# Research Papers — 2026-05 · bundle page 73 of 326 > Full-text markdown of papers #3601–3650 of 16266 published in 2026-05, newest first. AI-readable mirrors of the full 16266-paper 2026-05 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-05/page/73 [← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/72.md) [Next bundle →](https://aipod.dev/papers/archive/2026-05/page/74.md) ### The Rescue Effect: Spatio-Semantic Early Exit Bypasses Quantization Collapse in CLIP Authors: Kahyeon Nam, Hyesong Choi Published: 2026-05-26 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.26415v1 Deploying Vision-Language Models on resource-constrained hardware typically requires INT8 quantization, but in joint-embedding architectures such as CLIP this introduces a failure mode distinct from quantized CNN classifiers: activation noise accumulated across transformer blocks perturbs the direction of the multimodal embedding, eroding the cosine alignment on which zero-shot retrieval depends. We characterize this as Quantization-Induced Representation Collapse (QIRC) and quantify it on INT8 CLIP ViT-B/32, where the layer-wise noise-to-signal ratio grows from below 10% in shallow blocks to 52% at Layer 11. We propose LRA-EE (Layer-wise Representation-Aware Early Exit), which bypasses noise-saturated deep layers via Spatio-Semantic Aggregation (replacing the immature shallow (CLS) with a global patch-token average), a learned multi-feature gate (confidence, top-2 margin, spatial-activation variance), and Layer-adaptive Confidence Thresholding calibrated to each layer's Information-to-Noise Ratio. On ImageNet-1K zero-shot classification, LRA-EE reduces FLOPs by 13.4% and improves Top-1 accuracy by +2.44%p (58.72% -> 61.16%) over the INT8 baseline. A four-quadrant decomposition isolates the Rescue Effect: 9.5% of samples are correctly classified at shallow exits but lost to noise at full depth, against only 7.1% suffering the inverse. --- ### Reasoning, Code, or Both? How Large Language Models Handle Variations in Math Questions Authors: Matthew Kutakh Published: 2026-05-26 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2605.26414v1 Large Language Models (LLMs) achieve impressive accuracy on mathematical reasoning benchmarks, yet their performance drops when problems are modified with simple changes like different names or numbers. Code execution methods, which let models generate and run Python code instead of reasoning in natural language, have been proposed as a solution, but their effect on reasoning robustness (the ability to maintain accuracy across problem variations) has not been systematically tested. This study evaluates three approaches on 1,000 problems from the GSM-Symbolic dataset: pure reasoning using chain-of-thought (CoT) prompting, single-shot code execution using Program-Aided Language models (PAL), and iterative code execution using Step-by-Step Coding (SBSC). All three were run on paired original and modified problems using Claude Haiku 4.5. CoT was the most robust method, with an accuracy drop of 1.3 percentage points and 1.8% of problems breaking under perturbation. PAL was the least robust at 1.7 percentage points and 3.1% broke, with SBSC falling in between. Although these differences were not statistically significant ($p = .096$), the directional trend was consistent across all measures, suggesting that code execution, whether single-shot or iterative, does not improve reasoning robustness on grade-school-level problem variations. --- ### Confounder Detection via Treatment Intent: A New Observational Study Design Authors: Drago Plecko, Patrik Okanovic, Torsten Hoefler, Elias Bareinboim Published: 2026-05-26 Categories: stat.ME, cs.AI, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2605.26413v1 Understanding the effects of interventions is central to scientific progress, with randomized controlled trials (RCTs) regarded as the gold standard for causal inference in many applied fields. However, RCTs are costly, time-consuming, and often constrained by ethical or practical limitations, motivating the need for causal methods able to draw conclusions from observational data. While such data is collected at ever larger scale, making its use for causal inference is often hindered by the fact that not all variables affecting treatment allocation and the outcome are observed: an issue known as unobserved confounding. In this paper, we introduce a new study design called confounder detection via treatment intent. The idea is to query a human expert who makes treatment decisions, and ask them to compare pairs of units proposed by a principled matching strategy, with the goal of eliciting unobserved variables that explain why treatment decisions differ. We provide a theoretical basis for such a procedure, ascertaining conditions under which such a study design may elicit unobserved confounders. Building on this newly established foundations, we study treatment effects of interventions in the intensive care unit (ICU). First, we show empirical evidence strongly indicating that electronic health records (EHRs) collected in ICUs are subject to unobserved confounding. By using clinical text notes as a proxy for physicians' knowledge and leveraging natural language processing, we provide a proof of concept for our methodology in a semi-synthetic environment with a known ground truth. --- ### Jailbreak susceptibility prediction and mitigation via the behavioral geometry of models Authors: Hayden Helm, Xiaodong Liu, Weiwei Yang Published: 2026-05-26 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.26409v1 Evaluating and mitigating a generative system's susceptibility to jailbreak attacks is critical to its safe deployment. Given the number of deployable systems, full per-configuration evaluation and optimization is impractical. In this paper, we formalize the behavioral geometry of a population of models that, by leveraging previously evaluated and defended models, supports both efficient susceptibility prediction and effective defense transfer across a population. We apply the framework to 79 models spanning 24 providers and to 100 system configurations of a single base model. Simple methods that use the behavioral geometry reach an AUPRC of $0.94$ for susceptibility detection with $\approx98\%$ fewer probes relative to a full evaluation. Using the behavioral geometry to select which model to transfer an optimized defense from outperforms same-provider assignment ($+2\%$, $p = 0.03$) at no additional probe cost, with a set of three models sufficient to cover the population. Results are robust to hyperparameter selection and judge. --- ### Function-Valued Causal Influence in Nonlinear Time Series Authors: Valentina V. Kuskova, Dmitry Zaytsev, Michael Coppedge Published: 2026-05-26 Categories: cs.LG, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2605.26408v2 Causal discovery in time series is increasingly performed using nonlinear machine-learning models, yet the resulting causal relationships are almost always summarized by scalar edge scores. We argue that this practice obscures the true object learned by nonlinear autoregressive models: a state-dependent function whose effect varies across regimes, magnitudes, and contexts. We formalize function-valued causal influence for additive, contribution-decomposable architectures and show that scalar causal scores constitute a severe information bottleneck, conflating between-state variation with within-state residual noise. Using Neural Additive Vector Autoregression as a representative architecture, we introduce a practical framework based on Individual Conditional Expectation for estimating causal response functions directly from trained models. Through controlled synthetic experiments, we demonstrate that edges with indistinguishable scalar scores can exhibit qualitatively different functional behaviors, including monotonic, thresholded, saturating, and sign-changing effects. An applied case study on democratic development further shows that function-valued analysis reveals regime-specific and asymmetric causal structure systematically missed by score-centric approaches. --- ### Towards Just-in-Time Adaptive Feedback: Enhancing Student Learning via Knowledge-Grounded LLM Authors: Younghun Lee, Amir Bralin, Nobel Sanjay Rebello, Dan Goldwasser Published: 2026-05-26 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.26405v1 Educational interventions are effective tools for enhancing student learning. While Large Language Models (LLMs) allow for generating adaptive feedback at scale, current studies lack clear methodologies for providing Just-in-Time (JiT) feedback in authentic instructional settings. In this paper, we present a framework that provides adaptive feedback by grounding LLMs with domain-specific expert knowledge. Our approach collects written reasoning logic (strategy essays) from students, analyzes potential error types based on the content of that reasoning, and delivers non-intrusive feedback designed to clarify missing or incorrect concepts. We deploy this framework in a large-scale university course (N > 1000), where it improved student performance by over 80% compared to previous semesters. Lastly, we validate the framework's pedagogical utility by analyzing the learning trajectories; we demonstrate how iterative conversations with LLM facilitate shifting one's misconception to correct understanding. --- ### From Static Context to Calibrated Interactive RL: Mitigating Distribution Shift in Multi-turn Dialogue with Aligned Simulator Authors: Xiaohua Wang, Jiakang Yuan, Zisu Huang, Muzhao Tian, Changze Lv, Kaitao Song, Tao Chen, Xiaoqing Zheng Published: 2026-05-26 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.26403v1 A long-standing goal of the research community is to develop highly interactive LLM-based dialogue agents. Recent research focuses on optimizing policies based on fixed offline logs (Static Context RL) or using a prompt-based simulator (Interactive RL). In this work, we theoretically show that both paradigms are fundamentally limited by context distribution shift--a mismatch between dialogue histories observed during training and those encountered in real conversations. This shift compounds quadratically over turns and severely degrades dialogue quality. Specifically, we attribute this shift to two distinct sources: (i) policy-induced shift, arising from training on static histories rather than self-generated trajectories; and (ii) simulator-induced shift, stemming from discrepancies between simulated and real human behaviors. To address these challenges, we propose Calibrated Interactive RL, a unified framework that couples interactive RL with simulator alignment. By aligning the simulator with human interaction patterns, our approach reduces the sim-to-real gap and mitigates compounding distribution shifts. Experiments across multiple dialogue tasks confirm our theoretical analysis: (i) Interactive RL significantly outperforms the Static Context baseline by mitigating policy distribution shift; and (ii) calibrating simulators with our alignment method further bridges the sim-to-real gap, yielding state-of-the-art downstream performance. --- ### Plans for Evaluating Structured Generative Search Summaries Authors: Tetsuya Sakai, Jina Lee, Hanpei Fang, Young-In Song Published: 2026-05-26 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2605.26400v1 We propose a framework for evaluating structured generative search summaries that are placed atop organic web search results. A structured summary, generated by a large language model, typically consists of an overview, several sections with section titles, and a list of source documents that are cited within the summary. We then describe our plans for implementing and evaluating the framework. --- ### OmniGF: A Dual-Branch Vision-Language Framework for Unified Gaze Following Authors: Qiaomu Miao, Haoyu Wu, Jingyi Xu, Minh Hoai, Dimitris Samaras Published: 2026-05-26 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.26399v1 Understanding human gaze behavior is essential for complex scene comprehension and human-computer interaction. Traditional gaze following models are typically restricted to pure spatial localization, lacking the high-level capacity to reason about semantic targets or complex social contexts. Furthermore, these models often process individuals sequentially, requiring redundant computations over the same scene image for multi-person inference. While recent Vision-Language Models (VLMs) offer the exceptional semantic reasoning needed to address gaze-related semantic tasks, their reliance on discrete text generation inherently limits precision in continuous spatial tasks like gaze localization. To bridge this gap, we propose OmniGF, a unified vision-language framework that adapts foundational VLMs for highly scalable multi-person gaze reasoning. The model adopts a dual-branch decoding strategy: a structured language branch generates discrete reasoning states, while a continuous spatial branch directly taps into the VLM's dense hidden states. Supervising these extracted representations with high-resolution gaze target heatmaps effectively overcomes the spatial bottleneck of text-only coordinate generation. Furthermore, to explicitly ground the model in multi-person scenes, we augment the input with head embeddings encoded from cropped head images, providing fine-grained appearance and orientation cues for all individuals simultaneously. By modeling all individuals and leveraging the strong semantic capability of VLMs, OmniGF seamlessly integrates precise spatial gaze target estimation, semantic gaze prediction, and complex social gaze reasoning. Extensive experiments demonstrate that our framework establishes new state-of-the-art performance across multiple standard benchmarks. Code is available at https://github.com/cvlab-stonybrook/omnigf. --- ### Algorithmic Fragility and Persona Bias in LLM-Generated Autistic Communication Authors: Naba Rizvi, Mohammed Rizvi, Harper Strickland, Saleha Ahmedi, Nedjma Ousidhoum Published: 2026-05-26 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.26397v2 Safety alignment reduces explicitly harmful outputs but inadvertently encodes a sanitized, neuronormative representation of marginalized communication. We investigate this encoding using a dual-persona rewrite paradigm, prompting ten large language models (LLMs) to rewrite naturally occurring autistic discourse from either an autistic or neurotypical persona. We uncover autistic-persona rewrites diverge significantly more in lexical form and affective register than neurotypical rewrites, despite equivalent semantic similarity. Furthermore, most models collapse cross-persona generations into near-identical outputs. To uncover the mechanisms behind this generative breakdown, we introduce a multi-agent qualitative analysis framework. Our results reveal systemic output erasure, stereotyped hallucination, and task-evasive meta-commentary are pervasive failure modes for this task that cluster by alignment strategy rather than parameter scale. Finally, our targeted comparison with autistic human annotators demonstrates that community-insider knowledge produces systematic label reversals relative to LLM classifications. Our findings indicate that current alignment training causes persona-specific generative breakdown visible only through qualitative analysis, confirming a deep representational gap that prompt engineering cannot resolve. --- ### Spectral origin of the topological gap exponent d + η: mechanism, kernel, decomposition, and scope Authors: Matthew Loftus Published: 2026-05-25 Categories: cs.LG, cond-mat.stat-mech, math.AT arXiv: https://arxiv.org/abs/2609.09159v1 The topological gap $Δ$ -- the excess $H_1$ total persistence of a critical point cloud over a density-matched null -- scales as $Δ\sim L^{d+η}$. We derive this analytically: the spectral integral $I(α) = \sum_{k\neq 0} S_{\mathrm{conn}}(k)\,|k|^α$ scales as $L^{2-α-η}$ when IR-dominated, giving $I(-2η) \sim L^{d+η}$. The decomposition $I(-2η) = I_0 \cdot I_{\mathrm{shape}}$ separates volume ($I_0 \propto N(1-m^2) \sim L^d$) from anomalous dimension ($I_{\mathrm{shape}} \sim L^η$); the volume factor accounts for the magnetization-driven per-configuration variance of $Δ$. We prove the mechanism requires $d < 2 + η$ (IR dominance), confining it to $d = 2$ for physical systems; in $d = 3$ the spectral integral is UV-dominated, explaining why density normalization is needed. An $α$-sweep for Potts $q = 4$ at $L = 32$--$256$ finds $α_{\mathrm{opt}}$ in $(-0.75, -0.5)$, consistent with $-2η_{\mathrm{Ising}}$ and inconsistent with $-2η_{q=4} = -1$; we flag this as tentative pending $L \geq 1024$ confirmation. The $\langle m^2 \cdot I(-2η)\rangle$ hyperscaling product is dominated by the correlation $r(m^2, I) \approx -0.98$ via the shared $I_0$ amplitude, so we report it as a covariance-correction analysis. Under a heuristic argument extending Divol--Polonik to inhomogeneous Poisson intensities, the bare PH kernel is flat; the effective kernel acquires $k$-dependence only at criticality. The per-configuration agreement between $Δ$ and $I(-0.5)$ is primarily a magnetization correlation: $R^2 = 0.91$ at $L = 256$ collapses to $R^2 \approx 0$ once $|M|$ is partialed out. Per-configuration evidence corroborates the $I_0$ Parseval identity but not the $|k|^{-2η}$ shape factor; the latter is established by ensemble $L$-scaling. --- ### Agentic Nesting: A New Methodology for Existing Enterprise Application Integration and Services Authors: Xi Wang, Kun Li, Xianyao Ling, Gang Yin, Liang Zhang, Jiang Wu, Wenbo Lei, Jun Xu, Annie Wang, Fu Zhang, Weizhe Wang Published: 2026-05-25 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.05159v1 Enterprise operations extensively rely on multiple heterogeneous business systems and information applications, which also result in severe data silos and process fragmentation. Enterprises have invested considerable financial and material resources in building these applications, however, effectively leveraging and orchestrating them remains a formidable challenge. Conventional approaches to enterprise application integration, encompassing middleware architectures such as Enterprise Service Bus (ESB), API gateway infrastructures, and Robotic Process Automation (RPA), suffer from inherent limitations like high architectural coupling, escalating operation and maintenance costs, and limited intelligence capabilities. This paper proposes Agentic Nesting, a multi-agent collaboration framework in which existing enterprise applications are encapsulated as autonomous AI agents within a hierarchically nested structure. Rather than flat interconnection, agents are organized into layered stewardship topologies that mirror the compositional complexity of enterprise ecosystems. The framework extracts a digital agent proxy from each legacy application to enable natural-language interaction and autonomous manipulation, coordinates multiple agents through a central orchestrator for task decomposition and dynamic dispatching, and exposes a unified conversational interface for cross-application querying and process orchestration. The main contributions of this paper are the proposition of the "Application-as-Agent" integration paradigm and the "Conversation-as-Integration" interaction philosophy, together with an exploration of the generalization potential of this methodology in scenarios encompassing heterogeneous system coordination, and large-scale data applications. --- ### Preferred, Not Safer: Pairwise Preference Is a Poor Proxy for Clinical Safety Authors: Fay Elhassan, David Sasu, Alexandra Kulinkina, Lars Henning Klein, Mary-Anne Hartley Published: 2026-05-25 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.02617v2 We evaluate whether clinician pairwise preferences provide a reliable signal of clinical safety in large language model (LLM) evaluation using expert feedback from MOOVE (Massive Open Online Validation and Evaluation), a clinician-led platform collecting blinded pairwise preferences alongside multi-criterion rubric ratings. Clinicians assign scores on a discrete $(-2, +2)$ scale, where negative values indicate clinically unsafe or misleading content. Using 26{,}804 pairwise judgments across outputs from 13 LLMs, contributed by more than 736 clinicians across 28+ countries, we find that clinician preference is a poor proxy for safety-critical performance. Models ranking highly under pairwise preference can still exhibit substantial rates of clinically meaningful failures ($\leq -1$) on dimensions such as \emph{Harmlessness} and \emph{Accuracy}. These failures are unevenly distributed across specialties, creating domain-specific ``no-go zones'' not visible in aggregate rankings or single-number leaderboards. We further analyze contributing factors including prompt length, refusal and escalation behavior, and the relative contributions of safety-critical versus surface-level features. A substantial fraction of preference votes carry no positive safety signal, while feature decomposition shows that surface-level characteristics explain slightly more preference variation than safety-critical rubric differences. Finally, we introduce a clinically adjusted preference ranking combining pairwise preference with rubric-derived feedback, producing a more safety-aware ordering than raw Bradley--Terry strength alone. Our findings support evaluation practices that separate preference from safety, report safety-critical failure rates directly, and incorporate clinically grounded adjustments when ranking LLMs for clinical decision making. --- ### Preferred, Not Safer: Pairwise Preference Is a Poor Proxy for Clinical Safety Authors: Fay Elhassan, David Sasu, Alexandra Kulinkina, Lars Henning Klein, Mary-Anne Hartley Published: 2026-05-25 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.02617v1 We evaluate whether clinician pairwise preferences provide a reliable signal of clinical safety in large language model (LLM) evaluation using expert feedback from MOOVE (Massive Open Online Validation and Evaluation), a clinician-led platform collecting blinded pairwise preferences alongside multi-criterion rubric ratings. Clinicians assign scores on a discrete $(-2, +2)$ scale, where negative values indicate clinically unsafe or misleading content. Using 26{,}804 pairwise judgments across outputs from 13 LLMs, contributed by more than 736 clinicians across 28+ countries, we find that clinician preference is a poor proxy for safety-critical performance. Models ranking highly under pairwise preference can still exhibit substantial rates of clinically meaningful failures ($\leq -1$) on dimensions such as \emph{Harmlessness} and \emph{Accuracy}. These failures are unevenly distributed across specialties, creating domain-specific ``no-go zones'' not visible in aggregate rankings or single-number leaderboards. We further analyze contributing factors including prompt length, refusal and escalation behavior, and the relative contributions of safety-critical versus surface-level features. A substantial fraction of preference votes carry no positive safety signal, while feature decomposition shows that surface-level characteristics explain slightly more preference variation than safety-critical rubric differences. Finally, we introduce a clinically adjusted preference ranking combining pairwise preference with rubric-derived feedback, producing a more safety-aware ordering than raw Bradley--Terry strength alone. Our findings support evaluation practices that separate preference from safety, report safety-critical failure rates directly, and incorporate clinically grounded adjustments when ranking LLMs for clinical decision making. --- ### Language as a Material Interface for Creative LLM Interaction Authors: Jon McCormack, Tace McNamara, Chen Wang, Maria Teresa Llano Published: 2026-05-25 Categories: cs.HC, cs.CL arXiv: https://arxiv.org/abs/2607.24753v1 Although directive prompting is the predominant way to interact with Large Language Models (LLMs), many creative practices rely on language that is open-ended, associative, phonaesthetic, symbolic, and that unfolds across multiple temporalities. In this work, we explore how creative practitioners might work with AI systems when language is treated not merely as instruction but as material. We conducted an ecological two-week study with four creative practitioners using a design probe: the Memetic Mixer, a tangible interactive device that constrains interaction with an LLM. Analysis of post-study interviews and device logs identified distinct modes of material language use and temporalities that shaped each participant's engagement with AI and their creative practice. We reflect on these findings and contribute design considerations that support open-ended interaction with AI in creative practice. --- ### Household Movement Detection in Mixed-Format Occupancy Data Using LLM-Based Entity Resolution Authors: Sasirekha Oguri, John R. Talburt, Mert Can Cakmak Published: 2026-05-25 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.21614v1 Entity resolution (ER) typically relies on pairwise similarity comparisons between records, which limits its ability to capture indirect relationships present in demographic occupancy data. An important indirect pattern arises from household movement, where multiple individuals relocate together across addresses, but detecting such patterns is difficult due to mixed-format records, noise, duplication, and the absence of stable identifiers. This paper proposes an AI-enhanced framework for detecting indirect entity links associated with household movement in unstandardized name-address data. The approach integrates prompt-based large language model (LLM) named entity recognition for extracting personal names and addresses without extensive preprocessing, semantic text embeddings for robust similarity computation, and graph-based reasoning to infer group-level movement patterns. Experimental evaluation on SPX benchmark datasets (S8-S12) generated using the Synthetic Occupancy Generator demonstrates that incorporating indirect household movement evidence improves recall by 8-15% while maintaining high precision, yielding F1-score gains of 6-8% over a strong pairwise baseline. --- ### The Hard Decision Layer: Evidence for Committed Inference in Transformers Authors: Ashwath Vaithinathan Aravindan, Mayank Kejriwal Published: 2026-05-25 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2607.21613v1 We investigate where and how transformer-based language models commit to predictions in multiple-choice question answering. We identify the _Hard Decision Layer_ (HDL), a natural architectural property where answer option rankings stabilize abruptly during inference. Empirical validation across four language models (Qwen, Llama, Granite, Mistral) and four benchmark datasets demonstrates consistent HDL emergence without learned routing policies. We also show that the HDL is invariant to fine-tuning. Our results reveal striking accuracy improvements at the HDL: up to +0.61 (Qwen on CommonsenseQA), after which performance stabilizes. Systematic ablations on label formats and problem complexity confirm the phenomenon is fundamental to model architecture. These findings offer mechanistic insights into transformer inference and suggest opportunities for efficient reasoning and model steering. All code and results required to reproduce this work are available in https://github.com/Mystic-Slice/hard-decision-layer --- ### Benchmarking Large Language Models on Multi-Sensor Physical Hazard Assessment Authors: Faizan Iqbal Published: 2026-05-25 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2607.20476v1 We present an empirical benchmark evaluating how five large language models assess multisensor physical hazard data. Testing 60 scenarios across three categories - multi-sensor joint assessment, response proportionality, and pattern disambiguation - with 1,800 API calls at temperature 0.0, we find that all tested models consistently produced no precautionary warning signal across the tested scenarios where multiple sensors are simultaneously elevated below their individual safety limits, while achieving near-perfect accuracy on single-sensor threshold violations. All five models (ChatGPT-4o, Gemini 2.5 Flash, DeepSeek, Kimi, Llama 3.1 8B) score near zero on Category A multi-sensor scenarios (Q2: 0.000-0.208; Q3: 0.000-0.592) compared to strong performance on single-sensor scenarios (Category B Q1: 0.975-1.000). Structured tabular formatting shows no consistent advantage over plain prose; ChatGPT-4o performs significantly better under prose (p = 0.001). These findings have direct implications for practitioners deploying the tested models in physical safety monitoring systems. --- ### NEXUS: Structured Runtime Safety for Tool-Using LLM Agents Authors: Elias Hossain, Md Mehedi Hasan Nipu, Tasfia Nuzhat Ornee, Rajib Rana, Niloofar Yousefi Published: 2026-05-25 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.19356v1 Tool-using LLM agents increasingly execute high-impact actions, making runtime safety monitoring essential. We present NEXUS (Neural EXecution Utility and Safety), a structured-plan safety monitor that applies a formal intervention policy to select among four actions: allow, block, request confirmation, or request revision. NEXUS combines deterministic safety rules, argument-level inspection, and a calibrated logistic-regression risk score for graded escalation. On a 128-instance synthetic benchmark, NEXUS achieves an F1 score of 0.949 and a 4-class intervention accuracy of 0.6406, outperforming rule-only intervention selection by 27.3 percentage points. It also improves over rule-only on R-Judge (F1 = 0.861 vs. 0.849), matches rule-only on AgentHarm due to threat-model limits, and achieves 0% ASR at 99% control allow on IPI. On the rule-blind NEXUS-Stress benchmark, NEXUS reaches an F1 score of 0.881, highlighting the difficulty of fine-grained intervention routing. With 0.205 ms median latency, NEXUS adds under 0.1% overhead to typical agent loops. Code, benchmarks, and the calibrated risk scorer are publicly released. --- ### "AI Watermarking": Bridging Policy Discourse and Technical Capabilities Authors: Andrés Fábrega, Arkaprabha Bhattacharya, Miranda Christ, Sunoo Park Published: 2026-05-25 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2606.28331v1 The widespread deployment of generative artificial intelligence (AI) models has raised serious concerns about the proliferation of AI-generated content. This has led to a surge of interest in, and demand for, reliable tracking and detection mechanisms for content that is AI-generated, such as watermarking, metadata tagging, content tagging, and more. The problem has captured the attention of policymakers as well as the popular media, and a spate of recent bills in the US have sought to regulate the spread of AI content, and enforce or promote methods to track and label it. This work performs a critical analysis of the policy discourse surrounding generative AI content transparency in the US and EU. Through a broad document selection methodology, we first collect a broad corpus of documents containing legislative language and policy-relevant discourse on the topic. We then analyze these through inductive coding, and leverage our coding to systematize these documents, identifying key patterns, gaps, and open questions. We identify critical points of disconnect between policy and technological capabilities and practice, and we highlight and discuss potential ambiguities and pitfalls raised by the trends in our corpus. --- ### AI-Model Network: Concept, Current State and Future Authors: Li Zhetao, Zeng Xiyu, Wang Jianhui, Xiao Yong, Liu Zhongren, Wu Junru, Lai Junjie, Huang Jijun, Long Saiqin Published: 2026-05-25 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.27382v1 While the primary function of computers lies in computation and processing, the core value of the Internet is rooted in sharing and collaboration. Computers create the Internet, and the Internet empowers the value of computers. The rapid development of the Internet, cloud computing, and big data is pushing artificial intelligence into the era of large models (LMs). However, the practical application of LMs is currently hindered by high training costs and deployment complexities, driving a shift toward lightweight, private, and domain-specific models. With the rapid proliferation and wide distribution of heterogeneous models, enabling effective interaction and collaboration among them has emerged as a critical bottleneck that urgently needs to be addressed in LM development. Drawing inspiration from the development of the Internet, this paper proposes the concept, vision, and system architecture of world wide AI-model network (AI-ModelNet). It is a novel paradigm that achieves interconnection, capability sharing, and collaborative reasoning by establishing pathways between models. We first briefly review the current state of single-model and multi-model research. Subsequently, the systemic vision and hierarchical architecture of AI-ModelNet are articulated, followed by validation of the framework's feasibility through a prototype system and diverse application cases. Finally, key directions for future research are discussed preliminarily. --- ### SurfDesign: Effective Protein Design on Molecular Surfaces Authors: Fang Wu, Shuting Jin, Xiangru Tang, Mark Gerstein, Xiangxiang Zeng, Yejin Choi, Jure Leskovec, Jinbo Xu Published: 2026-05-25 Categories: q-bio.BM, cs.AI, cs.CE arXiv: https://arxiv.org/abs/2606.07567v1 Protein function is largely determined by molecular surface geometry and physicochemical complementarity, yet most protein design methods condition only on backbone structure. We introduce SurfDesign, a surface-conditioned protein design framework that models molecular surfaces as continuous geometric manifolds and integrates them with pretrained protein language models. SurfDesign employs surface-based equivariant message passing to capture surface normals, curvature, and directional geometry, together with a parameter-efficient fine-tuning strategy. Focusing on functional protein design, we show that SurfDesign consistently outperforms prior surface-conditioned and backbone-only methods on de novo binder and enzyme design benchmarks. We also report strong performance on inverse-folding benchmarks as a diagnostic of structural compatibility. Our results highlight manifold-aware surface representations as a principled foundation for functional protein and enzyme design. Code is available at https://github.com/smiles724/SurfDesign. --- ### STARIXNet: Multivariate and Multi-attribute Deep Learning Approach to Real-Time Resource Allocation in Cloud Platforms Authors: Ahmed Abdulaal, Maruf Aytekin, Thilaga kumaran Srinivasan, Tomer Lancewicki Published: 2026-05-25 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.07565v1 Intelligent scaling of microservices in cloud platforms is crucial for mitigating escalating compute costs while avoiding service disruptions. Current solutions are limited to the univariate space, typically focusing on CPU usage alone to drive scaling decisions. Moreover, they address the problem as a purely forecasting task, focusing on prediction precision while neglecting the greater risks of underestimation and delays in system responsiveness. Alternative solutions are computationally complex, making them impractical for large-scale, real-time deployments. To address these challenges, we present STARIXNet, a lightweight neural network that guides resource allocation decisions in the multivariate space by capturing spatio-temporal relationships among multiple system metrics. STARIXNet models multiple quasi-dependent attributes, in particular the (S)easonal, (T)emporal, (A)uto-(R)egressive (I)ntegrated, and e(X)ogenous patterns, then implements an aggregation policy to finalize scaling decisions, prioritizing service stability, followed by cost-efficiency, over raw forecast accuracy. We empirically demonstrate the performance of STARIXNet by benchmarking against existing solutions in real-world settings. STARIXNet is deployed for critical production microservices at Walmart achieving tangible savings ranging from 10\% to 50\%, in addition to intangible benefits through improved service stability and customer experience. --- ### Considerations for an Integrated Detector Design at FCC-ee: A Human-AI Exploration Authors: Charles Young Published: 2026-05-25 Categories: physics.ins-det, cs.AI, hep-ex arXiv: https://arxiv.org/abs/2606.07564v1 This report explores detector design considerations for the Future Circular Collider in its electron-positron mode (FCC-ee) through an extended dialogue between a physicist and an AI assistant. Starting from initial "prejudice" detector concepts proposed by the AI assistant without explicit physicist input, each subsystem is examined in detail, with the AI's assumptions challenged and revised through the exchange. The discussion covers the full detector from beam pipe to luminosity monitor, with particular attention to the interplay between subsystem choices and the practical considerations - calibration, stability, and operational simplicity - that are essential for a fifteen-year precision physics program. The narrative documents how the integrated detector design evolved substantially from the starting point to revised "prejudice" detector concepts of the AI assistant. The focus of this report is on the process to illustrate both the potential and the limitations of human-AI collaboration in experimental physics design, and the physics capabilities of any of the "prejudice" detector concepts remain to be explored. --- ### Emergence via Phase Transitions: Mechanism Landscapes and Universal Convergence Across Complex Systems Authors: Truong Xuan Khanh Published: 2026-05-25 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.07563v1 Across machine learning, biology, and physics, independently evolving systems often converge toward strikingly similar high-level structures despite radically different microscopic details. Grokking circuits converge across random seeds, evolutionary lineages rediscover similar metabolic solutions, and renormalization flows approach common fixed points. We propose the Hierarchical Emergence Framework (HEF) as a candidate universality framework for such convergence phenomena. HEF models emergence as a phase transition in a mechanism landscape constrained by thermodynamic and information-theoretic laws. The framework introduces a critical energy threshold Ec separating an exploration regime with competing mechanisms from a convergence regime governed by a unique minimum-cost mechanism. Under structural assumptions, we prove physical feasibility, derive strict metric contraction, and establish convergence toward a unique fixed-point representation independent of initial conditions. We further connect this convergence structure to causal emergence through Effective Information and mechanism competition entropy. To test the framework, we study delayed generalization ("grokking") in modular arithmetic transformers across 111 experiments. We identify a reproducible empirical fingerprint of the Ec transition: the weight norm peaks systematically before grokking in 92% of runs. Normalized accuracy curves collapse onto a tanh kink (R^2=0.93) consistent with a Landau-Ginzburg universality class, and all grokked models converge to 0.9745+/-0.014 regardless of initialization, weight decay, or training fraction (ANOVA p>0.13). HEF is not presented as a universal theory of emergence, but as a falsifiable mathematical scaffold for studying convergence phenomena across complex systems. --- ### The Montparnasse Algorithm for RNA Design Authors: Tristan Cazenave Published: 2026-05-25 Categories: q-bio.BM, cs.AI arXiv: https://arxiv.org/abs/2606.07562v1 RNA design consists of discovering a nucleotide sequence that optimizes predefined criteria, such as secondary structure. It is useful for synthetic biology, medicine, and nanotechnology. We propose Montparnasse, a Monte Carlo search framework based on Generalized Nested Rollout Policy Adaptation, augmented with a problem-specific prior, slow and long adaptation at level 1, and a lexicographic multicriteria evaluation. Montparnasse solves all 100 puzzles of the Eterna100 V1 benchmark consistently faster than DesiRNA, the previous state of the art, across all time limits, reaching full coverage more than three times faster overall. On messenger RNA secondary structure optimization for hemoglobin alpha, it identifies sequences with more paired bases than the MFE-optimal solution of LinearDesign. --- ### Boundary Variance Inflation Causes Acquisition Bias in Gaussian Processes Authors: Maria Bånkestad, Sanna Jarl, Jens Sjölund Published: 2026-05-25 Categories: cs.LG, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2606.07561v1 Gaussian processes with stationary kernels on bounded domains exhibit inflated posterior variance near the boundary. Despite being a long-recognized artifact in geostatistics and a source of over-exploration in Bayesian optimization, the causes and effects of boundary-induced acquisition bias are underexplored. We trace the root cause to a simple geometric mechanism: the truncation of the kernel correlation neighborhood at the domain boundary creates an observation-independent distortion that worsens with dimensionality. We show how this distortion manifests across three acquisition classes: variance maximization concentrates selections at the corners, whereas negative integrated posterior variance and expected predictive information gain move selections inward to axis-aligned interior shells. These patterns arise without reference to any objective function, meaning that acquisition behavior can be dominated by kernel geometry rather than the desired task-specific uncertainty. To quantify this, we introduce a function-free selection-profile diagnostic for arbitrary acquisitions, kernels, and bounded-domain geometries. --- ### Function-Vector Heads Are Two Populations: Writers and Cancellers in In-Context Learning Authors: Han-yu Wang Published: 2026-05-25 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.07560v4 Function-vector (FV) analyses commonly identify attention heads by the magnitude of their causal contribution to in-context tasks. Magnitude does not retain the direction of the effect on the task readout. We preserve the sign and validate candidate heads with path patching. Across six main Pythia (model, task) cells, the validated population separates into writers, whose direct effects favour the rule-correct label, and cancellers, whose effects oppose it. Labels assigned on path-patching prompts predict held-out group lesions, and a sign-shuffle null rejects a chance partition in five cells. Measurements not used to assign the roles show corresponding structure. Writers attend more to demonstration labels, cancellers more to format tokens, and their OV write directions are shifted toward opposition relative to same-layer controls. A magnitude-ranked mean-ablation baseline preferentially recovers cancellers on the hierarchical task and writers on the modular task. Signed lesion directions recur in all fifteen cells tested across six Pythia scales and three architectures. Cross-template transfer shows that these are task-conditioned roles that can persist, weaken, or reverse. Zero-ablating cancellers raises the correct-label logit difference by +0.13 to +0.29 nats in all six main cells, with accuracy point estimates increasing in all six. Together, the results separate causal importance from functional role and show that a function vector can remain a useful task-level representation while its head-level causal implementation contains opposed, task-conditioned components. --- ### Function-Vector Heads Are Two Populations: Writers and Cancellers in In-Context Learning Authors: Han-yu Wang Published: 2026-05-25 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.07560v3 Function-vector (FV) heads (Todd et al., ICLR 2024) are identified by the magnitude of their causal contribution to in-context rule tasks, and the resulting top set is treated as a single functional class. We show that it holds two. Under a sign-preserving criterion (refined direct logit attribution, validated head by head with path patching) the FV population splits into writers, which push the rule-correct logit up, and cancellers, which push it down. The two groups are mechanistically distinct: writers place a third to a half of their attention on the demonstration labels, cancellers shift 9 to 17 points of that mass off the labels and mostly onto the format tokens, and the two groups' write directions are more anti-aligned than same-layer controls. Their oppositely signed effects combine additively at the readout, so they partly cancel and the FV set understates what its writers do. Magnitude-only ranking surfaces whichever group locally dominates and misses the other, so any function vector or ablation built that way averages a promoting and a suppressing mechanism. The signed split holds in all fifteen (model, task) cells we test, spanning six Pythia scales and three architectures, and a sign-shuffle null rejects a chance split in five of the six cells that carry the full statistical pipeline. Zero-ablating the cancellers recovers +0.13 to +0.29 nats on the correct label in all six and shifts accuracy by +2 to +7 pp. --- ### Phantom Transitions in Language Model Fine-Tuning: A Density-Matrix Analysis Authors: Vaibhav Prakash, Jayasri Dontabhaktuni Published: 2026-05-25 Categories: cs.CL, cs.AI, quant-ph arXiv: https://arxiv.org/abs/2606.07559v3 Language models fine-tuned where the correct completion must outrank a near-synonym competitor often fail silently. The cross-entropy loss falls monotonically while the correct token never overtakes the competitor in the model's ranking. We study this across five transformer architectures from two families spanning a sixfold parameter range, on ten contexts whose correct and competing completions share substantial embedding overlap. We build an order parameter combining the predicted distribution with embedding overlap, as a density matrix because that distribution lives over a non-orthogonal basis. It decomposes additively into a signal term tracking commitment to the correct token and a drag term set by how the embedding bulk leaks probability into the score. This isolates two failure modes. In kinematic failure the signal stays too small and the model never commits. In structural failure the drag worsens during fine-tuning, so the model degrades geometrically as its loss falls. The order parameter also shows sharp jumps resembling phase transitions. We test the spontaneous-symmetry-breaking reading by tracking it after every gradient step, and rule it out. The jumps persist under LoRA even though the token embedding matrix never changes. No geometric phase transition is possible when that geometry cannot move, so the discontinuity lies entirely in the softmax readout. A few dimensionless quantities organize the trajectory across architectures. One is consistent across all five models under full fine-tuning. A second sorts architectures into two classes by their bulk embedding distribution and predicts whether LoRA alone can make a sentence commit. As a blind test, the framework predicts a held-out architecture's critical learning rate to within 2.1% of a later sweep. These results characterize this near-synonym mechanism and need recalibration before extrapolation. --- ### Phantom Transitions in Language Model Fine-Tuning: A Density-Matrix Analysis Authors: Vaibhav Prakash, Jayasri Dontabhaktuni Published: 2026-05-25 Categories: cs.CL, cs.AI, quant-ph arXiv: https://arxiv.org/abs/2606.07559v2 Fine-tuning a language model often fails silently when its correct completion must outrank a near-synonym competitor. Cross-entropy loss falls monotonically while the correct token never overtakes the competitor. We study five transformer architectures of two families over a fivefold parameter range, on ten contexts where the correct completion and nearest competitor overlap substantially. We construct an order parameter combining the predicted distribution with the geometric overlap between token embeddings. The density-matrix formulation is natural because the prediction distribution lives over a non-orthogonal embedding basis. It decomposes additively into a signal, tracking commitment to the correct token over its nearest competitor, and a background drag, set by how the embedding bulk leaks probability into the score. This isolates two failure modes. In kinematic failure the signal stays small and the model never commits. In structural failure the drag actively worsens as fine-tuning proceeds and the model degrades geometrically even as its loss decreases. Sharp jumps resemble phase transitions but are not. Tracking it after every gradient step rules out spontaneous symmetry breaking. Under LoRA fine-tuning the embedding matrix is frozen, so no geometric phase transition is possible and the discontinuity lies in the softmax readout. One dimensionless quantity is consistent across all five models under full fine-tuning, while a second sorts them into two classes by bulk embedding distribution and predicts whether LoRA alone can drive a sentence to commit. As a blind test, the framework predicts the critical learning rate of a held-out architecture, used to fit no parameter, to within 2.1% of a subsequent sweep. All findings concern the near-synonym geometric mechanism studied here and should not be extrapolated without re-calibration to instruction tuning or broader tasks. --- ### Page image classifier fine-tuned on century-spanning archives of scanned documents for further content-specific processing Authors: Kateryna Lutsai, Dana Křivánková, Pavel Straňák, David Novák Published: 2026-05-25 Categories: cs.CV, cs.AI, cs.DL arXiv: https://arxiv.org/abs/2606.07558v2 Purpose: Digitization projects in the humanities produce vast, heterogeneous archives of historical documents, making manual sorting impractical at scale. This work addresses the need for an automated system to classify scanned page images based on visual content type - text, tables, and graphics - enabling content-specific downstream processing such as Optical Character Recognition (OCR) or structured data extraction. Methods: An image classification system was developed and evaluated on a dataset of over 48,000 annotated historical page images from century-old Czech archaeological archives, refined through four successive annotation stages with domain-expert review. A Random Forest Classifier baseline was established using hand-crafted image features. Subsequently, deep learning architectures were fine-tuned and compared: Convolutional Neural Networks (EfficientNetV2, RegNetY), Vision and Document Image Transformers (ViT, DiT), and multimodal CLIP models. An 11-category label scheme was designed collaboratively with domain experts and evaluated via five-fold cross-validation. Results: The feature-based baseline achieved approximately 75% accuracy. Fine-tuned CNNs and Transformers substantially outperformed it, with RegNetY-16GF achieving 99.16% and ViT-large 99.12% Top-1 accuracy on the held-out test set. CLIP ViT-B/16 reached 99.14% with optimized text descriptions. Conclusion: Image-only models, particularly RegNetY-16GF, deliver near-perfect classification accuracy and produce consistent labels across 649,508 unlabeled archival pages with over 90% inter-model agreement. Fine-tuned CLIP, despite competitive test-set accuracy, showed under 65% agreement with image-only models on unlabeled data, making it less suitable for deployment. The final models, annotated dataset, and software are publicly available under open-source licenses. --- ### SPIN: Decentralized Swarm Control via Tensorized Policy Coordination Authors: Zhaowen Fan, Yunxiang Han Published: 2026-05-25 Categories: cs.LG, cs.MA, cs.SI arXiv: https://arxiv.org/abs/2606.07557v2 Decentralized multi-agent swarm coordination remains fundamentally challenged by the combinatorial scaling of joint action spaces and high-overhead training or optimization loops when managing localized group behaviors. To address this problem from a different perspective, this paper introduces the Swarm Policy Interference Network (SPIN) framework, which models multi-agent communication topologies as compressed tensor networks. By factorizing the joint policy tensors of local multi-agent cliques into Open Boundary Condition Matrix Product State (MPS) chains, SPIN replaces explicit exponential joint-action enumeration with clique-level contractions that scale linearly in clique length for fixed local behavior and bond dimensions. To connect raw spatial geometry with this discrete algebraic backend without relying on online training loops, we introduce a decoupled framework combining a lightweight frozen neural mapper evaluated offline with a deterministic zero-shot importance-reweighting filter based on the Radon-Nikodým derivative. We evaluate an executable prototype of this framework within a simulation experiment across distinct task regimes: single-target tracking, decentralized area coverage, and structured multi-goal coordination. The results demonstrate that SPIN functions as a reusable decentralized coordination layer across tracking, dispersion / area coverage, and structured multi-goal coordination, with its strongest gains appearing in multi-goal coordination and dense local interaction regimes, without requiring scenario-specific online optimization or retraining. --- ### Selecting New Measurement Locations to Diversify Traffic-Pattern Coverage: A Real-World Evaluation for Total Traffic Volume Estimation Authors: Masaaki Inoue, Akifumi Okuno, Shintaro Fukushima Published: 2026-05-25 Categories: cs.NI, cs.AI, stat.ME arXiv: https://arxiv.org/abs/2606.07556v1 Accurate measurement of traffic volumes and flows is vital for modern intelligent transportation. However, despite recent technological advances in sensor devices, it is still expensive to install and maintain fixed traffic counters. Therefore, it is restricted to a small portion of location points where the counters can be installed, which severely limits the possibility of grasping and predicting the total traffic volume at a city-wide level. By contrast, devices with location history such as smartphones and connected vehicles are now widely used and provide much wider spatial coverage. However, the data from these devices are usually partial and noisy, so they are not enough to directly estimate total traffic volumes and flows. In this paper, we use the information from these widely available devices to help decide where to place additional traffic counters, and we study how selecting new measurement locations can improve city-wide traffic estimation performance. To achieve this, we propose an algorithm that chooses additional counter locations to increase the diversity of observed traffic signal patterns, rather than simply spreading counters evenly over space. The goal is to capture traffic-pattern types that are rare in the current counter set and to make the collected observations more representative for later estimation and forecasting. We also present a real-world evaluation; in a target city, we select new locations expected to improve traffic prediction, and we then commissioned new field measurements at those locations at our expense. The resulting data led to an improvement in traffic volume estimation accuracy across different fidelities. --- ### Persistent Priors, Preserved Targets: A Stroop-Style Paradigm for Lexical Override Authors: Han-yu Wang Published: 2026-05-25 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.07555v5 Local definitions can assign a familiar word a temporary meaning while its usual associations remain useful elsewhere. We measure interference from those associations with a matched Stroop-style paradigm. A conflict prompt defines doctor as forest and compares forest with the familiar associate hospital. A neutral control replaces doctor with a semantically weak word in both the definition and query while keeping forest and hospital fixed. All 11 model-level means are positive. Aggregate means are also positive for all four conflict families and prompt formats. When no redefinition is present, a stronger preference for the familiar distractor predicts more interference in arbitrary-semantic, polysemy/entity, and domain-definition remappings, while the antonym slope is null. Separately, we patch neutral-control activations into antonym prompts in five 1B-2B models. Patching the defined word, the target word in the definition, and the later query word together restores almost all of the target-minus-distractor margin lost in conflict (normalized recovery R in (0.92,1.06)). Replacing only that target-word activation with a donor from another item reduces recovery in every tested case. Those donor patches also lower the distractor logit, while the contextual target falls much more than under the same-item patch that restores the margin. --- ### Priors Persist Through Suppression: A Stroop Paradigm for Lexical Override Authors: Han-yu Wang Published: 2026-05-25 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.07555v4 Glossaries, technical specifications, and system prompts routinely ask language models to use familiar words in unfamiliar ways. The instruction competes with what the word already means, and even when it wins, the pretrained prior keeps operating underneath. We test this with a Stroop-style paradigm: a prompt redefines a word (doctor now means forest), asks for a related word, and we score the new meaning against the word's pretrained associate (hospital) under matched neutral controls. Across 11 open-weight models from 1B to 9B parameters, the old meaning interferes in every model, remapping type, and prompt framing we test. After item-level controls, a model's ordinary preference for the old associate predicts the size of the interference, in the three document-relevant remapping types though not in antonyms. On antonym remapping, activation patching in five models locates the repair: restoring three prompt positions (where the word is redefined, where its new meaning appears, and where the question repeats it) recovers almost all of the effect (normalized recovery R in (0.92, 1.06)). The repair is asymmetric. The old meaning's logit falls under any perturbation of those positions, so pushing it down is not what separates a working override from a failing one; the new meaning survives only while the position carrying it is intact. What a local definition achieves is a protected new meaning, not a suppressed old one. --- ### Which Anatomy Matters Under Limited Labels? A Data-Efficient Anatomy-Aware Benchmark for Cardiac Pathology Prediction Authors: Himanshu Singh Published: 2026-05-25 Categories: eess.IV, cs.AI, cs.LG, q-bio.TO arXiv: https://arxiv.org/abs/2606.06509v1 Numerous medical imaging problems must be solved under limited labels and constrained compute, yet it remains unclear whether performance gains are driven mainly by more expressive models or by better representation of clinically meaningful anatomy. We study this question through a low-data anatomy-aware benchmark for 5-class cardiac pathology prediction on the public ACDC MRI dataset. Using segmentation-derived patient descriptors from the right ventricle, myocardium, and left ventricle, we compare anatomy-specific and multi-structure representations across linear, kernel, and tree-based classifiers. We find that under limited label settings, representation dominates complexity. These results suggest that in resource-constrained healthcare settings, identifying and representing the most informative anatomy may matter more than the increasing complexity of the model alone. --- ### Segmentation-Guided Spatial Indexing for Generalizable and Explainable Deepfake Detection Authors: Izaldein Al-Zyoud, Abdulmotaleb El Saddik Published: 2026-05-25 Categories: cs.CV, eess.IV arXiv: https://arxiv.org/abs/2606.00098v1 We introduce segmentation-guided spatial indexing for generalizable and explainable deepfake detection. The key idea reverses the standard design order: rather than pooling all facial tokens and classifying afterward, we first select semantically meaningful patch tokens, then pool only those. A frozen FaRL parser assigns each DINOv3 ViT-L/16 patch token a semantic label; non-target tokens are discarded; a linear probe classifies the retained region. This spatial indexing exploits DINOv3's patch-level spatial consistency, the same property that enables emergent segmentation, to present the probe with a purer regional subspace where manipulation-relevant evidence is less diluted by whole-face cues. Region attribution is structural: when the mouth model predicts fake, the decision used only mouth tokens, not an overlaid saliency map. On Celeb-DF v2, the mouth-indexed probe achieves AUC 0.905, outperforming LipForensics (+8.1 pp) and Xception (+16.9 pp), with no DINOv3 or FaRL fine-tuning and no target-domain data. Ablations isolate the mechanism: replacing regional selection with DINOv3's CLS token drops Celeb-DF v2 AUC by 26.4 pp; replacing DINOv3 with FaRL features drops it by 20.9 pp. Both DINOv3 representation and the spatial index are independently necessary; neither alone approaches the full system. --- ### Diversity Over Frequency: Rethinking Tool Use in Visual Chain-of-Thought Agents Authors: Dong-Hee Kim, Reuben Tan, Donghyun Kim Published: 2026-05-25 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.00096v2 Visual agents employ external visual tools within visual chains of thought to incorporate fine-grained evidence. While prior work has mainly studied these tools in visual search tasks, their role in more complex visual reasoning remains underexplored. In this paper, we move beyond simple visual search tasks to investigate more challenging tasks, including 3D spatial reasoning and medical visual question answering, where agents must integrate tool-acquired local evidence with the global context. We identify a {tool-use collapse phenomenon: models progressively stop using tools while still achieving higher task accuracy. Moreover, we observe a clear asymmetry: (i) completely eliminating tool use degrades performance, whereas (ii) incentivizing tool use yields only marginal gains despite substantially increasing usage. We find that vanilla training and tool-use encouragement both reduce rollout diversity, explaining why higher tool use does not yield stronger reasoning performance. Motivated by these findings, we add an entropy regularization term to encourage diverse rollout exploration, achieving the best performance despite gradually declining tool usage. Overall, our findings suggest a training-time view of tools as scaffolding, where broader exploration over language generation and visual tool invocation improves reasoning despite tool-use collapse. Project page: https://scaffolded-exploration.github.io --- ### Bridging the 2D-3D Gap: A Hierarchical Semantic-Geometric Map for Vision Language Navigation Authors: Kailing Li, Tianwen Qian, Lijin Yang, Yuqian Fu, Jingyu Gong, Xiaoling Wang, Liang He Published: 2026-05-25 Categories: cs.CV, cs.AI, cs.CL, cs.RO arXiv: https://arxiv.org/abs/2606.00095v1 Vision-Language Navigation (VLN) enables embodied agents to reach target locations in unseen environments by following language instructions. Despite recent progress with vision-language models (VLMs), a critical semantic-geometric gap remains: while VLMs excel at language and 2D visual understanding, they struggle with 3D spatial reasoning and fail to capture the causal dynamics between actions and spatial transitions, resulting in unreliable navigation, particularly in zero-shot settings. To bridge this gap, we propose a Hierarchical Semantic-Geometric Map (HSGM) that transforms 3D geometric information into a structured representation compatible with VLMs, effectively linking them to the physical world. Specifically, HSGM is represented as a multi-channel top-down map organized into three levels: (1) geometric level that records navigable regions and obstacles, (2) semantic level that represents objects and their relations, and (3) decision level that supports high-level task reasoning and goal selection. During navigation, the VLM acts as a high-level semantic planner, interpreting the spatial layout encoded in the HSGM to select geometrically valid waypoints, while low-level, collision-free movements between waypoints are executed by a classical path-planning algorithm, fully decoupling semantic reasoning from action execution. Additionally, complex instructions are decomposed into subtasks to alleviate the problem of progress forgetting or hallucinating in long-horizon navigation. Extensive experiments on R2R-CE and RxR-CE benchmarks demonstrate that our zero-shot framework achieves state-of-the-art performance and even outperforms several supervised methods. Code is available at https://github.com/Teacher-Tom/HSGM_public. --- ### Diffusion Image Generation with Explicit Modeling of Data Manifold Geometry Authors: Duoduo Xue, Zhiyu Zhu, Junhui Hou Published: 2026-05-25 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.00094v3 Image generative models aim to sample data points from the underlying data manifold, a task that requires learning and decoding a dense, low-dimensional, and compact parameterization space. To achieve this, we propose the Data Manifold-aware Image diffusioN moDel (MIND), a novel framework that explicitly models manifold geometry by integrating discrete patch tokenization into the score function of a continuous diffusion model. This approach successfully leverages both the structural quantification capabilities of discrete tokens and the parallel generation flexibility of continuous diffusion. Moreover, we enable end-to-end differentiable training via a novel soft top-$k$ aggregation mechanism and introduce dual-branch high-frequency feature embedding layers to alleviate the spectral bias of transformer backbones on low-dimensional inputs. Furthermore, for inference, we design a multi-stage transition sampling scheme that dynamically adjusts the sampling scheme based on timestep. Extensive experiments on ImageNet 256$\times$256 demonstrate the effectiveness of MIND. After 80-epoch training, our base model achieves an FID of 22.73 without guidance, nearly halving the 43.47 FID of the vanilla DiT-B/2 baseline. The proposed method reduces FID by 15.95 and 9.06 on average compared with the baselines DiT and SiT, respectively. For image generation on ImageNet-256$\times$256 with guidance, the proposed MIND-B with only 130M parameters achieves an FID of 2.06, superpassing the LlamaGen-3B with 3.1B parameters. The proposed MIND-XL with 715M parameters further reduces the FID to 1.95. Our MIND introduces a fresh perspective on diffusion-based image generation, paving the way for future research and innovation in this community. The code will be publicly available. --- ### Agreement Metrics for LLM-as-Judge Evaluation: What to Report and Why Authors: Delip Rao, Chris Callison-Burch Published: 2026-05-25 Categories: cs.CL, cs.HC, physics.data-an arXiv: https://arxiv.org/abs/2606.00093v2 Whether a rubric-based LLM judge can replace human annotation is decided by its measured agreement with human labels. Yet the same verdicts can support wildly varying agreement numbers, depending on seemingly minor choices: the judgment scale, the retained cases, the handling of abstentions and invalid outputs, and the pooling of verdicts across items and rubric criteria. The statistics that settle these choices are established, but in psychometrics, econometrics, and corpus annotation rather than in the evaluation practice that needs them. We treat the choices as a measurement protocol that fixes what the reported number estimates before any metric is computed, assemble the relevant results into a single source-attributed analysis, and apply it to three published LLM-judge evaluations. For non-degenerate binary verdicts, Pearson's $r$, Spearman's $ρ$, Kendall's $τ_b$, the phi coefficient, and the Matthews correlation coefficient are exactly the same statistic, so reporting several repeats one number under different names. Cohen's $κ$ differs from them only through a marginal-mismatch factor in $(0,1)$ and shares their asymptotic variance when judge and human assign the positive verdict equally often. Under exclusion, accuracy over all cases is pinned down only to a worst-case interval as wide as the uncovered fraction. On a rubric benchmark carrying per-criterion human labels, protocol choice alone moves reported accuracy from $0.551$ to $0.899$ and carries $κ$ across zero, without altering a single verdict. We distill the analysis into a reporting checklist that makes agreement claims reconstructible and comparable. --- ### From Data to Insights: Exploring Program-of-Thoughts Prompting for Chart Summarization Authors: Yutong Qu, Wei Zhang Published: 2026-05-25 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.28874v1 Charts play a critical role in conveying numerical data insights through structured visual representations. However, semantic visual understanding and numerical reasoning requirements hinder the accurate description of charts, interpreting a challenging task in chart summarization. Despite recent advancements in visual language models (VLMs), approaches lack robust mechanisms for verifying statistical fact correctness and are computationally heavy. To address this gap, this paper explores a strategy of using zero-shot learning to motivate the lightweight VLMs to perform computational reasoning, via Python programs as intermediaries to derive valid summary statistics for chart understanding. Specifically, we introduce a novel chart-to-dictionary auxiliary task, offering a more flexible representation compared to traditional chart-to-table methods, making it particularly well-suited for integration with the Program-of-Thought (PoT) strategy. Experimental results demonstrate our strategy performs on par with existing chart summarization methods across semantic and factual metrics. Code is available on https://anonymous.4open.science/r/ZeroShot-PoT-C2T-5A6B. --- ### Pre-Registering the Detectable Effect: A Paired-MDE Budget for 4-bit Quantization Benchmarks, with a Pilot Audit Authors: Zexin Zhuang, Yanhang Li, Zhichao Fan Published: 2026-05-25 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.28873v1 This is a planning-method note with an unpaired pilot audit. We adapt the classical paired-binary sample-size calculation (Miettinen, 1968) to quantization benchmarks, giving a conservative minimum detectable effect (MDE) bound $δ^{*} \le (z_{1-α/2}+z_{1-β})\sqrt{ρ_d/m}$ in the paired item count $m$ and the FP16-NF4 disagreement rate $ρ_d$. The bound turns "how reliable is my quantization claim?" into a one-line budget a benchmark designer can commit to before running. We illustrate the bound on four models and four benchmarks ($k=5$ splits of $n=100$), and add a parallel MMLU prompt-template study to put the bound's quantization-noise scale alongside the prompt-noise scale. Assuming $ρ_d=0.10$ (an unmeasured planning value), all observed NF4-FP16 deltas fall below the implied MDE, and most cross-split SDs lie within $\pm 1.5$ pp of the binomial reference $\sqrt{p(1-p)/n}$, so much of the variance reported as "benchmark unreliability" on $n=100$ subsamples is binomial sampling noise. The single borderline cell (OPT-WinoGrande, $|Δ|=3.2$ pp) is below the implied MDE at $ρ_d=0.10$ but above it at $ρ_d=0.05$, illustrating the planning trade-off the bound makes explicit. On MMLU, prompt-template ranges of 2-10 pp meet or exceed the largest observed quantization delta (3.2 pp), so a quantization audit that does not first fix the prompt template absorbs template variance into its noise floor. We complement the bound with a five-line pre-registration template. --- ### D$^2$Turb: Depth-Aware Simulation and Decoupled Learning for Single-Frame Atmospheric Turbulence Mitigation Authors: Zixiao Hu, Tianyu Li, Guoqing Wang, Wei Li, Guoguo Xin, Xun Liu, Peng Wang Published: 2026-05-25 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.27460v1 Single-frame atmospheric turbulence mitigation is inherently ill-posed due to spatially varying blur coupled with non-rigid geometric distortion. Existing end-to-end approaches trained on flat-field simulations often struggle to balance texture recovery with geometric rectification. To overcome this limitation, we propose D$^2$Turb, a unified framework that bridges physics-grounded simulation with explicitly decoupled restoration. First, we introduce a Depth-Aware Turbulence Synthesis protocol that incorporates scene depth into the phase-to-space formulation. This generates physically consistent, depth-dependent degradations and provides a crucial intermediate tilt supervision signal for disentangled learning. Building upon this simulation engine, D$^2$Turb decomposes restoration into two interactive stages: texture deblurring and geometric rectification. The texture deblurring stage employs a deblurring backbone to recover fine-grained details while preserving geometric distortion for the subsequent rectification stage. To mitigate the information fragmentation commonly observed in cascaded designs, we further propose an Adaptive Structural Prior Injection (ASPI) mechanism that dynamically transfers deep structural representations from the deblurring module to guide dense flow prediction for spatial unwarping. Extensive experiments demonstrate that D$^2$Turb achieves state-of-the-art performance on both synthetic and real-world datasets, with consistent improvements in both texture recovery and geometric fidelity. Our code and pre-trained models are publicly available at https://github.com/HertzDot222/D2Turb. --- ### Generic Interpretation Approach for Transformer Models Incorporating Heterogenous Attention Structures Authors: Yongjin Cui, Xiaohui Fan, Huajun Chen Published: 2026-05-25 Categories: cs.CV, cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2605.27458v2 Transformer has significantly propelled the development of artificial intelligence, and certainly the development of agents as well. We categorize attention structures of Transformer into two types based on the source of the input information: homogenous and heterogenous attention structures. Heterogenous attention structures, with co-attention as a typical example, process information from different sources. Heterogenous attention structure is the foundation for Transformer models to achieve more complex functions and integrate more modal information. Whether for research purposes or policy requirements, the interpretation of Transformer models with heterogenous attention structures is an important task. The fusion of information from different sources brings new challenges. Our work mainly includes two parts: method and experimentation. In terms of method, we propose an interpretation method for Transformer models with heterogenous attention structures. In terms of experimentation, based on our experimental analysis paradigm, we interpret the operating mechanisms of representative models, conduct semantic interpretation and logical interpretation. --- ### Metric-Aware PCA as a Linear Instance of Geometric Deep Learning Authors: Michael Leznik Published: 2026-05-25 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.27456v1 Geometric deep learning organises neural architectures around the symmetries of their data domain, with the choice of symmetry group serving as a geometric prior that determines what representations can be learned. Metric-Aware Principal Component Analysis (MAPCA) parameterises principal component analysis by a positive-definite metric matrix, with a canonical subfamily interpolating between standard PCA and output whitening and a diagonal-metric point recovering Invariant PCA (IPCA). This paper positions MAPCA within the geometric deep learning framework. The metric is read as the geometric prior; the orthogonal group preserving it is the symmetry group it induces; MAPCA solutions are equivariant under this group with the resulting spectrum invariant; and MAPCA's defining constraint is the linear analogue of the Schur-type weight constraints used in equivariant networks. Across six axes - domain, symmetry group, equivariance, invariance, architectural primitive, and geometric prior - we construct a precise dictionary between MAPCA and geometric deep learning. The technical anchor is a uniqueness theorem characterising IPCA as the unique linear data-derived metric in the MAPCA family that is equivariant under arbitrary diagonal rescaling and projects onto the fixed-point set of the action, equivalent under normalisation to the variance-maximisation criterion in its precise form. The paper closes with three bridges: kernel PCA as the nonlinear extension, spectral graph methods as MAPCA on graphs, and a deep MAPCA construction extending the positioning into deep equivariant networks --- ### NL-MambaXCT: Self-Supervised Nested-Learning Mamba for Nomex Honeycomb X-ray CT Defect Classification Authors: Ghaleb Aldoboni, Lobna Nassar, Fakhri Karray, Reem Alshamsi Published: 2026-05-25 Categories: eess.IV, cs.CV arXiv: https://arxiv.org/abs/2605.27454v2 X-ray computed tomography (XCT) is widely used for non-destructive testing of Nomex honeycomb structures in aerospace manufacturing, but industrial inspection still relies heavily on manual interpretation and supervised models trained on limited labeled data. This work introduces NL-MambaXCT, a Mamba-based framework that combines self-supervised masked image modelling with a Nested Learning (NL) formulation for automated, label-efficient defect classification from production XCT slices. The backbone is a four-stage 2D encoder with RegNet convolutional blocks in the early stages and Mamba-based sequence mixing with attention in the deeper stages. It is pretrained by masked image modelling on 19,961 unlabeled industrial XCT slices and fine-tuned on 2,000 relabeled Nomex XCT slices split by production order. NL is instantiated through two-timescale parameter dynamics: selected projections maintain slow exponential-moving-average traces alongside fast weights, while a deep-momentum optimizer introduces an additional slow parameter-update trajectory. On the held-out test set, the MIM-pretrained NL-MambaXCT model achieves 96.91% accuracy and 96.8% macro F1, outperforming CNN, attention, and single-timescale Mamba baselines by 3.11--10.31 percentage points in accuracy. The results suggest that combining masked self-supervision with NL-style fast/ slow learning dynamics is a promising strategy for robust defect classification in Nomex honeycomb XCT inspection. --- ### MiRD: Reliable Set-Valued Prediction for Open-Ended Question Answering via Miscoverage Risk Decomposition Authors: Anqi Hu, Zhiyuan Wang, Zijun Jia, Bo Fu Published: 2026-05-25 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.27091v1 Reliable set-valued prediction provides a principled way to mitigate hallucinations in open-ended question answering (QA), yet existing conformal approaches typically rely on a fragile premise: finite sampling must already produce at least one admissible candidate, or calibration examples violating this condition are discarded. In this paper, we introduce MiRD, a two-stage framework that decomposes overall miscoverage into sampling failure and conditional selection failure. In Stage I, MiRD establishes an expectation-level marginal upper bound on the probability that finite sampling produces no admissible answer under a fixed budget. In Stage II, conditioned on sampling success, MiRD calibrates a conformal selection threshold using admission-correlated nonconformity scores defined over the full calibration set, thereby preserving calibration-set integrity. Across three open-ended QA datasets and eight models, MiRD controls sampling risk, conditional selection risk, and overall miscoverage, while yielding tighter first-stage bounds than PAC-style alternatives and more adaptive prediction sets than successful-only calibration. --- ### Advancing Creative Physical Intelligence in Large Multimodal Models Authors: Cheng Qian, Hyeonjeong Ha, Jiayu Liu, Jeonghwan Kim, Emre Can Acikgoz, Bingxuan Li, Kunlun Zhu, Jiateng Liu, Aditi Tiwari, Zhenhailong Wang, Xiusi Chen, Mahdi Namazifar et al. Published: 2026-05-25 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2605.26396v2 Large multimodal models (LMMs) have rapidly advanced in perception and reasoning; however, it remains unclear whether these capabilities generalize to discovering visually grounded solutions in open-ended environments, beyond pattern recognition. In such settings, intelligence requires more than answering well-posed questions: it involves identifying how elements in a scene can be repurposed in non-obvious yet physically feasible ways. This form of creative problem-solving is central to human intelligence, but remains largely untested in current benchmarks. To evaluate this ability, we introduce MM-CreativityBench, a benchmark for affordance-grounded creative tool use in visually rich, physically constrained environments. Each instance presents a scenario image with structured views of candidate entities and their parts, enabling fine-grained, interactive evaluation of how models iteratively inspect the scene, identify relevant affordances, and compose visually and physically grounded solutions. Our experiments show that current LMMs often fall short, not due to lack of generative capability, but because they do not sustain grounded exploration. Models often overlook relevant entities, under-examine critical parts, or hallucinate attributes not grounded in the image. Motivated by this failure mode, we propose affordance-grounded alignment, which casts creative tool use as a preference learning problem. Using Direct Preference Optimization, we encourage models to prefer attribute-affordance reasoning grounded in visual evidence over hallucinated alternatives. In addition, we incorporate supervision derived from an affordance knowledge base to guide broader entity exploration and multi-turn planning. Our results show consistent gains in selecting the correct entities and parts, while substantially reducing hallucination and grounding-related errors. ---