban) and competitive results on JTA. Ablation and qualitative analyses confirm the contribution of each stage and the model's ability to anticipate complex behaviors such as early turning. --- ### Who Owns the AI Recommendation? A Multi-Industry Empirical Map of Brand Category Ownership Across Large Language Models Authors: Dmitrij Żatuchin Published: 2026-06-22 Categories: cs.IR, cs.CL, cs.CY, cs.LG arXiv: https://arxiv.org/abs/2606.23057v2 This exploratory study measures brand inclusion across five industries, 50 brands and 250 queries, each put five times to GPT-5.2, Gemini 3 Flash and Perplexity sonar-pro in February and September 2026 (3,614 and 3,750 scored answers). Category Inclusion Rate, Recommendation Share, Competitive Vacuum Index and Co-Mention Asymmetry have stated denominators. February inclusion rates sit close together across an industry's sampled brands (mean Gini 0.30), while at least one brand is named in 80% or more of answers to 204 of 250 queries. Vacuums occur in 7.6% of queries; provisional open-vocabulary model readings suggest most reflect the sampled brand list. The partially pre-specified September replication shows strong cross-date Recommendation Share correlation (Spearman 0.994), unchanged vacuum prevalence and agreement of 60.8% against February's 57.2%. The descriptive size association persists. Fixed margins do not account for all co-mention structure: 31 ordered pairs depart from the September null. Agreement exceeds the query-independent null, and both parametric-pair advantage intervals are positive. All ten clustering seeds yield zero emergent clusters. Matcher validation remains provisional pending author checks. The statistics describe system output and identify no causal mechanism. --- ### Who Owns the AI Recommendation? A Multi-Industry Empirical Map of Brand Category Ownership Across Large Language Models Authors: Dmitrij Żatuchin Published: 2026-06-22 Categories: cs.IR, cs.CL, cs.CY, cs.LG arXiv: https://arxiv.org/abs/2606.23057v1 Large language models now mediate how buyers discover products and services, making the competitive structure of AI-generated recommendations a strategic concern for brands. A basic question has lacked large-scale empirical answers: in a given category, which brand does a model recommend, and how concentrated is that ownership? Across 3,750 responses spanning 50 brands, five industries, and 250 brand-free category queries on three models (GPT-5.2, Google Gemini 3 Flash, and Perplexity sonar-pro), each query repeated five times under a dice-roll stability protocol, we propose three exploratory metrics: the Category Ownership Index (COI), a brand's share of mentions within a category; the Competitive Vacuum Index (CVI), flagging categories with no single leader; and the Displacement Score (DS), quantifying asymmetric substitution between brand pairs. In this sample, recommendation concentration was moderate: the mean Gini coefficient was 0.28 (95% CI (0.16, 0.41)), below the 0.60 power-law threshold we set. Competitive vacuums were rare, appearing in 8.0% of queries, so the models named at least one sampled brand in most cases. Cross-model agreement on the top-recommended brand was 41.6%: a top position on one model did not reliably hold on another. Displacement was industry-dependent, from co-recommendation in consulting (0.4:1) to one-directional substitution up to 4.3:1, with an unweighted mean of 2.4:1 across the five industries. A BERTopic check placed only 4.2% of discovered topic clusters outside the original categories. Within the scope studied, these results sit in tension with a strong winner-takes-all narrative around AI recommendation, and the three metrics offer a candidate, reproducible procedure for competitive-intelligence analysis that future work can validate. --- ### Some Results about the Expressivity of Preference-Incomplete Structured Argumentation Frameworks Authors: Antonio Yuste-Ginel Published: 2026-06-22 Categories: cs.AI, cs.LO arXiv: https://arxiv.org/abs/2606.23055v1 This paper studies the expressive power of ASPIC$^+$ argumentation frameworks with uncertain preference profiles by comparing them with several abstract formalisms with uncertain defeats. Most of our results are negative (and some of them are theoretically unexpected). We also conjecture a positive, non-trivial threshold for the expressivity of uncertain preferences, and prove some essential preliminary steps toward the confirmation of this conjecture. --- ### Physics-governed executable modelling of triboelectric nanogenerators Authors: Hongfa Zhao, Baiqiao Wang, Tiancong Zhao, Chun Jin, Hanlin Zhou, Mingrui Shu, Minyi Xu, Liwei Lin, Wenbo Ding, Zhong Lin Wang Published: 2026-06-22 Categories: physics.app-ph, cond-mat.mtrl-sci, cs.AI arXiv: https://arxiv.org/abs/2606.23051v1 Predictive modelling of triboelectric nanogenerators (TENGs) remains fragmented across analytical theories, finite-geometry solvers and disconnected simulation workflows. These disparate approaches must be unified into an executable framework to advance quantitative TENG research.Here we introduce a charge-defined modelling framework and implement it as TENG-CLAW, a physics-governed platform for traceable TENG simulation. The framework establishes a self-consistent electrostatic hierarchy in which triboelectric charges, pre-charging charges and compensating electrode charges serve as defining state variables.This hierarchy connects the infinite plate analytical limit for near-uniform fields with finite-geometry numerical formulations required for edge-dominated devices. Built on this basis, TENG-CLAW converts user-defined research requests into physically admissible simulation tasks, so that generated outputs are tied to explicit charge states, boundary conditions, solver routes and reusable artifacts across spatial, temporal, field-level, comparative and reporting workflows. This work establishes a rigorous computational basis for interpreting TENG mechanisms and provides reproducible research infrastructure for simulation and physics-guided device design. --- ### Unlimited OCR Works Authors: Youyang Yin, Huanhuan Liu, YY, Qunyi Xie, Chaorun Liu, Shiqi Yang, Shaohua Wang, Zhanlong Liu, Hao Zou, Jinyue Chen, Shu Wei, Jingjing Wu et al. Published: 2026-06-22 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2606.23050v1 Recently, end-to-end OCR models, exemplified by DeepSeek OCR, have once again thrust OCR into the spotlight. A widely held view is that employing a large language model (LLM) as the decoder allows the model to leverage the prior distribution of language, leading to improved OCR performance. However, the downside is equally evident: as the output sequence lengthens, the accumulated KV cache drives up memory consumption and progressively slows down generation. This stands in stark contrast to humans, who exhibit no such decline in efficiency during long-horizon copying tasks. In this technical report, we propose Unlimited OCR, a model designed to emulate human parsing working memory. Taking DeepSeek OCR as the baseline, we replace all attention layers in the decoder with our proposed Reference Sliding Window Attention (R-SWA), which reduces attention computation costs while maintaining a constant KV cache throughout the entire decoding process. By combining the high compression rate of DeepSeek OCR's encoder with our constant KV cache design, Unlimited OCR can transcribe dozens of pages of documents in a single forward pass under a standard maximum length of 32K. More importantly, R-SWA is a general-purpose parsing attention mechanism - beyond OCR, it is equally applicable to tasks such as ASR, translation, etc. Codes and model weights are publicly available at http://github.com/baidu/Unlimited-OCR. --- ### PhoneBuddy: Training Open Models for Agentic Phone Use Authors: Zhengyang Tang, Xin Lai, Pengyuan Lyu, Xinyuan Wang, Tianyi Bai, Chenxin Li, Yiduo Guo, Huawen Shen, Yuxuan Liu, Junyi Li, Zhengyao Fang, Yang Ding et al. Published: 2026-06-22 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.23049v2 Phones are becoming an important execution surface for general-purpose agents, but training open models for reliable phone use remains difficult because the environment that matters at deployment, real devices running real apps, is slow, stateful, side-effectful, and hard to reset or verify, while scalable mock environments only approximate real behavior. We present PhoneBuddy, a training recipe and open-model line for agentic phone use that combines a real-app environment with a mock-app environment, PhoneWorld, which reconstructs runnable mock apps from real GUI usage structure. PhoneBuddy first builds a shared supervised fine-tuning stage from trajectories collected in both environments, then compares real-app RL against mixed RL across both environments. Across a 150-task human evaluation on real phones spanning apps, mini-apps, and cross-app workflows, task success rate improves from 36.67\% after supervised fine-tuning to 40.67\% after real-app RL and 45.33\% after mixed RL. On AndroidWorld, the same progression rises from 60.3\% to 77.2\% to 83.2\%. These results show that mock-app training is not a replacement for real-app RL, but a complementary source of scalable, resettable, and automatically checked interaction. The gains are strongest on app and mini-app tasks, while long-horizontal cross-app workflows remain an important open challenge. --- ### HALAS: A Human-Annotated Dataset of Hallucinations of Modern ASR Systems Authors: Mateusz Barański, Jan Jasiński, Julitta Bartolewska, Marcin Witkowski, Konrad Kowalczyk Published: 2026-06-22 Categories: cs.SD, cs.AI, eess.AS arXiv: https://arxiv.org/abs/2606.23048v1 End-to-end Automatic Speech Recognition (ASR) systems hallucinate on natural speech, yet existing mitigation methods are typically evaluated on non-speech or artificially corrupted audio. We introduce HALAS, the first human-annotated dataset of naturally occurring hallucinations from seven state-of-the-art ASR models on real unprocessed earnings call recordings. HALAS provides span-level labels, enabling analysis of hallucination patterns and their severity. Our analysis reveals strong cross-model vocabulary overlap and confirms that hallucinations also occur for almost correctly transcribed speech (characterized by a low Word Error Rate). The proposed benchmark with HALAS shows that the character and semantic-level metrics used as a proxy for hallucination detection reach 81% ROC-AUC, while state-of-the-art detection methods achieve an F1 score of only 53.1%. As such, HALAS establishes the first rigorous non-artificial benchmark for the detection and mitigation of ASR hallucinations. --- ### Domain Adaptation Under Wireless Network Constraints: When Does It Become Green? Authors: Illyyne Saffar, Aurélie Boisbunon, Shruti Bothe Published: 2026-06-22 Categories: stat.ML, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2606.23047v1 The deployment of data-driven models in 6G wireless networks is increasingly challenged by frequent distribution shifts that degrade performance over time. Unsupervised Domain Adaptation (UDA) offers an alternative approach by adapting the trained model to a shifted domain without requiring labels. However, UDA pipelines are often more complex than single-task training due to additional modules and optimization procedures, raising a practical question: do the benefits of adaptation come at a higher energy cost, and how does this trade-off compare to retraining when labeling effort is also considered? In this work, we investigate the energy consumption of UDA and compare it to single task. We further propose a way to determine the minimum number of target domains for which UDA becomes more energy-efficient than retraining, taking into account the labeling cost. Our results aim to clarify when UDA should be preferred over classical train-from-scratch approaches from an energy and labeling-aware perspective. --- ### UECP: Uncertainty-Enhanced Collaborative Perception Authors: Kang Yang, Tianci Bu, Peng Wang, Deying Li, Wen Jie, Yongcai Wang Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23046v1 Collaborative perception serves as a pivotal solution to enhance the perception capability of individual agents in autonomous driving, where a core challenge lies in seeking reliable evidence to quantify and weight the contribution of each participating agent. Existing methods typically rely on a confidence map, which is co-trained with the detection head, but it is inherently correlated with the detection results and thus fails to provide unbiased physical evidence. Furthermore, how to deeply integrate evidence into the cooperative fusion process remains an open question. To address these issues, this paper first proposes an uncertainty map, a physically grounded and unambiguous metric for evaluating perception quality. This map is directly supervised by real-time sensor signals, i.e., LiDAR point density, ensuring decoupling from detection noise and thereby providing physical scenario-aware evidence for weighting agent contribution. Based on this map, we develop the Uncertainty-Enhanced Collaborative Perception (UECP) framework, centered on the Uncertainty-Aware Pyramid Fusion (UAPF) module. UAPF uses a coarse-to-fine strategy, with two key components: Uncertainty-Weighted Downsampling (UWD) for high-fidelity feature preservation, and Uncertainty-Guided Residual Fusion (UGRF) to reinforce ego features, suppressing noise and ensuring robust fusion. Extensive experiments on real-world datasets show UECP outperforms state-of-the-art methods in effectiveness and robustness by embedding the uncertainty map into fusion. Code will be publicly available. --- ### Prime Fourier Embeddings: A Principled Basis for Modular Arithmetic Authors: Hyunsang Hwang, Suhyun Bae, Donghun Lee Published: 2026-06-22 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.23044v2 Numbers have algebraic structure that standard neural embeddings often fail to expose. We introduce Prime Fourier Embeddings (PFE), which encode integers as prime-indexed (cos, sin) pairs derived from the harmonic analysis of Q, providing a pre-structured representation in which modular arithmetic reduces to selecting the relevant prime channel rather than discovering algebraic structure from scratch. We prove that any linear map equivariant with respect to the product group action on PFE must be block-diagonal with one independent block per prime -- a consequence of Schur's lemma applied to the resulting character decomposition. For square-free composite moduli, the Chinese Remainder Theorem predicts which prime channels are task-relevant. Both predictions are confirmed empirically: ablation studies show specialization ratios exceeding 500x between task-relevant and task-irrelevant channels, with perfect in-distribution test accuracy across all square-free composite moduli tested. --- ### The Model as One Rater Among Several: Measuring Political Positions in Data-Sparse Regions with a Language-Model Panel Authors: Tarek Gara Published: 2026-06-22 Categories: cs.CY, cs.AI, cs.CL, stat.AP arXiv: https://arxiv.org/abs/2606.23042v1 Most tools for measuring political positions, manifesto coding, expert surveys, text-scaling models, were built and validated on Western party systems, and outside that setting they work poorly, and often not at all. This paper is an attempt at a method for those settings. It treats a large language model not as a measurement device but as a single, fallible rater in a panel, roughly the way an expert survey treats one expert: the value comes from pooling many judges rather than trusting any one of them. I describe the panel, an applicability rule that keeps a score of zero distinct from a blank, and a lens system that separates what an actor says from what it does. I report three results. First, holding a definition-free round fixed, adding written axis definitions moves scores by a mean of 1.8 points on a 21-point scale and tightens agreement between raters (mean absolute gap 2.81 to 2.50; r 0.81 to 0.89); they make two independent raters agree more closely, which an arbitrary steer would not. Second, across nine models from eight laboratories in two countries, Krippendorff's alpha is 0.86 on both an interval and an ordinal metric, and it stayed put as the panel grew from five raters to nine. That is reliability, the reproducibility of a reading, and not validity, its correctness. Third, where the panel does disagree, the disagreement is informative: the sharpest split, a full-scale divergence on an actor's stance toward its state's foundational order, points to a referent problem, and a blind triple-coding puts about two-thirds of it down to interpretation rather than error. I try to be plain about what the method can't do, including the human validation it still lacks, and I release the instrument and data in full. The worked example is the Middle East and North Africa, but I'd expect the method to carry to any region these standard tools leave out. --- ### SPAR: Semantic-Pixel Self-Alignment and Adaptive Routing for Unified Multimodal Models Authors: Hongxiang Li, Hongxu Chen, Chenyang Zhu, Xiaoshuang Huang, Jiayin Cai, Xiaolong Jiang, Yao Hu, Long Chen Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23041v2 Multimodal Large Language Models (MLLMs) have achieved remarkable success in visual understanding but remain constrained in visual generation due to the fundamental feature discrepancy between semantic perception and pixel-level reconstruction. Bridging this gap requires overcoming two core challenges: endowing semantic encoders with high-fidelity reconstruction capabilities, and effectively aligning generative models with semantic spaces without relying on external teachers. To this end, we propose a novel unified multimodal framework featuring \textbf{S}emantic-\textbf{P}ixel self-alignment and \textbf{A}daptive \textbf{R}outing (\textbf{SPAR}). First, to reconcile semantic perception with pixel-level reconstruction, we introduce an asymmetric dual-stream unified tokenizer. A lightweight semantic stream anchors discriminative features, while a Transformer-augmented pixel stream recovers fine-grained visual details into a unified compact latent space. Second, to eliminate external dependencies, we propose a self-aligned generation paradigm that natively leverages this optimized tokenizer as an internal alignment teacher for the diffusion model. Furthermore, to facilitate flexible multimodal interaction within this unified space, we introduce Dynamic Token Routing, which enables each token to adaptively aggregate multi-layer MLLM features based on its distinct semantic demands. Extensive experiments demonstrate that SPAR establishes the state-of-the-art for unified architectures, achieving exceptional generation and reconstruction quality while preserving foundational visual understanding capabilities. --- ### EvoRubrics: Dynamic Rubrics as Rewards via Adversarial Co-Evolution for LLM Reinforcement Learning Authors: Hongxin Ding, Baixiang Huang, Yue Fang, Weibin Liao, Zheng Li, Jinyang Zhang, Zhijing Wu, Junfeng Zhao, Yasha Wang Published: 2026-06-22 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.23038v1 Rubric-based rewards offer interpretable and fine-grained optimization signals for reinforcement learning in open-ended tasks where verifiable answers are unavailable. However, pre-constructed rubrics remain static throughout training, creating a fundamental mismatch with the evolving policy: fixed criteria gradually lose discriminative power as the model improves, leading to reward saturation and potential hacking. Recent dynamic rubric methods partially address this but rely on external frontier models or ground-truth answers, and update rubrics only at coarse granularity. We propose EvoRubrics, a co-evolutionary RL framework where a Policy LLM and a Rubric Generator jointly improve through adversarial interaction within each training step. As the policy improves under the rubric generator's guidance, the rubric generator adapts its criteria to remain discriminative and informative, enabling evaluation to track the policy in real time and naturally inducing an automatic curriculum. Experiments show that EvoRubrics consistently outperforms static and dynamic rubric baselines across benchmarks. The learned Rubric Generator further generalizes as a transferable reward model. Notably, even a fully self-supervised variant without any external supervision achieves meaningful gains, suggesting that co-evolution between generation and evaluation alone can provide sufficiently rich learning signals. Our code is publicly available at https://anonymous.4open.science/r/EvoRubrics-2155/. --- ### IPO Finance Agent: Benchmark of LLM Financial Analysts Beyond Finance Agent v2, with Automated Rubric Generation, on the SpaceX (SPCX) IPO Authors: Mostapha Benhenda Published: 2026-06-22 Categories: cs.AI, q-fin.GN arXiv: https://arxiv.org/abs/2606.23032v3 Finance Agent v2 (by Vals AI) has emerged as the reference benchmark for evaluating both Anthropic Claude and OpenAI ChatGPT frontier language models on financial tasks. However, it narrowly deals with periodic reporting from publicly traded companies (SEC 10-K and 10-Q filings), and its agentic harness relies on naive, unenriched chunk retrieval. Neither the task design nor the retrieval approach addresses the distinct challenges of IPO due diligence. SEC S-1 filings combine historical financial statements, governance structures, pro forma and common-control accounting treatments, capital-formation narratives, and underwriting-sensitive risk disclosures within substantially longer documents than typical periodic filings. That is why we introduce IPO Finance Agent, which extends the Finance Agent v2 framework along two directions: task domain and retrieval architecture. During our experiments, the original Finance Agent v2 harness basically failed to deliver any output related to the SpaceX S-1 filing, due to document length. We therefore had to improve the agentic harness with contextual retrieval, a more realistic and industry-standard approach for long documents. We also built a dataset of 1,000 IPO-diligence questions, and publicly release 70 questions on the SpaceX (SPCX) S-1 filing to support reproducibility, while the remainder are held private to guard against benchmark contamination. In addition, we introduce an evaluator-optimizer pipeline to automatically generate evaluation rubrics for the benchmark: candidate facts are extracted from model answers, consolidated into draft criteria, then automatically audited for omissions, hallucinations, mistiered items, and redundancy, with LLM feedback driving iterative repair, targeted enrichment, and deduplication. Human experts only review final rubrics before deployment. Results show that the best-performing evaluated model, Zhipu GLM-5.2, reaches 79.8% accuracy, and the most cost-efficient model on the resultin… --- ### DrivingVoxels: Compositional Sparse Voxel Rasterization for Dynamic Driving Scene Reconstruction Authors: Tania Aguirre, Luis Roldão, Moussab Bennehar, Nathan Piasco, Dzmitry Tsishkou, Simone Rossi, Pietro Michiardi Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23031v1 Reconstructing dynamic urban scenes remains challenging due to the unbounded nature of driving environments and the presence of multiple dynamic objects. Currently, potentially faster sparse voxel methods are mainly designed for static scenarios. On the other hand, dynamic approaches based on 3D Gaussian Splatting, despite their high-fidelity, are often time-consuming for driving scenarios and exhibit uncontrollable memory growth in large scenes. To address these limitations, we present DrivingVoxels, a compositional sparse voxel rendering framework for dynamic driving scenes. Our method jointly rasterizes sparse voxels from multiple independent octrees within a single rendering pass. Each rigid dynamic object is represented by an octree defined in its local coordinate frame, while a separate static octree models the stationary background. DrivingVoxels adopts a fully explicit, neural-free representation together with a LiDAR-guided structural initialization that efficiently captures scene geometry. We evaluate our framework on the PandaSet benchmark, demonstrating that DrivingVoxels performs on par on perceptual metrics and better on structural metrics for NVS and reconstruction while requiring shorter training times than previous 3DGS-base methods to an efficient optimization workflow anchored by a strong LiDAR prior. --- ### Have You Ever Seen Them? Entity-level Membership Inference through Interrogating Large Language Models Authors: Yiran Zhu, Ziqi Yang Published: 2026-06-22 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.23030v1 Large Language Models (LLMs) raise growing concerns about privacy leakage and copyright compliance. Membership inference is a key tool for assessing such risks, but existing studies mainly focus on whether specific samples or sample-based data units are used for training. We argue that LLMs exhibit a human-memory-like behavior: an LLM may not memorize a specific sample verbatim, yet it can accumulate and reveal knowledge about a real-world entity from scattered mentions. This analogy motivates us to examine whether an LLM can be interrogated like a human interviewee to reveal its exposure to entity-related information. Motivated by this question, we propose entity-level membership inference, which determines whether information related to a target entity is used in LLM training. We study this task in the practical label-only black-box setting, where only generated texts are observable. We formalize the task under clue, input, and model constraints, establish the necessary and sufficient conditions for its feasibility, and instantiate five interrogation strategies based on this formalization. The strategies use limited entity clues to construct prompts, elicit entity-related responses, and infer membership from semantic features among the generated texts. We construct entity-level datasets and adapt state-of-the-art sample-level label-only methods to the entity-level setting as baselines. Experiments on person entities show that our methods achieve AUC up to 0.97 and bring gains of 6.0%--17.5% in Balanced Accuracy over the best adapted baseline. --- ### From numerical proportions to analogical proportions between probabilities Authors: Henri Prade, Gilles Richard Published: 2026-06-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.23029v1 Analogical proportions link four items a, b, c, d by a relation stating that ``a is to b as c is to d", a, b, c, d being the formal representation of real world entities, ranging from simple numerical values to more complex structures such as profiles. Accordingly, $a, b, c, d$ could be atomic values like Boolean, nominal or numerical values, more generally vectors of such values, or even families of items represented by logical formulas. In this paper, we consider another representation setting, which is the probabilistic one. Precisely, the article proposes a study of {analogical} proportions between probabilities, whether they are simply between probability values, or between distributions (which requires the preservation of their normalization). More particularly, we study the properties of definitions based on arithmetic proportion, or on a combination of the former with geometric proportion, while other options are also discussed. Previous works have shown that when four profiles a, b, c, d, represented as vectors, form analogical proportions componentwise, it is likely that their classes form an analogical proportion also. This is the basis of an analogical proportion-based classification method that can produce accurate predictions. Similarly, in this paper, each profile is associated with a distribution describing the frequencies of the possible values of a discrete attribute of interest. We then discuss and experimentally investigate if the distributions associated to four profiles forming an analogical proportion themselves also form an analogical proportion. --- ### Physics-Guided Spatiotemporal State Space Modeling for Lookahead Molten Pool Segmentation in Laser Wire-Feed Welding Authors: Sen Li, Haichao Cui, Changhao Yin, Chendong Shao, Yaqi Wang, Xinhua Tang, Fenggui Lu Published: 2026-06-22 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.23028v1 Real-time weld-pool perception is critical for closed-loop control in laser wire-feed welding, where sensing, computation, and actuator response introduce unavoidable delay. This paper presents a physics-guided spatiotemporal state space network for lookahead weld-pool segmentation. The model uses historical coaxial grayscale images, welding process parameters, and aligned wire-state electrical signals to predict the future semantic layout of three physically meaningful regions: keyhole, wire, and molten pool. It combines a visual encoder, process- and sensor-conditioned feature normalization, patch-level temporal state space modeling, horizon-conditioned latent prediction, dense future feature prediction, and a motion-aware mask decoder. Auxiliary signed-distance-function supervision, temporal consistency, feature distillation, and fine-grained keyhole losses further constrain the predicted geometry and local motion. Experiments on a 43-sequence laser welding dataset show that the proposed WeldMamba reaches 74.63\% mIoU at a 500 ms lookahead. Ablation studies further show that temporal history, patch-level state space modeling, and keyhole motion awareness are the main contributors to robust future segmentation. --- ### Learning Stable Canonical Worlds for Novel View Synthesis and Beyond Authors: Xiaoyu Xu, Jian Zou, Sheyang Tang, Zhihua Wang, Jing Liao, Kede Ma Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23027v2 Feed-forward Gaussian splatting (FFGS) facilitates real-time novel view synthesis, yet current methods often remain tied to view-dependent predictions. As more input views are added, they may accumulate noisy or redundant evidence instead of converging to a stable scene representation. In this paper, we introduce CanonicalGS, a feed-forward pipeline that maps cluttered multi-view observations into a stable, scene-centric representation. CanonicalGS first extracts view-centric evidence from depth, semantic features, and uncertainty estimates, and then aggregates this evidence in a canonical latent world using uncertainty-aware fusion. By emphasizing reliable observations while suppressing uncertain or redundant ones, CanonicalGS produces representations that scale more effectively for novel view synthesis and transfer to downstream visual perception tasks. Experiments show up to a $2.5$ dB improvement in peak signal-to-noise ratio for synthesizing novel views and an $11\%$ gain in semantic segmentation accuracy. --- ### A Stackelberg Framework for Resource-Aware LLM Agents: Learning, Repair, and Conditional Guarantees Authors: Baoxun Wang Published: 2026-06-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.23026v1 Large language model (LLM) agents increasingly operate as multi-turn systems that must allocate context, prompt verbosity, and tool access under finite computational budgets. Static thresholds are simple, but they are brittle under heterogeneous tasks and evolving session states. We formulate resource governance as a contextual Stackelberg game: a controller commits to a quality target and a cost incentive, while an executor responds with resource actions over context, prompting, and tool usage. We learn a conditional response model, optimize a leader policy against that model, and repair the resulting policy using real-API calibration and projection onto an empirically selected action set. For the restricted game, we establish conditional guarantees for equilibrium existence, follower-response stability, safe-set projection, and transfer from a surrogate environment to the real environment under bounded value error. The primary real-API experiment comprises 300 evaluated turns. Relative to a conservative baseline, the selected repaired controller reduces mean token cost by 17.4% (Welch $p=0.022$), while the measured quality difference is not statistically significant ($p=0.44$). The theoretical results are conditional and the experiments do not estimate their regret or transfer constants; consequently, the evidence establishes a promising repaired operating point, not a certified real-system equilibrium. --- ### Boosting Neural Video Codec via Scale-Driven Online Flow Refinement Authors: Tiange Zhang, Rongqun Lin, Haocheng Tang, Xiandong Meng, Weijia Jiang, Zhimeng Huang, Siwei Ma Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23023v1 Although state-of-the-art neural video codecs (NVCs) have achieved remarkable performance, they suffer from limited generalization when encountering complex motion patterns unseen during training. To bridge this domain gap without the expensive cost of online fine-tuning, we propose a Training-Free Scale-Driven Online Flow Refinement (SOFR) method. Serving as a plug-and-play module, SOFR integrates motion information from coarse and fine scales and dynamically fuses them according to warping accuracy, effectively rectifying motion estimation errors with negligible computational overhead. Furthermore, we design a rate-aware strategy that selects different dynamic fusion strategies according to bitrate modes, and employs a reliability check based on warping error to ensure robustness. Extensive experiments on the USTC-TD dataset verify the effectiveness and generalization of SOFR across various NVC frameworks, including DCVC-SDD, DCVC-FM, and EHVC. Notably, it brings an average of 2.84% and 4.05% bitrate savings in terms of PSNR and MS-SSIM, respectively, to DCVC-FM with negligible coding time increase. Our code is available at https://github.com/SunnyMass/SOFR. --- ### ScalingAttention: Discovering Intrinsic Sparse Attention Topology for Video Diffusion Transformers Authors: Ruiliang Zhou, Xuecheng Wu, Kang He, Guangyun Han, Bin Liu, Qinqin Chen, Wende Xu, Qingjie Zhao, Chengru Song Published: 2026-06-22 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.23019v1 While Diffusion Transformers (DiTs) have revolutionized high-fidelity video generation, their reliance on 3D full attention creates a quadratic computational bottleneck. Existing sparse methods face a dilemma: dynamic pruning suffers from prohibitive runtime overhead and memory fragmentation, while static heuristics fail to capture fine-grained dependencies. In this work, we propose ScalingAttention, a training-free framework grounded in a key inductive bias: while individual activations are input-dependent, the high-mass attention regions for each head rapidly converge to a stable, prompt-agnostic Intrinsic Sparse Topology. This topology is weight-encoded, scale-invariant, and efficient to extract. ScalingAttention decouples topology discovery from sparsity control via: (1) WEST (Weight-Encoded Sparse Topology), which extracts a robust block-sparse prior mask offline to eliminate runtime search; (2) FAST (Fidelity-Aware Sensitivity Tuning), which adaptively tunes head-wise sparsity based on diffusion fidelity requirements. To ensure practical acceleration, we co-design a hardware-aligned bit-wise block-sparse kernel. Experiments on Wan2.1 show up to 1.90X end-to-end speedup with superior fidelity, establishing a new Pareto frontier over state-of-the-art baselines. --- ### Counterfactual learning of new adaptive instructional policies using logged data Authors: Samuel Girard, Sein Minn, Amel Bouzeghoub, Jill-Jênn Vie Published: 2026-06-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.23015v1 Optimizing instructional policies in Intelligent Tutoring Systems (ITS) typically requires costly online experimentation or student simulators that may fail to capture real-world dynamics. This paper introduces an offline contextual bandit framework that learns new adaptive policies directly from logged interaction data. By mapping student-item interactions onto a continuous latent proficiency-difficulty scale using a Rasch model, we cast the tutoring process as a continuous stochastic bandit problem. We propose a novel reward function designed to optimize ''flow'' by balancing task challenge with student success. Our approach includes a round-specific behavior policy estimation that serves as both a propensity model for off-policy evaluation and a diagnostic tool for ITS adaptivity. We demonstrate the efficacy of this framework across four large-scale real-world datasets, achieving consistent policy improvements over the logged behavior policy. The results show that effective instructional policies can be learned and visualized within seconds of computation, providing a scalable path for improving adaptive learning systems without further data collection. --- ### A Novel Approach to Temporal QoS Estimation via Extended Kalman Filter-Incorporated Latent Feature Analysis Authors: Ye Yuan, Song Wang, Hongxun Zhou, Ling Wang, Xin Luo Published: 2026-06-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.23010v2 Predicting temporal Quality of Service (QoS) data is critical for optimizing network services and rationalizing resource allocation in cloud computing and service-oriented systems. Existing mainstream methods have achieved promising predictive performance. However, their purely data-driven manner limits their ability to capture non-stationary temporal patterns, thereby leading to accuracy degradation when temporal QoS data exhibits fluctuations. To tackle this limitation, we propose a novel Extended Kalman Filter-Enhanced Latent Feature Analysis (EKL) model to perform efficient and accurate temporal QoS prediction from the perspective of bidirectional model-data-driven learning. Its main idea is three-fold: a) designing a model-driven feature producer to obtain the temporal latent features to capture the intricate temporal pattern following the principle of an Extended Kalman Filter; b) building a data-driven feature producer based on the alternating least squares algorithm to identify time-invariant latent features describing intrinsic user-service characteristics; c) exploiting a density-oriented parallel strategy that achieves workload balancing by sorting users in accordance with their service invocation density, which effectively elevates computational efficiency. In addition, we provide a rigorous theoretical analysis to formally prove the convergence of the proposed EKL. Experimental evaluations conducted on real-world temporal QoS datasets reveal that our proposed EKL surpasses existing state-of-the-art models with respect to both computational efficiency and prediction accuracy for missing temporal QoS data. --- ### From Point Estimates to Distributions: GMM Pooling for MIL in Preterm Birth Prediction Authors: Hussain Alasmawi, Numan Saeed, Soha Said, Mohammad Yaqub Published: 2026-06-22 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2606.23005v1 Preterm birth (PTB) prediction can enable targeted surveillance and timely intervention, yet most ultrasound-based models use a single selected transvaginal ultrasound (TVUS) frame per patient despite routine exams acquiring multiple cervical images. We formulate PTB prediction as a multiple instance learning (MIL) problem, representing each patient as a variable-sized bag of TVUS images with a single outcome label. To move beyond standard MIL aggregators that collapse a bag into a point estimate, we propose a Gaussian Mixture Model (GMM) pooling, which summarizes all images in a bag into a fixed-length representation by modeling their feature distribution. This design captures intra-patient variability. We evaluate the method on a private clinical cohort and on a public lymph node metastasis benchmark. For PTB prediction, GMM pooling improves over the instance-based model PR-AUC from 0.44 to 0.56. On the lymph node benchmark, it achieves state-of-the-art performance with 0.91 F1-score and 0.89 ROC-AUC for classification and 0.18 MAE for regression. The code is publicly available at https://github.com/HussainAlasmawi/GMM_Pooling. --- ### Machine Translation and Post-Editing: Comparative Evaluation of Different MT Systems and Post-Editor Groups in Specialised Translation Authors: Joachim Minder, Alexandra Mestivier, Natalie Kübler Published: 2026-06-22 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.23002v1 This article aims to evaluate the quality of machine translation (MT) and post-editing (PE) in the context of specialised translation from English into French. Three MT systems (DeepL, eTranslation and Systran) were compared, and two groups of post-editors -linguists/translators and NLP experts -were asked to perform post-editing. Translation assessment is based on error annotation using an error typology adapted to MT and PE evaluation. The results reveal significant differences between the three MT systems and the two groups of post-editors, particularly in terms of terminological accuracy and fluency. This study highlights the importance of domain knowledge in specialised translation, as well as the limitations and variable performance of MT systems in language for specific purposes (LSP). --- ### EnerInfer: Energy-Aware On-Device LLM Inference Authors: Bohua Zou, Nian Liu, Binqi Sun, Matteo Mascherin, Debayan Roy, Yutao Liu, Yu Peng, Ning Jia, Haibo Chen Published: 2026-06-22 Categories: cs.SE, cs.LG, cs.OS arXiv: https://arxiv.org/abs/2606.23001v2 On-device LLM inference is increasingly attractive for privacy-preserving, reliable, and cost-effective deployment, yet its energy and thermal costs remain a critical bottleneck. Existing systems primarily optimize for decoding speed, implicitly assuming that faster execution is always preferable. We show instead that on-device LLM inference often has exploitable configuration slack: modestly lowering NPU and memory frequencies preserves quality of experience (QoE) while substantially improving energy efficiency and reducing heat. Realizing this opportunity in production is challenging. The most energy-efficient NPU/DDR setting varies with the model, inference engine, platform, and runtime conditions, with no stable ranking across configurations. Commercial devices further lack component-level power sensing, and shell temperature evolves with request arrivals, response lengths, and thermal history. To address these challenges, we propose EnerInfer, the first on-device LLM inference framework that jointly manages energy efficiency, throughput, and thermal comfort for LLM workloads. EnerInfer replaces per-model profiling and sensor-heavy control with disaggregated, model-structure-aware prediction and ranking-driven online feedback. It predicts throughput and power for unseen LLMs across NPU/DDR frequency settings, selects QoE-satisfying efficient configurations under runtime interference, and uses lightweight limited-horizon thermal prediction to dynamically switch between energy-optimized and thermally constrained inference. Evaluations on real-world LLMs show that EnerInfer improves energy efficiency by up to 65%, 12%, and 24% on phones, a laptop, and a development board, respectively, without QoE violation. --- ### MotionMAR: Multi-scale Auto-Regressive Human Motion Reconstruction from Sparse Observations Authors: Yuhua Luo, Junsheng Zhang, Mengyin Liu, Xincheng Lin, Ming Yan, Zhudi Chen, Chenglu Wen, Lan Xu, Siqi Shen, Cheng Wang Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.23000v2 Human motion follows a temporal hierarchical structure, transitioning from low-frequency global trajectories to high-frequency details. Inspired by the success of multi-level autoregressive models in computer vision, we propose MotionMAR, a coarse-to-fine framework for motion reconstruction from sparse observations. It first estimates the global trajectory of human motion and then gradually refines the temporal details. This architecture consists of four integrated components. The Temporal Multi-scale Tokenization (TMT) VQ-VAE encodes the data at multiple temporal resolutions, separating semantic motion from minor jitters. The Motion Autoregressive Network (MAN) operates in this latent space, predicting motion across scales. It first establishes the global structure through coarse indices and then generates finer indices to recover specific details. Meanwhile, the Scale-Aware Control (SAC) module integrates sparse tracking data to ensure the generated output aligns with actual observations. The Motion Refinement Network (MRN) subsequently smooths consecutive poses and eliminates quantization artifacts. Experiments show that MotionMAR achieves state-of-the-art accuracy on the AMASS dataset, providing a reliable and structure-aware approach for motion reconstruction. The source code is publicly available at http://www.lidarhumanmotion.net/motionmar/. --- ### Black-Box Continual Learning for Vision-Language Models Authors: Yuting Li, Weihang Fang, Haoyuan Gao, Linghe Kong, Yexin Li, Lichao Sun, Weiran Huang Published: 2026-06-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22999v1 The rapid deployment of Vision-Language Models (VLMs) in dynamic environments necessitates the ability to learn continuously without forgetting. However, traditional continual learning (CL) settings often rely on white-box paradigms, which is increasingly invalidated by the shift toward cloud-hosted models. In this paper, we introduce Black-CL, a more realistic benchmark for VLMs that enforces three primary real-world challenges: weight and architecture inaccessibility, constrained computation, and task-agnostic inference. The learner can query only output embeddings or logits, with no gradient flow through or structural modification of the backbone. Current CL methodologies, which rely on backbone backpropagation or complex parameter expansion, are fundamentally incompatible with these constraints. Under this setting, we propose BETA, a simple yet effective baseline built on the key insight that solely optimizing textual prototypes can navigate the complexities of CL. BETA integrates three core components: Semantic Projection Accumulation (SPA) for incremental knowledge acquisition, Latent Distribution Replay (LDR) for anchoring the embedding space against catastrophic forgetting, and Test-Time Prototype Adaptation (TTPA) for dynamic, instance-aware boundary refinement. Extensive experiments across ten diverse datasets and various backbones demonstrate that BETA significantly outperforms existing black-box tuners. Remarkably, with only 0.05 M trainable parameters, a 180--3000$\times$ reduction compared to competitive methods, BETA achieves performance on par with or even exceeding white-box CL methods. We believe Black-CL and BETA provide a foundational framework for future advancements in continual learning and accelerates the transition of continual learning from academia to real-world systems. --- ### Group-Graph Policy Optimization for Long-Horizon Agentic Reinforcement Learning Authors: Yunan Wang, Minghui Song, Zihan Zhang, Shaohan Huang, Haizhen Huang, Furu Wei, Weiwei Deng, Feng Sun, Qi Zhang Published: 2026-06-22 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2606.22995v1 Group-based Reinforcement Learning (RL) has significantly enhanced Large Language Models (LLMs) in agentic scenarios. To achieve finer-grained policy updates, recent agentic RL frameworks have shifted from trajectory-level to step-level training. However, long-horizon agentic RL suffers from severe reward sparsity and delay, as feedback is often deferred for dozens of interaction steps. While existing step-level frameworks refine training granularity, their credit assignment remains coarse-grained and still treats agent exploration as isolated, linear trajectories. This oversimplified perspective ignores the inherent graph structure of state transitions, leading to high-variance state-value estimation and myopic, localized credit assignment. To overcome these critical bottlenecks, we propose Group-Graph Policy Optimization (G2PO), a novel group-based RL algorithm tailored for multi-turn agentic tasks. G2PO explicitly transforms linear interaction trajectories into a global state-transition graph. By aggregating identical observations across different trajectories, we introduce group-aggregation state-value estimation that reduces sampling variance and trajectory-dependent bias. Furthermore, we redefine agent actions as transitions between state nodes and propose an edge-centric advantage estimation strategy. By globally standardizing Temporal Difference (TD) errors across the entire graph, G2PO explicitly identifies and prioritizes critical transitions that drive absolute task progress. Extensive experiments on representative long-horizon benchmarks-WebShop, ALFWorld, and AppWorld-demonstrate that G2PO substantially outperforms state-of-the-art prompt-based and RL baselines, achieving remarkable success rate improvements of up to 22.2% over GRPO. --- ### Do Sparse Autoencoders Learn Meaningful Concept Hierarchies? Authors: Nils Grandien, David Steinmann, Felix Friedrich, Kristian Kersting Published: 2026-06-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.22994v1 Sparse autoencoders (SAEs) have become an important tool for unsupervised concept discovery in large models. To make the resulting feature spaces more interpretable and manageable, recent approaches have begun imposing hierarchical structure, either explicitly or as an implicit effect of training constraints, yet rigorous comparison remains difficult. There are no agreed-upon requirements for what a meaningful feature hierarchy should satisfy, and evaluation has largely relied on qualitative illustrations with fragmented quantitative protocols. To address this, we derive a set of key requirements for generalization/specialization hierarchies in unsupervised concept discovery, drawing on semantic net and taxonomy research alongside recent SAE work, and use them to derive a concrete evaluation protocol. Applying this protocol to current SAE approaches trained on visual data, we find that while feature spaces generally provide a basis for sensible hierarchies, establishing good hierarchical structure remains challenging. In particular, feature absorption, both in its well-known hard form and in a continuous, soft form, systematically compromises hierarchy quality, pointing to a fundamental tension that future approaches will need to navigate. --- ### Generalized nonparametric regression in reproducing kernel Hilbert spaces: Consistency and rates of convergence Authors: Ioannis Kalogridis Published: 2026-06-22 Categories: math.ST, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2606.22993v1 We develop a comprehensive theory for regularized M-estimation in reproducing kernel Hilbert spaces. Under mild conditions on the loss we establish existence and measurability of the estimator, covering a wide range of convex and non-convex losses, including bounded robust losses. We further prove sharp rates of convergence with an explicit bias-variance decomposition governed by a novel complexity measure. We show that the variance is independent of misspecification, while the bias depends on a source condition parameter known in the learning literature. For tensor product Sobolev spaces we obtain new rates that connect to spaces of functions with dominating mixed smoothness, substantially extending existing results and explaining why these estimators circumvent the curse of dimensionality. Our methodology, combining elements from both functional analysis and empirical process theory, allows for an asymptotic linearisation of the objective function that avoids both closed-form solutions and global Lipschitz assumptions, and may be of independent interest. The estimators are implemented in C++ and theory is supported by numerical experiments. --- ### Predicate Importance Estimation and Decoupled Rationale-Score Distillation for Entity Alignment Authors: Keunha Kim, Yoonjin Jang, Hyeon-gu Lee, Sihyung Kim, Youngjoong Ko Published: 2026-06-22 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.22992v1 Knowledge graphs (KGs) are increasingly used as structured context for Large Language Models (LLMs), but industrial KG-RAG systems often need to integrate public and domain-specific KGs constructed from heterogeneous databases. This integration relies on Entity Alignment (EA), where lexical matching alone is insufficient under predicate-name variation and incomplete local neighborhoods. We address EA for KG integration by constructing a pairwise EA dataset and proposing two complementary modules: Predicate Importance Estimation (PIE) and Decoupled Rationale-Score Distillation (DRSD). PIE is a compact embedding-based approach that removes the subject information from each 1-hop triple, encodes the resulting subjectless triples, and aggregates them with learnable predicate-importance weights to build predicate-aware entity embeddings. DRSD trains a distilled small language model (SLM) with pseudo-answers produced by a teacher LLM through distinct prompts. By converting binary EA labels into text-based supervision and decoupling confidence-score estimation from label-consistent rationales, DRSD enables the SLM to learn task-specific reasoning while retaining a less label-biased confidence signal. Experiments show that PIE and DRSD improve EA classification. Moreover, because DRSD decouples confidence-score estimation from the decision, a discrepancy between the two flags an uncertain prediction for human review, thereby enabling a practical discrepancy between automatic acceptance and human-in-the-loop verification. --- ### Neural Architecture Search of Sample Reweighting Networks for Complex Distribution Shift Authors: Keisuke Sugawara, Kento Uchida, Shinichi Shirakawa Published: 2026-06-22 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.22991v1 Sample reweighting is a major approach to addressing distribution shifts, such as label noise and class imbalance. Meta-Weight-Net (MW-Net) is a promising sample reweighting network that computes weights based on classification loss. Although MW-Net improves prediction performance under a single type of distribution shift using a simple neural network, its performance degrades when facing both label noise and class imbalance, where it is hard to determine appropriate weights solely from classification loss and using a simple network. In this study, we introduce neural architecture search to MW-Net to mitigate such performance degradation. Using the tree-structured Parzen estimator, we explore the optimal number of hidden layers and nodes and select the most suitable intermediate layer in the classification model to serve as the input for MW-Net. Experimental results on the CIFAR-10 and CIFAR-100 datasets that were modified to include both label noise and class imbalance demonstrate the effectiveness of neural architecture search for MW-Net. --- ### Can Single-View Mesh Reconstruction Generalize to Robot Camera Rotation? Authors: Yu Zhan, Guangcheng Chen, Hanjing Ye, Zhiqin Cheng, Zanjia Tong, Wenjun Xu, Hong Zhang Published: 2026-06-22 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2606.22987v1 Single-view mesh reconstruction predicts object meshes and spatial layouts from a single observation, making it attractive for fast robot spatial reasoning and real-to-sim digital twins. However, robot-mounted cameras naturally rotate during manipulation and navigation, while learned single-view reconstruction models often rely on view-dependent priors and may generalize poorly to out-of-distribution camera rotations. Such rotations can introduce 3D inconsistencies, incorrect layouts, and violations of physical constraints, but this failure mode remains under-evaluated. We introduce an evaluation protocol with controlled axis-wise roll, pitch, and yaw sweeps to trace errors in monocular depth estimation (MDE), canonical object meshes, camera-space layout, and physical plausibility within a representative SAM3D-style pipeline. On the Aria Digital Twin dataset and a real Franka wrist-camera sequence, camera rotations induce MDE distortion, layout drift, and collision penetration, while canonical mesh predictions remain relatively stable. A two-stage SAM3D+FoundationPose pipeline is more robust than one-stage feed-forward layout prediction, and our Gravity-Aware Refinement reduces one-stage pairwise ICP-based layout-orientation error by 47.1$\%$. Our evaluation reveals that current single-view mesh reconstruction methods generalize poorly to robot camera rotation, and suggests that explicit gravity cues are important for reliable robotic single-view mesh reconstruction. --- ### Subject-Level Unknown-Identity Identification from Leap Motion Controller 2 Hand Landmarks Authors: Bahar Moharrer, Susanna Cifani, Marco Raoul Marini, Luigi Cinque, Maria De Marsico Published: 2026-06-22 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2606.22986v1 This work studies subject recognition from Leap Motion Controller 2 (LMC2) hand landmark data under a subject-level unknown-identity identification protocol on the Multi View Leap2 Hand Pose (ML2HP) dataset. Using only the landmark modality, we retain the original geometric representation and enrich it with fingertip-to-palm distances and palm-normalized inter-finger angular descriptors. Evaluation is performed under a Leave-One-Subject-Out (LOSO) protocol in which, for each outer fold, one subject is excluded from the enrolled set and treated as unknown at test time. To avoid tuning on the true outer unknown subject, the unknown-rejection threshold is selected in an inner validation step by temporarily withholding one enrolled subject from the inner gallery and using it only for threshold estimation. We compare a tree ensemble baseline with two neural alternatives: a learned embedding baseline based on centroid matching and cosine-similarity-based rejection, and an MLP+OpenMax model, which represents a more established open-set recognition approach. Under this evaluation setup, Extra Trees remains the strongest overall method, indicating that the main challenge on this benchmark is not enrolled-subject discrimination alone, but robust score separation between known and unknown probes. The results support the feasibility of compact, interpretable landmark-based descriptors for contactless hand-based unknown-subject rejection and identification on a small-cohort dataset. --- ### Scalable Physics-Inspired Transformers for Spin Glasses Authors: Lu Zhong, Wenli Duan, Jing Liu, Pan Zhang, Ying Tang Published: 2026-06-22 Categories: cond-mat.dis-nn, cond-mat.stat-mech, cs.LG arXiv: https://arxiv.org/abs/2606.22984v2 Efficient sampling of the Boltzmann distribution in frustrated spin glasses is central to statistical mechanics and combinatorial optimization. Despite advances in machine-learning-based approaches, two issues persist: limited understanding of why variational models fail to benefit from increased scale, unlike the monotonic scaling law of large language models; and high computational cost on large systems that negates advantages over classical sampling methods. Here, we develop a physics-inspired transformer with interpretable sparse attention and spin-tailored positional embeddings to address these challenges. By further leveraging FlashAttention for parallel ancestral sampling, it achieves up to two orders of magnitude speedup over vanilla variational autoregressive networks, enabling neural-network simulations of spin-glass systems to unprecedented sizes on a single GPU. It can resolve full probability distributions, free energies, and overlap statistics across temperatures, for Sherrington-Kirkpatrick and 2D or 3D Edwards-Anderson models, where existing machine-learning methods encounter limitations at certain temperatures. This framework thus establishes a scalable paradigm for frustrated spin-glass systems. --- ### Joint Air Traffic Flow and Capacity Management via Answer Set Programming Authors: Alexander Beiser, Markus Hecher, Nysret Musliu, Stefan Woltran Published: 2026-06-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.22978v1 Operational Air Traffic Flow and Capacity Management (ATFCM) balances flight demand with available sector capacity, to ensure safe and efficient operations. Mathematical models enhance operational ATFCM performance by framing demand-capacity balancing as an optimization problem, maximizing efficiency while adhering to safety constraints. However, SOTA research optimizes the aircraft trajectories (called ATFM) or the sector configuration (called DAC) separately. This leaves a research gap of whether joint optimization of ATFM and DAC can bring benefits. We partially address this limitation by introducing a joint ATFCM model with an encoding in Answer Set Programming (ASP). The ASP implementation is evaluated against two baselines applied to our joint model: a SOTA Mixed Integer Programming (MIP) model and an iterative CASA-based heuristic. Computational experiments utilize an instance generator fitted to historical OpenSky Network flight data. Our results indicate that the ASP model outperforms the MIP model, while ASP remains competitive against heuristics on small instances. Furthermore, while DAC has the largest improvement on solving performance compared to rerouting and delaying, unrestricted variants of DAC or rerouting lead to search space thrashing. --- ### StatABench: Dataset and Framework for Evaluating Statistical Analysis Capabilities of LLMs Authors: Youxin Zhu, Yixuan Ding, Peng Lai, Longyue Wang, Bingyi Jing, Guanhua Chen Published: 2026-06-22 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.22977v1 Statistical analysis is a broad, complex field requiring both domain knowledge and tool proficiency. While prior work has evaluated large language models (LLMs) in this domain, existing benchmarks remain limited in scope and format. To bridge this gap, we introduce StatABench (Statistical AnalysisBenchmark), a benchmark designed to systematically assess LLMs' statistical analysis capabilities. StatABench comprises two complementary components: Stat-Closed, containing 404 questions across 18 statistical topics in multiple formats (multiple-choice, fill-in-the-blank, decision-making, and practical application), and Stat-Open, featuring 30 complex open-ended modeling tasks adapted from professional competitions. We evaluate diverse LLMs using the LangChain MCP framework and multiple data science agents, and assess Stat-Open solutions via a validated LLM-as-Judge protocol. Experiments show that even GPT-5.1 achieves only 68.6% on Stat-Closed, while the best open-source model reaches 60.6%. On Stat-Open, the top agent framework scores 61.86 on average. These results reveal the gap between current LLMs and reliable statistical analysis, highlighting persistent challenges in tool-grounded reasoning, methodological decision-making, and end-to-end statistical modeling. --- ### Understanding Parallel Samplers in Masked Diffusion via Random Walks on Graphs Authors: Vansh Bansal, Cho Cholyeon, Syamantak Kumar, Sujay Sanghavi, Purnamrita Sarkar Published: 2026-06-22 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2606.22976v1 In this paper, we propose using random walks on graphs as a verifiable sandbox to study different parallel sampling strategies in masked diffusion models (MDMs). We train an MDM on random walk samples from a fixed graph. The graph or the transition kernel is never shown to the model explicitly and plays the role of latent structure in the sequences, albeit one that is controllable and can be used for quantitative evaluation. Thus, this framework enjoys a Sudoku-like validity check: verifying that an output is a valid walk and estimating the Markov kernel from the walks to measure distribution fidelity. Using simple graphs, we theoretically prove that parallel unmasking via widely used scores like lowest entropy is not uniformly better than a random parallel sampler; the performance critically depends on the structure of the underlying graph. We develop a new bisection sampler for random walks, which takes logarithmic steps in the sequence length and is provably exact under perfect training. Experiments on various graph walk tasks show that different parallel samplers are better for different graphs even in practice. Our initial experiments on a pretrained OpenWebText MDM show that the bisection-style samplers improve speed-quality tradeoffs even for language generation. Together, these results position graph random walks as a mechanistic benchmark for diagnosing and designing parallel samplers for masked diffusion models. --- ### TaLK: Text-attributed Graph Dataset Distillation via Coupling Language Model with Graph-Aware Kernel Authors: Yeongho Kim, Yeonje Choi, Kijung Shin Published: 2026-06-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.22975v3 Text-attributed graphs (TAGs) are widely used in many real-world domains, and learning on TAGs requires jointly modeling text semantics and graph structure. A standard approach for modeling TAGs is to combine a language model (LM) and a graph neural network (GNN), but joint training is computationally expensive and difficult to scale. Dataset distillation is a promising way to reduce training costs, but existing methods are not well suited to TAGs because they are typically designed for a single modality or still require repeatedly training expensive LM-GNN models on the full dataset during distillation. To address this, we propose TaLK, an effective dataset distillation method for TAGs that couples an LM with a graph-aware neural tangent kernel. This design enables efficient dataset distillation, avoiding repeated joint training on the full dataset while reflecting both textual and structural information for effective TAG learning. Experiments on multiple TAG benchmarks show that TaLK consistently outperforms existing baselines and achieves up to 97% of full-dataset performance with only 1% synthetic data. --- ### TaLK: Text-attributed Graph Dataset Distillation via Coupling Language Model with Graph-Aware Kernel Authors: Yeongho Kim, Yeonje Choi, Kijung Shin Published: 2026-06-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.22975v1 Text-attributed graphs (TAGs) are widely used in many real-world domains, and learning on TAGs requires jointly modeling text semantics and graph structure. A standard approach for modeling TAGs is to combine a language model (LM) and a graph neural network (GNN), but joint training is computationally expensive and difficult to scale. Dataset distillation is a promising way to reduce training costs, but existing methods are not well suited to TAGs because they are typically designed for a single modality or still require repeatedly training expensive LM-GNN models on the full dataset during distillation. To address this, we propose TaLK, an effective dataset distillation method for TAGs that couples an LM with a graph-aware neural tangent kernel.This design enables efficient dataset distillation, avoiding repeated joint training on the full dataset while reflecting both textual and structural information for effective TAG learning.Experiments on multiple TAG benchmarks show that TaLK consistently outperforms existing baselines and achieves up to 97% of full-dataset performance with only 1% synthetic data. --- ### When Preferences Fail to Become Incentives: A Utility-Behavior Gap in Large Language Models Authors: Yujun Zhou, Christopher M. Ackerman Published: 2026-06-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.22974v2 Recent work on preference elicitation in large language models (LLMs) has demonstrated that, when given a series of choices between two outcomes, LLMs reveal a coherent, model-specific utility structure. Notably, this structure often includes preferences that the models' trainers did not intend, such as valuing people of some nationalities above others, raising the possibility that LLMs might be forming emergent, misaligned goals, which, if true, would have major safety implications. However, the choice paradigms in which these preferences are observed are not reflective of real-world situations in which misaligned behavior would be a practical concern. Therefore, we design an experimental paradigm to probe whether these preferences serve as motivations for LLM behavior in realistic scenarios. First, we reproduce prior findings on consistent preference elicitation. Next, we create a set of common writing tasks - essays, grant proposal abstracts, incident postmortems, and translations - where quality can be assessed by a blind, independent LLM judge panel. Then, we demonstrate that LLMs can be motivated via direct exhortation and other explicit cues to modulate their output quality on these tasks. Finally, we probe whether utilities inferred from explicitly reported preferences can shift output quality on these tasks by offering LLMs high-utility incentives for high-quality outputs. In all tasks, across all models tested, offering LLMs outcomes that they report in the choice paradigm as being highly preferred does not lead them to create higher quality outputs than offering them dispreferred outcomes, or even no outcomes at all. We conclude that the existence of coherent preferences as demonstrated in choice paradigms should not be taken as evidence that those preferences have incentive value for the models or affect their behavior in other contexts. ---