🤖 AI 资讯

arXiv:2603.00267v1 Announce Type: new Abstract: Misinformation spreading over the Internet poses a significant threat to both societies and individuals, necessitating robust and scalable fact-checking that relies on retrieving accurate and trustworthy evidence. Previous methods rely on semantic and social-contextual patterns learned from training data, which limits their generalization to new data distributions. Recently, Retrieval Augmented Generation (RAG) based methods have been proposed to utilize the reasoning capability of LLMs with retrieved grounding evidence documents. However, these methods largely rely on textual similarity for evidence retrieval and struggle to retrieve evidence that captures multi-hop semantic relations within rich document contents. These limitations lead to overlooking subtle factual correlations between the evidence and the claims to be fact-checked during evidence retrieval, thus causing inaccurate veracity predictions. To address these issues, we propose WKGFC, which exploits authorized open knowledge graph as a core resource of evidence. LLM-enabled retrieval is designed to assess the claims and retrieve the most relevant knowledge subgraphs, forming structured evidence for fact verification. To augment the knowledge graph evidence, we retrieve web contents for completion. The above process is implemented as an automatic Markov Decision Process (MDP): A reasoning LLM agent decides what actions to take according to the current evidence and the claims. To adapt the MDP for fact-checking, we use prompt optimization to fine-tune the agentic LLM.

arXiv:2603.00285v1 Announce Type: new Abstract: Evaluating AI agents in finance faces two key challenges: static benchmarks require costly expert annotation yet miss the dynamic decision-making central to real-world trading, while LLM-based judges introduce uncontrolled variance on domain-specific tasks. We introduce TraderBench, a benchmark that addresses both issues. It combines expert-verified static tasks (knowledge retrieval, analytical reasoning) with adversarial trading simulations scored purely on realized performance-Sharpe ratio, returns, and drawdown-eliminating judge variance entirely. The framework features two novel tracks: crypto trading with four progressive market-manipulation transforms, and options derivatives scoring across P&L accuracy, Greeks, and risk management. Trading scenarios can be refreshed with new market data to prevent benchmark contamination. Evaluating 13 models (8B open-source to frontier) on ~50 tasks, we find: (1) 8 of 13 models score ~33 on crypto with <1-point variation across adversarial conditions, exposing fixed non-adaptive strategies; (2) extended thinking helps retrieval (+26 points) but has zero impact on trading (+0.3 crypto, -0.1 options). These findings reveal that current agents lack genuine market adaptation, underscoring the need for performance-grounded evaluation in finance.

arXiv:2603.00309v1 Announce Type: new Abstract: The increasingly popular agentic AI paradigm promises to harness the power of multiple, general-purpose large language model (LLM) agents to collaboratively complete complex tasks. While many agentic AI systems utilize predefined workflows or agent roles in order to reduce complexity, ideally these agents would be truly autonomous, able to achieve emergent collaboration even as the number of collaborating agents increases. Yet in practice, such unstructured interactions can lead to redundant work and cascading failures that are difficult to interpret or correct. In this work, we study multi-agent systems composed of general-purpose LLM agents that operate without predefined roles, control flow, or communication constraints, relying instead on emergent collaboration to solve problems. We introduce the Dynamic Interaction Graph (DIG), which captures emergent collaboration as a time-evolving causal network of agent activations and interactions. DIG makes emergent collaboration observable and explainable for the first time, enabling real-time identification, explanation, and correction of collaboration-induced error patterns directly from agents' collaboration paths. Thus, DIG fills a critical gap in understanding how general LLM agents solve problems together in truly agentic multi-agent systems. The project webpage can be found at: https://happyeureka.github.io/dig.

arXiv:2603.00312v1 Announce Type: new Abstract: While multimodal large language models offer a promising solution to the "black box" nature of health AI by generating interpretable reasoning traces, verifying the validity of these traces remains a critical challenge. Existing evaluation methods are either unscalable, relying on manual clinician review, or superficial, utilizing proxy metrics (e.g. QA) that fail to capture the semantic correctness of clinical logic. In this work, we introduce a reproducible framework for evaluating reasoning in ECG signals. We propose decomposing reasoning into two distinct, components: (i) Perception, the accurate identification of patterns within the raw signal, and (ii) Deduction, the logical application of domain knowledge to those patterns. To evaluate Perception, we employ an agentic framework that generates code to empirically verify the temporal structures described in the reasoning trace. To evaluate Deduction, we measure the alignment of the model's logic against a structured database of established clinical criteria in a retrieval-based approach. This dual-verification method enables the scalable assessment of "true" reasoning capabilities.

arXiv:2603.00349v1 Announce Type: new Abstract: Real-world scenarios increasingly require multiple embodied agents to collaborate in dynamic environments under embodied constraints, as many tasks exceed the capabilities of any single agent. Recent advances in large language models (LLMs) enable high-level cognitive coordination through reasoning, planning, and natural language communication. However, fine-grained analyses of how such collaboration emerges, unfolds, and contributes to task success in embodied multi-agent systems are difficult to conduct with existing benchmarks. In this paper, we introduce EmCoop, a benchmark framework for studying cooperation in LLM-based embodied multi-agent systems. Our framework separates a high-level cognitive layer from a low-level embodied interaction layer, allowing us to characterize agent cooperation through their interleaved dynamics over time. Given a cooperation-constrained embodied task, we propose generalizable, process-level metrics that diagnose collaboration quality and failure modes, beyond final task success. We instantiate our framework in two embodied environments that scale to arbitrary numbers of agents and support diverse communication topologies, and use these instantiations to demonstrate how EmCoop enables systematic analysis of cooperation dynamics across team sizes and task settings. The project web page can be found at: https://happyeureka.github.io/emcoop.

arXiv:2603.00350v1 Announce Type: new Abstract: The prevailing paradigm in artificial intelligence research equates progress with scale: larger models trained on broader datasets are presumed to yield superior capabilities. This assumption, while empirically productive for general-purpose applications, obscures a fundamental epistemological tension between breadth and depth of knowledge. We introduce the concept of \emph{Monotropic Artificial Intelligence} -- language models that deliberately sacrifice generality to achieve extraordinary precision within narrowly circumscribed domains. Drawing on the cognitive theory of monotropism developed to understand autistic cognition, we argue that intense specialization represents not a limitation but an alternative cognitive architecture with distinct advantages for safety-critical applications. We formalize the defining characteristics of monotropic models, contrast them with conventional polytropic architectures, and demonstrate their viability through Mini-Enedina, a 37.5-million-parameter model that achieves near-perfect performance on Timoshenko beam analysis while remaining deliberately incompetent outside its domain. Our framework challenges the implicit assumption that artificial general intelligence constitutes the sole legitimate aspiration of AI research, proposing instead a cognitive ecology in which specialized and generalist systems coexist complementarily.

arXiv:2603.00374v1 Announce Type: new Abstract: Offline learning of strategies takes data efficiency to its extreme by restricting algorithms to a fixed dataset of state-action trajectories. We consider the problem in a mixed-motive multiagent setting, where the goal is to solve a game under the offline learning constraint. We first frame this problem in terms of selecting among candidate equilibria. Since datasets may inform only a small fraction of game dynamics, it is generally infeasible in offline game-solving to even verify a proposed solution is a true equilibrium. Therefore, we consider the relative probability of low regret (i.e., closeness to equilibrium) across candidates based on the information available. Specifically, we extend Policy Space Response Oracles (PSRO), an online game-solving approach, by quantifying game dynamics uncertainty and modifying the RL objective to skew towards solutions more likely to have low regret in the true game. We further propose a novel meta-strategy solver, tailored for the offline setting, to guide strategy exploration in PSRO. Our incorporation of Conservatism principles from Offline reinforcement learning approaches for strategy Exploration gives our approach its name: COffeE-PSRO. Experiments demonstrate COffeE-PSRO's ability to extract lower-regret solutions than state-of-the-art offline approaches and reveal relationships between algorithmic components empirical game fidelity, and overall performance.

arXiv:2603.00376v2 Announce Type: new Abstract: NeuroHex is a hexagonal coordinate system designed to support highly efficient world models and reference frames for online adaptive AI systems. Inspired by the hexadirectional firing structure of grid cells in the human brain, NeuroHex adopts a cubic isometric hexagonal coordinate formulation that provides full 60{\deg} rotational symmetry and low-cost translation, rotation and distance computation. We develop a mathematical framework that incorporates ring indexing, quantized angular encoding, and a hierarchical library of foundational, simple, and complex geometric shape primitives. These constructs allow low-overhead point-in-shape tests and spatial matching operations that are expensive in Cartesian coordinate systems. To support realistic settings, the NeuroHex framework can process OpenStreetMap (OSM) data sets using an OSM-to-NeuroHex (OSM2Hex) conversion tool. The OSM2Hex spatial abstraction processing pipeline can achieve a reduction of 90-99% in geometric complexity while maintaining the relevant spatial structure map for navigation. Our initial results, based on actual city and neighborhood scale data sets, demonstrate that NeuroHex offers a highly efficient substrate for building dynamic world models to enable adaptive spatial reasoning in autonomous AI systems with continuous online learning capability.

arXiv:2603.00451v1 Announce Type: new Abstract: Accurate and unambiguous guidelines are critical for large language model (LLM) based graders, yet manually crafting these prompts is often sub-optimal as LLMs can misinterpret expert guidelines or lack necessary domain specificity. Consequently, the field has moved toward automated prompt optimization to refine grading guidelines without the burden of manual trial and error. However, existing frameworks typically aggregate independent and unstructured error samples into a single update step, resulting in "rule dilution" where conflicting constraints weaken the model's grading logic. To address these limitations, we introduce Confusion-Aware Rubric Optimization (CARO), a novel framework that enhances accuracy and computational efficiency by structurally separating error signals. CARO leverages the confusion matrix to decompose monolithic error signals into distinct modes, allowing for the diagnosis and repair of specific misclassification patterns individually. By synthesizing targeted "fixing patches" for dominant error modes and employing a diversity-aware selection mechanism, the framework prevents guidance conflict and eliminates the need for resource-heavy nested refinement loops. Empirical evaluations on teacher education and STEM datasets demonstrate that CARO significantly outperforms existing SOTA methods. These results suggest that replacing mixed-error aggregation with surgical, mode-specific repair yields robust improvements in automated assessment scalability and precision.

arXiv:2603.00460v1 Announce Type: new Abstract: Clinical decision-making requires synthesizing heterogeneous evidence, including patient histories, clinical guidelines, and trajectories of comparable cases. While large language models (LLMs) offer strong reasoning capabilities, they remain prone to hallucinations and struggle to integrate long, structured medical documents. We present MED-COPILOT, an interactive clinical decision-support system designed for clinicians and medical trainees, which combines guideline-grounded GraphRAG retrieval with hybrid semantic-keyword similar-patient retrieval to support transparent and evidence-aware clinical reasoning. The system builds a structured knowledge graph from WHO and NICE guidelines, applies community-level summarization for efficient retrieval, and maintains a 36,000-case similar-patient database derived from SOAP-normalized MIMIC-IV notes and Synthea-generated records. We evaluate our framework on clinical note completion and medical question answering, and demonstrate that it consistently outperforms parametric LLM baselines and standard RAG, improving both generation fidelity and clinical reasoning accuracy. The full system is available at https://huggingface.co/spaces/Cryo3978/Med_GraphRAG , enabling users to inspect retrieved evidence, visualize token-level similarity contributions, and conduct guided follow-up analysis. Our results demonstrate a practical and interpretable approach to integrating structured guideline knowledge with patient-level analogical evidence for clinical LLMs.

arXiv:2603.00465v1 Announce Type: new Abstract: Automated assessment of open-ended student responses is a critical capability for scaling personalized feedback in education. While large language models (LLMs) have shown promise in grading tasks via in-context learning (ICL), their reliability is heavily dependent on the selection of few-shot exemplars and the construction of high-quality rationales. Standard retrieval methods typically select examples based on semantic similarity, which often fails to capture subtle decision boundaries required for rubric adherence. Furthermore, manually crafting the expert rationales needed to guide these models can be a significant bottleneck. To address these limitations, we introduce GUIDE (Grading Using Iteratively Designed Exemplars), a framework that reframes exemplar selection and refinement in automated grading as a boundary-focused optimization problem. GUIDE operates on a continuous loop of selection and refinement, employing novel contrastive operators to identify "boundary pairs" that are semantically similar but possess different grades. We enhance exemplars by generating discriminative rationales that explicitly articulate why a response receives a specific score to the exclusion of adjacent grades. Extensive experiments across datasets in physics, chemistry, and pedagogical content knowledge demonstrate that GUIDE significantly outperforms standard retrieval baselines. By focusing the model's attention on the precise edges of rubric, our approach shows exceptionally robust gains on borderline cases and improved rubric adherence. GUIDE paves the way for trusted, scalable assessment systems that align closely with human pedagogical standards.

arXiv:2603.00469v1 Announce Type: new Abstract: Operators of Earth observation satellites need justifications for scheduling decisions: why a request was selected, rejected, or what changes would make it schedulable. Existing approaches construct post-hoc reasoning layers independent of the optimizer, risking non-causal attributions, incomplete constraint conjunctions, and solver-path dependence. We take a faithfulness-first approach: every explanation is a certificate derived from the optimization model itself: minimal infeasible subsets for rejections, tight constraints and contrastive trade-offs for selections, and inverse solves for what-if queries. On a scheduling instance with structurally distinct constraint interactions, certificates achieve perfect soundness with respect to the solver's constraint model (15/15 cited-constraint checks), counterfactual validity (7/7), and stability (Jaccard = 1.0 across 28 seed-pairs), while a post-hoc baseline produces non-causal attributions in 29% of cases and misses constraint conjunctions in every multi-cause rejection. A scalability analysis up to 200 orders and 30 satellites confirms practical extraction times for operational batches.

arXiv:2603.00472v1 Announce Type: new Abstract: Agentic AI systems exhibit numerous crosscutting concerns -- security, observability, cost management, fault tolerance -- that are poorly modularized in current implementations, contributing to the high failure rate of AI projects in reaching production. The goals-to-aspects methodology proposed at RE 2004 demonstrated that aspects can be systematically discovered from i* goal models by identifying non-functional soft-goals that crosscut functional goals. This paper revisits and extends that methodology to the agentic AI domain. We present a pattern language of 12 reusable patterns organized across four NFR categories (security, reliability, observability, cost management), each mapping an i* goal model to a concrete aspect implementation using an AOP framework for Rust. Four patterns address agent-specific crosscutting concerns absent from traditional AOP literature: tool-scope sandboxing, prompt injection detection, token budget management, and action audit trails. We extend the V-graph model to capture how agent tasks simultaneously contribute to functional goals and non-functional soft-goals. We validate the pattern language through a case study analyzing an open-source autonomous agent framework, demonstrating how goal-driven aspect discovery systematically identifies and modularizes crosscutting concerns. The pattern language offers a principled approach for engineering reliable agentic AI systems through early identification of crosscutting concerns.

arXiv:2603.00490v1 Announce Type: new Abstract: The rapid progress of Multimodal Large Language Models (MLLMs) marks a significant step toward artificial general intelligence, offering great potential for augmenting human capabilities. However, their ability to provide effective assistance in dynamic, real-world environments remains largely underexplored. Existing video benchmarks predominantly assess passive understanding through retrospective analysis or isolated perception tasks, failing to capture the interactive and adaptive nature of real-time user assistance. To bridge this gap, we introduce LifeEval, a multimodal benchmark designed to evaluate real-time, task-oriented human-AI collaboration in daily life from an egocentric perspective. LifeEval emphasizes three key aspects: task-oriented holistic evaluation, egocentric real-time perception from continuous first-person streams, and human-assistant collaborative interaction through natural dialogues. Constructed via a rigorous annotation pipeline, the benchmark comprises 4,075 high-quality question-answer pairs across 6 core capability dimensions. Extensive evaluations of 26 state-of-the-art MLLMs on LifeEval reveal substantial challenges in achieving timely, effective and adaptive interaction, highlighting essential directions for advancing human-centered interactive intelligence.

arXiv:2603.00495v1 Announce Type: new Abstract: We introduce AI Runtime Infrastructure, a distinct execution-time layer that operates above the model and below the application, actively observing, reasoning over, and intervening in agent behavior to optimize task success, latency, token efficiency, reliability, and safety while the agent is running. Unlike model-level optimizations or passive logging systems, runtime infrastructure treats execution itself as an optimization surface, enabling adaptive memory management, failure detection, recovery, and policy enforcement over long-horizon agent workflows.

arXiv:2603.00532v1 Announce Type: new Abstract: Autonomous agents are increasingly entrusted with complex, long-horizon tasks, ranging from mathematical reasoning to software generation. While agentic workflows facilitate these tasks by decomposing them into multi-step reasoning chains, reliability degrades significantly as the sequence lengthens. Specifically, minor interpretation errors in natural-language instructions tend to compound silently across steps. We term this failure mode accumulated semantic ambiguity. Existing approaches to mitigate this often lack runtime adaptivity, relying instead on static exploration budgets, reactive error recovery, or single-path execution that ignores uncertainty entirely. We formalize the multi-step reasoning process as a Noisy MDP and propose DenoiseFlow, a closed-loop framework that performs progressive denoising through three coordinated stages: (1)Sensing estimates per-step semantic uncertainty; (2)Regulating adaptively allocates computation by routing between fast single-path execution and parallel exploration based on estimated risk; and (3)Correcting performs targeted recovery via influence-based root-cause localization. Online self-calibration continuously aligns decision boundaries with verifier feedback, requiring no ground-truth labels. Experiments on six benchmarks spanning mathematical reasoning, code generation, and multi-hop QA show that DenoiseFlow achieves the highest accuracy on every benchmark (83.3% average, +1.3% over the strongest baseline) while reducing cost by 40--56% through adaptive branching. Detailed ablation studies further confirm framework-level's robustness and generality. Code is available at https://anonymous.4open.science/r/DenoiseFlow-21D3/.

arXiv:2603.00540v1 Announce Type: new Abstract: The evolution of Large Language Models (LLMs) from static instruction-followers to autonomous agents necessitates operating within complex, stateful environments to achieve precise state-transition objectives. However, this paradigm is bottlenecked by data scarcity, as existing tool-centric reverse-synthesis pipelines fail to capture the rigorous logic of real-world applications. We introduce \textbf{LOGIGEN}, a logic-driven framework that synthesizes verifiable training data based on three core pillars: \textbf{Hard-Compiled Policy Grounding}, \textbf{Logic-Driven Forward Synthesis}, and \textbf{Deterministic State Verification}. Specifically, a Triple-Agent Orchestration is employed: the \textbf{Architect} compiles natural-language policy into database constraints to enforce hard rules; the \textbf{Set Designer} initializes boundary-adjacent states to trigger critical policy conflicts; and the \textbf{Explorer} searches this environment to discover causal solution paths. This framework yields a dataset of 20,000 complex tasks across 8 domains, where validity is strictly guaranteed by checking exact state equivalence. Furthermore, we propose a verification-based training protocol where Supervised Fine-Tuning (SFT) on verifiable trajectories establishes compliance with hard-compiled policy, while Reinforcement Learning (RL) guided by dense state-rewards refines long-horizon goal achievement. On $\tau^2$-Bench, LOGIGEN-32B(RL) achieves a \textbf{79.5\% success rate}, substantially outperforming the base model (40.7\%). These results demonstrate that logic-driven synthesis combined with verification-based training effectively constructs the causally valid trajectories needed for next-generation agents.

arXiv:2603.00546v1 Announce Type: new Abstract: Using Multimodal Large Language Models (MLLMs) as judges to achieve precise and consistent evaluations has gradually become an emerging paradigm across various domains. Evaluating the capability and reliability of MLLM-as-a-judge systems is therefore essential for ensuring trustworthy assessment. Existing judge benchmarks categorize samples by task types but fail to capture the fundamental judgment capabilities required for reliable evaluation. In this work, we introduce M-JudgeBench, a ten-dimensional capability-oriented benchmark designed to comprehensively assess the judgment abilities of MLLMs. Our benchmark decomposes evaluation into pairwise Chain-of-Thought (CoT) comparison, length bias avoidance, and process error detection tasks, jointly covering ten fine-grained subtasks. This design enables diagnosis of model reliability across reasoning styles, response lengths, and cross-model variations. Systematic evaluation uncovers the systematic weaknesses in existing MLLM-as-a-judge systems. To address this issue, we further propose Judge-MCTS, a data construction framework generating pairwise reasoning trajectories with various correctness and length. Using Judge-MCTS, we construct an MCTS-augmented dataset and train M-Judger, a series of strong judge models. Extensive experiments demonstrate the superiority of M-Judger on existing judge benchmarks as well as M-JudgeBench. Overall, our work establishes a more principled foundation for evaluating MLLM-as-a-judge through M-JudgeBench and Judge-MCTS framework, paving the way for future research on judge model evaluation and capability-driven judge training.

arXiv:2603.00552v1 Announce Type: new Abstract: Evaluating persona-aligned empathy in LLM-based dialogue agents remains challenging. User states are latent, feedback is sparse and difficult to verify in situ, and seemingly supportive turns can still accumulate into trajectories that drift from persona-specific needs. We introduce EMPA, a process-oriented framework that evaluates persona-aligned support as sustained intervention rather than isolated replies. EMPA distills real interactions into controllable, psychologically grounded scenarios, couples them with an open-ended multi-agent sandbox that exposes strategic adaptation and failure modes, and scores trajectories in a latent psychological space by directional alignment, cumulative impact, and stability. The resulting signals and metrics support reproducible comparison and optimization of long-horizon empathic behavior, and they extend to other agent settings shaped by latent dynamics and weak, hard-to-verify feedback.

arXiv:2603.00575v1 Announce Type: new Abstract: Progress in software-engineering agents is increasingly constrained by the scarcity of executable, scalable, and realistic data for training and evaluation. This scarcity stems from three fundamental challenges in existing pipelines: environments are brittle and difficult to reproduce across languages; synthesizing realistic, system-level bugs at scale is computationally expensive; and existing data predominantly consists of short-horizon repairs, failing to capture long-horizon competencies like architectural consistency. We introduce \textbf{SWE-Hub}, an end-to-end system that operationalizes the data factory abstraction by unifying environment automation, scalable synthesis, and diverse task generation into a coherent production stack. At its foundation, the \textbf{Env Agent} establishes a shared execution substrate by automatically converting raw repository snapshots into reproducible, multi-language container environments with standardized interfaces. Built upon this substrate, \textbf{SWE-Scale} engine addresses the need for high-throughput generation, combining cross-language code analysis with cluster-scale validation to synthesize massive volumes of localized bug-fix instances. \textbf{Bug Agent} generates high-fidelity repair tasks by synthesizing system-level regressions involving cross-module dependencies, paired with user-like issue reports that describe observable symptoms rather than root causes. Finally, \textbf{SWE-Architect} expands the task scope from repair to creation by translating natural-language requirements into repository-scale build-a-repo tasks. By integrating these components, SWE-Hub establishes a unified production pipeline capable of continuously delivering executable tasks across the entire software engineering lifecycle.

arXiv:2603.00578v1 Announce Type: new Abstract: Long chain-of-thought~(CoT) has become a dominant paradigm for enhancing the reasoning capability of large reasoning models~(LRMs); however, the performance gains often come with a substantial increase in reasoning budget. Recent studies show that existing CoT paradigms tend to induce systematic overthinking, unnecessarily coupling reasoning capability with reasoning cost. Most prior approaches reduce token usage through post hoc techniques such as token compression, truncation, or length penalties, without explicitly addressing the core mechanisms of reasoning. We propose \textbf{Draft-Thinking}, which guides models to first learn a concise \textit{draft-style} reasoning structure that retains only the critical reasoning steps. Through a \textit{progressive curriculum learning}, the model stably internalizes this efficient reasoning pattern as its capability scales. Moreover, Draft-Thinking introduces adaptive prompting, which elevates reasoning depth to a flexible, model-selectable behavior. Extensive experiments demonstrate that Draft-Thinking substantially reduces reasoning budget while largely preserving reasoning performance; for example, on MATH500, it achieves an 82.6\% reduction in reasoning budget at the cost of only a 2.6\% performance drop.

arXiv:2603.00585v1 Announce Type: new Abstract: Recent advances in video generation have opened new avenues for macroscopic simulation of complex dynamic systems, but their application to microscopic phenomena remains largely unexplored. Microscale simulation holds great promise for biomedical applications such as drug discovery, organ-on-chip systems, and disease mechanism studies, while also showing potential in education and interactive visualization. In this work, we introduce MicroWorldBench, a multi-level rubric-based benchmark for microscale simulation tasks. MicroWorldBench enables systematic, rubric-based evaluation through 459 unique expert-annotated criteria spanning multiple microscale simulation task (e.g., organ-level processes, cellular dynamics, and subcellular molecular interactions) and evaluation dimensions (e.g., scientific fidelity, visual quality, instruction following). MicroWorldBench reveals that current SOTA video generation models fail in microscale simulation, showing violations of physical laws, temporal inconsistency, and misalignment with expert criteria. To address these limitations, we construct MicroSim-10K, a high-quality, expert-verified simulation dataset. Leveraging this dataset, we train MicroVerse, a video generation model tailored for microscale simulation. MicroVerse can accurately reproduce complex microscale mechanism. Our work first introduce the concept of Micro-World Simulation and present a proof of concept, paving the way for applications in biology, education, and scientific visualization. Our work demonstrates the potential of educational microscale simulations of biological mechanisms. Our data and code are publicly available at https://github.com/FreedomIntelligence/MicroVerse

arXiv:2603.00590v1 Announce Type: new Abstract: As artificial intelligence (AI) is increasingly deployed across domains, ensuring fairness has become a core challenge. However, the field faces a "Tower of Babel'' dilemma: fairness metrics abound, yet their underlying philosophical assumptions often conflict, hindering unified paradigms-particularly in unified Multimodal Large Language Models (UMLLMs), where biases propagate systemically across tasks. To address this, we introduce the IRIS Benchmark, to our knowledge the first benchmark designed to synchronously evaluate the fairness of both understanding and generation tasks in UMLLMs. Enabled by our demographic classifier, ARES, and four supporting large-scale datasets, the benchmark is designed to normalize and aggregate arbitrary metrics into a high-dimensional "fairness space'', integrating 60 granular metrics across three dimensions-Ideal Fairness, Real-world Fidelity, and Bias Inertia & Steerability (IRIS). Through this benchmark, our evaluation of leading UMLLMs uncovers systemic phenomena such as the "generation gap'', individual inconsistencies like "personality splits'', and the "counter-stereotype reward'', while offering diagnostics to guide the optimization of their fairness capabilities. With its novel and extensible framework, the IRIS benchmark is capable of integrating evolving fairness metrics, ultimately helping to resolve the "Tower of Babel'' impasse. Project Page: https://iris-benchmark-web.vercel.app/

arXiv:2603.00599v1 Announce Type: new Abstract: Hypergraphs are the natural description of higher-order interactions among objects, widely applied in social network analysis, cross-modal retrieval, etc. Hypergraph Neural Networks (HGNNs) have become the dominant solution for learning on hypergraphs. Traditional HGNNs are extended from message passing graph neural networks, following the homophily assumption, and thus struggle with the prevalent heterophilic hypergraphs that call for long-range dependence modeling. In this paper, we achieve heterophily-agnostic message passing through the lens of Riemannian geometry. The key insight lies in the connection between oversquashing and hypergraph bottleneck within the framework of Riemannian manifold heat flow. Building on this, we propose the novel idea of locally adapting the bottlenecks of different subhypergraphs. The core innovation of the proposed mechanism is the design of an adaptive local (heat) exchanger. Specifically, it captures the rich long-range dependencies via the Robin condition, and preserves the representation distinguishability via source terms, thereby enabling heterophily-agnostic message passing with theoretical guarantees. Based on this theoretical foundation, we present a novel Heat-Exchanger with Adaptive Locality for Hypergraph Neural Network (HealHGNN), designed as a node-hyperedge bidirectional systems with linear complexity in the number of nodes and hyperedges. Extensive experiments on both homophilic and heterophilic cases show that HealHGNN achieves the state-of-the-art performance.

arXiv:2603.00608v1 Announce Type: new Abstract: Student repetition in secondary education imposes significant resource burdens, particularly in resource-constrained contexts. Addressing this challenge, this study introduces a unified machine learning framework that simultaneously predicts pass/fail outcomes and continuous grades, a departure from prior research that treats classification and regression as separate tasks. Six models were evaluated: Logistic Regression, Decision Tree, and Random Forest for classification, and Linear Regression, Decision Tree Regressor, and Random Forest Regressor for regression, with hyperparameters optimized via exhaustive grid search. Using academic and demographic data from 4424 secondary school students, classification models achieved accuracies of up to 96%, while regression models attained a coefficient of determination of 0.70, surpassing baseline approaches. These results confirm the feasibility of early, data-driven identification of at-risk students and highlight the value of integrating dual-task prediction for more comprehensive insights. By enabling timely, personalized interventions, the framework offers a practical pathway to reducing grade repetition and optimizing resource allocation.

arXiv:2603.00623v1 Announce Type: new Abstract: Agentic systems augment large language models with external tools and iterative decision making, enabling complex tasks such as deep research, function calling, and coding. However, their long and intricate execution traces make failure diagnosis and root cause analysis extremely challenging. Manual inspection does not scale, while directly applying LLMs to raw traces is hindered by input length limits and unreliable reasoning. Focusing solely on final task outcomes further discards critical behavioral information required for accurate issue localization. To address these issues, we propose TraceSIR, a multi-agent framework for structured analysis and reporting of agentic execution traces. TraceSIR coordinates three specialized agents: (1) StructureAgent, which introduces a novel abstraction format, TraceFormat, to compress execution traces while preserving essential behavioral information; (2) InsightAgent, which performs fine-grained diagnosis including issue localization, root cause analysis, and optimization suggestions; (3) ReportAgent, which aggregates insights across task instances and generates comprehensive analysis reports. To evaluate TraceSIR, we construct TraceBench, covering three real-world agentic scenarios, and introduce ReportEval, an evaluation protocol for assessing the quality and usability of analysis reports aligned with industry needs. Experiments show that TraceSIR consistently produces coherent, informative, and actionable reports, significantly outperforming existing approaches across all evaluation dimensions. Our project and video are publicly available at https://github.com/SHU-XUN/TraceSIR.

arXiv:2603.00631v1 Announce Type: new Abstract: LiTS is a modular Python framework for LLM reasoning via tree search. It decomposes tree search into three reusable components (Policy, Transition, and RewardModel) that plug into algorithms like MCTS and BFS. A decorator-based registry enables domain experts to extend to new domains by registering components, and algorithmic researchers to implement custom search algorithms. We demonstrate composability on MATH500 (language reasoning), Crosswords (environment planning), and MapEval (tool use), showing that components and algorithms are orthogonal: components are reusable across algorithms within each task type, and algorithms work across all components and domains. We also report a mode-collapse finding: in infinite action spaces, LLM policy diversity (not reward quality) is the bottleneck for effective tree search. A demonstration video is available at https://youtu.be/nRGX43YrR3I. The package is released under the Apache 2.0 license at https://github.com/xinzhel/lits-llm, including installation instructions and runnable examples that enable users to reproduce the demonstrated workflows.

arXiv:2603.00656v1 Announce Type: new Abstract: Real-world user requests to LLM agents are often underspecified. Agents must interact to acquire missing information and make correct downstream decisions. However, current multi-turn GRPO-based methods often rely on trajectory-level reward computation, which leads to credit assignment problems and insufficient advantage signals within rollout groups. A feasible approach is to identify valuable interaction turns at a fine granularity to drive more targeted learning. To address this, we introduce InfoPO (Information-Driven Policy Optimization), which frames multi-turn interaction as a process of active uncertainty reduction and computes an information-gain reward that credits turns whose feedback measurably changes the agent's subsequent action distribution compared to a masked-feedback counterfactual. It then combines this signal with task outcomes via an adaptive variance-gated fusion to identify information importance while maintaining task-oriented goal direction. Across diverse tasks, including intent clarification, collaborative coding, and tool-augmented decision making, InfoPO consistently outperforms prompting and multi-turn RL baselines. It also demonstrates robustness under user simulator shifts and generalizes effectively to environment-interactive tasks. Overall, InfoPO provides a principled and scalable mechanism for optimizing complex agent-user collaboration. Code is available at https://github.com/kfq20/InfoPO.

arXiv:2603.00676v1 Announce Type: new Abstract: Existing mobile device control agents often perform poorly when solving complex tasks requiring long-horizon planning and precise operations, typically due to a lack of relevant task experience or unfamiliarity with skill execution. We propose K2-Agent, a hierarchical framework that models human-like cognition by separating and co-evolving declarative (knowing what) and procedural (knowing how) knowledge for planning and execution. K2-Agent's high level reasoner is bootstrapped from a single demonstration per task and runs a Summarize-Reflect-Locate-Revise (SRLR) loop to distill and iteratively refine task-level declarative knowledge through self-evolution. The low-level executor is trained with our curriculum-guided Group Relative Policy Optimization (C-GRPO), which (i) constructs a balanced sample pool using decoupled reward signals and (ii) employs dynamic demonstration injection to guide the model in autonomously generating successful trajectories for training. On the challenging AndroidWorld benchmark, K2-Agent achieves a 76.1% success rate using only raw screenshots and open-source backbones. Furthermore, K2-Agent shows powerful dual generalization: its high-level declarative knowledge transfers across diverse base models, while its low-level procedural skills achieve competitive performance on unseen tasks in ScreenSpot-v2 and Android-in-the-Wild (AitW).

arXiv:2603.00680v1 Announce Type: new Abstract: Long-horizon agents face the challenge of growing context size during interaction with environment, which degrades the performance and stability. Existing methods typically introduce the external memory module and look up the relevant information from the stored memory, which prevents the model itself from proactively managing its memory content and aligning with the agent's overarching task objectives. To address these limitations, we propose the self-memory policy optimization algorithm (MemPO), which enables the agent (policy model) to autonomously summarize and manage their memory during interaction with environment. By improving the credit assignment mechanism based on memory effectiveness, the policy model can selectively retain crucial information, significantly reducing token consumption while preserving task performance. Extensive experiments and analyses confirm that MemPO achieves absolute F1 score gains of 25.98% over the base model and 7.1% over the previous SOTA baseline, while reducing token usage by 67.58% and 73.12%.

arXiv:2603.00691v1 Announce Type: new Abstract: The world is undergoing a major demographic shift as older adults become a rapidly growing share of the population, creating new challenges for driving safety. In car-dependent regions such as the United States, driving remains essential for independence, access to services, and social participation. At the same time, aging can introduce gradual changes in vision, attention, reaction time, and driving control that quietly reduce safety. Today's assessment methods rely largely on infrequent clinic visits or simple screening tools, offering only a brief snapshot and failing to reflect how an older adult actually drives on the road. Our work starts from the observation that everyday driving provides a continuous record of functional ability and captures how a driver responds to traffic, navigates complex roads, and manages routine behavior. Leveraging this insight, we propose AURA, an Artificial Intelligence of Things (AIoT) framework for continuous, real-world assessment of driving safety among older adults. AURA integrates richer in-vehicle sensing, multi-scale behavioral modeling, and context-aware analysis to extract detailed indicators of driving performance from routine trips. It organizes fine-grained actions into longer behavioral trajectories and separates age-related performance changes from situational factors such as traffic, road design, or weather. By integrating sensing, modeling, and interpretation within a privacy-preserving edge architecture, AURA provides a foundation for proactive, individualized support that helps older adults drive safely. This paper outlines the design principles, challenges, and research opportunities needed to build reliable, real-world monitoring systems that promote safer aging behind the wheel.

arXiv:2603.00730v1 Announce Type: new Abstract: Deep reinforcement learning (RL) has been applied extensively to solve complex decision-making problems. In many real-world scenarios, tasks often have several conflicting objectives and may require multiple agents to cooperate, which are the multi-objective multi-agent decision-making problems. However, only few works have been conducted on this intersection. Existing approaches are limited to separate fields and can only handle multi-agent decision-making with a single objective, or multi-objective decision-making with a single agent. In this paper, we propose MO-MIX to solve the multi-objective multi-agent reinforcement learning (MOMARL) problem. Our approach is based on the centralized training with decentralized execution (CTDE) framework. A weight vector representing preference over the objectives is fed into the decentralized agent network as a condition for local action-value function estimation, while a mixing network with parallel architecture is used to estimate the joint action-value function. In addition, an exploration guide approach is applied to improve the uniformity of the final non-dominated solutions. Experiments demonstrate that the proposed method can effectively solve the multi-objective multi-agent cooperative decision-making problem and generate an approximation of the Pareto set. Our approach not only significantly outperforms the baseline method in all four kinds of evaluation metrics, but also requires less computational cost.

arXiv:2603.00801v1 Announce Type: new Abstract: Language agents increasingly act as web-enabled systems that search, browse, and synthesize information from diverse sources. However, these sources can include unreliable or adversarial content, and the robustness of agents to adversarial ranking - where misleading information appears prominently in search results - remains poorly understood. Existing benchmarks evaluate functional navigation or static factuality but cannot causally isolate this vulnerability, and current mitigation strategies for retrieval-augmented generation remain largely untested under such conditions. We introduce Synthetic Web Benchmark, a procedurally generated environment comprising thousands of hyperlinked articles with ground-truth labels for credibility and factuality, process-level interaction traces, and contamination filtering to eliminate training-data leakage. By injecting a single high-plausibility misinformation article into a controllable search rank, we measure the causal effect of adversarial exposure in six frontier models. The results reveal catastrophic failures: accuracy collapses despite unlimited access to truthful sources, with minimal search escalation and severe miscalibration. These findings expose fundamental limitations in how current frontier models handle conflicting information, with immediate implications for deployment in high-stakes domains. Our benchmark enables systematic analysis of these failure modes and provides a controlled testbed for evaluating mitigation strategies under adversarial ranking - a gap in current research. This work establishes a reproducible baseline for developing search-robust and epistemically humble agents capable of resisting manipulation in high-stakes domains.

arXiv:2603.00808v1 Announce Type: new Abstract: A major challenge for world models in multi-agent systems is to understand interdependent agent dynamics, predict interactive multi-agent trajectories, and plan over long horizons with collective awareness, without centralized supervision or explicit communication. In this paper, MetaMind, a general and cognitive world model for multi-agent systems that leverages a novel meta-theory of mind (Meta-ToM) framework, is proposed. Through MetaMind, each agent learns not only to predict and plan over its own beliefs, but also to inversely reason goals and beliefs from its own behavior trajectories. This self-reflective, bidirectional inference loop enables each agent to learn a metacognitive ability in a self-supervised manner. Then, MetaMind is shown to generalize the metacognitive ability from first-person to third-person through analogical reasoning. Thus, in multi-agent systems, each agent with MetaMind can actively reason about goals and beliefs of other agents from limited, observable behavior trajectories in a zero-shot manner, and then adapt to emergent collective intention without an explicit communication mechanism. Extended simulation results on diverse multi-agent tasks demonstrate that MetaMind can achieve superior task performance and outperform baselines in few-shot multi-agent generalization.

arXiv:2603.00873v1 Announce Type: new Abstract: With the increasing demand for step-wise, cross-modal, and knowledge-grounded reasoning, multimodal large language models (MLLMs) are evolving beyond the traditional fixed retrieve-then-generate paradigm toward more sophisticated agentic multimodal retrieval-augmented generation (MM-RAG). Existing benchmarks, however, mainly focus on simplified QA with short retrieval chains, leaving adaptive planning and multimodal reasoning underexplored. We present MC-Search, the first benchmark for agentic MM-RAG with long, step-wise annotated reasoning chains spanning five representative reasoning structures. Each example specifies sub-questions, retrieval modalities, supporting facts, and intermediate answers, with fidelity ensured by HAVE (Hop-wise Attribution and Verification of Evidence), resulting in 3,333 high-quality examples averaging 3.7 hops. Beyond answer accuracy, MC-Search introduces new process-level metrics for reasoning quality, stepwise retrieval and planning accuracy. By developing a unified agentic MM-RAG pipeline, we benchmark six leading MLLMs and reveal systematic issues such as over- and under-retrieval and modality-misaligned planning. Finally, we introduce Search-Align, a process-supervised fine-tuning framework leveraging verified reasoning chains, showing that our data not only enables faithful evaluation but also improves planning and retrieval fidelity in open-source MLLMs.

arXiv:2603.00876v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated significant reasoning capabilities in scientific discovery but struggle to bridge the gap to physical execution in wet-labs. In these irreversible environments, probabilistic hallucinations are not merely incorrect, but also cause equipment damage or experimental failure. To address this, we propose \textbf{BioProAgent}, a neuro-symbolic framework that anchors probabilistic planning in a deterministic Finite State Machine (FSM). We introduce a State-Augmented Planning mechanism that enforces a rigorous \textit{Design-Verify-Rectify} workflow, ensuring hardware compliance before execution. Furthermore, we address the context bottleneck inherent in complex device schemas by \textit{Semantic Symbol Grounding}, reducing token consumption by $\sim$6$\times$ through symbolic abstraction. In the extended BioProBench benchmark, BioProAgent achieves 95.6\% physical compliance (compared to 21.0\% for ReAct), demonstrating that neuro-symbolic constraints are essential for reliable autonomy in irreversible physical environments. \footnote{Code at https://github.com/YuyangSunshine/bioproagent and project at https://yuyangsunshine.github.io/BioPro-Project/}

arXiv:2603.00977v1 Announce Type: new Abstract: Large language model (LLM) agents have recently demonstrated strong capabilities in interactive decision-making, yet they remain fundamentally limited in long-horizon tasks that require structured planning and reliable execution. Existing approaches predominantly rely on flat autoregressive policies, where high-level reasoning and low-level actions are generated within a single token sequence, leading to inefficient exploration and severe error propagation over extended trajectories. In this work, we propose HiMAC, a hierarchical agentic RL framework that explicitly decomposes long-horizon decision-making into macro-level planning and micro-level execution. HiMAC models reasoning as a structured blueprint generation process followed by goal-conditioned action execution, enabling robust long-horizon planning within LLM-based agents. To train this hierarchy efficiently, we introduce a critic-free hierarchical policy optimization paradigm that extends group-based reinforcement learning to bi-level structures through hierarchical relative advantage estimation. Furthermore, we propose an iterative co-evolution training strategy that alternates between planner exploration and executor adaptation, mitigating the non-stationarity inherent in hierarchical learning. Extensive experiments on ALFWorld, WebShop, and Sokoban demonstrate that HiMAC consistently outperforms strong prompting and reinforcement learning baselines, achieving state-of-the-art performance and substantially improved sample efficiency across both text-based and visually grounded environments. Our results show that introducing structured hierarchy, rather than increasing model scale alone, is a key factor for enabling robust long-horizon agentic intelligence.

arXiv:2603.00991v1 Announce Type: new Abstract: AI agents that interact with the real world through tool calls pose fundamental safety challenges: agents might leak private information, cause unintended side effects, or be manipulated through prompt injection. To address these challenges, we propose to put the agent in a programming-language-based "safety harness": instead of calling tools directly, agents express their intentions as code in a capability-safe language: Scala 3 with capture checking. Capabilities are program variables that regulate access to effects and resources of interest. Scala's type system tracks capabilities statically, providing fine-grained control over what an agent can do. In particular, it enables local purity, the ability to enforce that sub-computations are side-effect-free, preventing information leakage when agents process classified data. We demonstrate that extensible agent safety harnesses can be built by leveraging a strong type system with tracked capabilities. Our experiments show that agents can generate capability-safe code with no significant loss in task performance, while the type system reliably prevents unsafe behaviors such as information leakage and malicious side effects.

arXiv:2603.00993v1 Announce Type: new Abstract: Large Language Models (LLMs) have revolutionized AI-generated content evaluation, with the LLM-as-a-Judge paradigm becoming increasingly popular. However, current single-LLM evaluation approaches face significant challenges, including inconsistent judgments and inherent biases from pre-training data. To address these limitations, we propose CollabEval, a novel multi-agent evaluation framework that implements a three-phase Collaborative Evaluation process: initial evaluation, multi-round discussion, and final judgment. Unlike existing approaches that rely on competitive debate or single-model evaluation, CollabEval emphasizes collaboration among multiple agents with strategic consensus checking for efficiency. Our extensive experiments demonstrate that CollabEval consistently outperforms single-LLM approaches across multiple dimensions while maintaining robust performance even when individual models struggle. The framework provides comprehensive support for various evaluation criteria while ensuring efficiency through its collaborative design.

arXiv:2603.01055v1 Announce Type: new Abstract: We present MMCOMET, the first multimodal commonsense knowledge graph (MMKG) that integrates physical, social, and eventive knowledge. MMCOMET extends the ATOMIC2020 knowledge graph to include a visual dimension, through an efficient image retrieval process, resulting in over 900K multimodal triples. This new resource addresses a major limitation of existing MMKGs in supporting complex reasoning tasks like image captioning and storytelling. Through a standard visual storytelling experiment, we show that our holistic approach enables the generation of richer, coherent, and contextually grounded stories than those produced using text-only knowledge. This resource establishes a new foundation for multimodal commonsense reasoning and narrative generation.

arXiv:2603.01092v1 Announce Type: new Abstract: Large language models are adept at synthesizing and recombining familiar material, yet they often fail at a specific kind of creativity that matters most in research: producing ideas that are both coherent and non-obvious to the current community. We formalize this gap through cognitive availability, the likelihood that a research direction would be naturally proposed by a typical researcher given what they have worked on. We introduce a pipeline that (i) decomposes papers into granular conceptual units, (ii) clusters recurring units into a shared vocabulary of idea atoms, and (iii) learns two complementary models: a coherence model that scores whether a set of atoms constitutes a viable direction, and an availability model that scores how likely that direction is to be generated by researchers drawn from the community. We then sample "alien" directions that score high on coherence but low on availability. On a corpus of $\sim$7,500 recent LLM papers from NeurIPS, ICLR and ICML, we validate that (a) conceptual units preserve paper content under reconstruction, (b) idea atoms generalize across papers rather than memorizing paper-specific phrasing, and (c) the Alien sampler produces research directions that are more diverse than LLM baselines while maintaining coherence.

arXiv:2603.01106v1 Announce Type: new Abstract: Reinforcement learning (RL) with group relative policy optimization (GRPO) has become a widely adopted approach for enhancing the reasoning capabilities of multimodal large language models (MLLMs). While GRPO enables long-chain reasoning without a critic, it often suffers from sparse rewards on difficult problems and advantage vanishing when group-level rewards are too consistent for overly easy or hard problems. Existing solutions (sample expansion, selective utilization, and indirect reward design) often fail to maintain enough variance in within-group reward distributions to yield clear optimization signals. To address this, we propose DIVA-GRPO, a difficulty-adaptive variant advantage method that adjusts variant difficulty distributions from a global perspective. DIVA-GRPO dynamically assesses problem difficulty, samples variants with appropriate difficulty levels, and calculates advantages across local and global groups using difficulty-weighted and normalized scaling. This alleviates reward sparsity and advantage vanishing while improving training stability. Extensive experiments on six reasoning benchmarks demonstrate that DIVA-GRPO outperforms existing approaches in training efficiency and reasoning performance. Code: https://github.com/Siaaaaaa1/DIVA-GRPO

arXiv:2603.01121v1 Announce Type: new Abstract: While deep learning-based weather forecasting paradigms have made significant strides, addressing extreme weather diagnostics remains a formidable challenge. This gap exists primarily because the diagnostic process demands sophisticated multi-step logical reasoning, dynamic tool invocation, and expert-level prior judgment. Although agents possess inherent advantages in task decomposition and autonomous execution, current architectures are still hampered by critical bottlenecks: inadequate expert knowledge integration, a lack of professional-grade iterative reasoning loops, and the absence of fine-grained validation and evaluation systems for complex workflows under extreme conditions. To this end, we propose HVR-Met, a multi-agent meteorological diagnostic system characterized by the deep integration of expert knowledge. Its central innovation is the ``Hypothesis-Verification-Replanning'' closed-loop mechanism, which facilitates sophisticated iterative reasoning for anomalous meteorological signals during extreme weather events. To bridge gaps within existing evaluation frameworks, we further introduce a novel benchmark focused on atomic-level subtasks. Experimental evidence demonstrates that the system excels in complex diagnostic scenarios.

arXiv:2603.01135v1 Announce Type: new Abstract: Large Language Models have achieved remarkable success in language understanding and reasoning, and their multimodal extensions enable comprehension of images, video, and audio. Inspired by this, foundation models for brain functional connectivity networks derived from resting-state fMRI have shown promise in clinical tasks. However, existing methods do not align FCNs with the text modality, limiting the ability of LLMs to directly understand FCNs. To address this, we propose FCN-LLM, a framework that enables LLMs to understand FCNs through graph-level, multi-task instruction tuning. Our approach employs a multi-scale FCN encoder capturing brain-region, functional subnetwork, and whole-brain features, projecting them into the semantic space of LLM. We design multi-paradigm instruction tasks covering 19 subject-specific attributes across demographics, phenotypes, and psychiatric conditions. A multi-stage learning strategy first aligns FCN embeddings with the LLM and then jointly fine-tunes the entire model to capture high-level semantic information. Experiments on a large-scale, multi-site FCN database show that FCN-LLM achieves strong zero-shot generalization on unseen datasets, outperforming conventional supervised and foundation models. This work introduces a new paradigm for integrating brain functional networks with LLMs, offering a flexible and interpretable framework for neuroscience.

arXiv:2603.01145v1 Announce Type: new Abstract: In practical LLM applications, users repeatedly express stable preferences and requirements, such as reducing hallucinations, following institutional writing conventions, or avoiding overly technical wording, yet such interaction experience is seldom consolidated into reusable knowledge. Consequently, LLM agents often fail to accumulate personalized capabilities across sessions. We present AutoSkill, an experience-driven lifelong learning framework that enables LLM agents to automatically derive, maintain, and reuse skills from dialogue and interaction traces. AutoSkill abstracts skills from user experience, supports their continual self-evolution, and dynamically injects relevant skills into future requests without retraining the underlying model. Designed as a model-agnostic plugin layer, it is compatible with existing LLMs and introduces a standardized skill representation for sharing and transfer across agents, users, and tasks. In this way, AutoSkill turns ephemeral interaction experience into explicit, reusable, and composable capabilities. This paper describes the motivation, architecture, skill lifecycle, and implementation of AutoSkill, and positions it with respect to prior work on memory, retrieval, personalization, and agentic systems. AutoSkill highlights a practical and scalable path toward lifelong personalized agents and personal digital surrogates.

arXiv:2603.01152v1 Announce Type: new Abstract: Deep-research agents are capable of executing multi-step web exploration, targeted retrieval, and sophisticated question answering. Despite their powerful capabilities, deep-research agents face two critical bottlenecks: (1) the lack of large-scale, challenging datasets with real-world difficulty, and (2) the absence of accessible, open-source frameworks for data synthesis and agent training. To bridge these gaps, we first construct DeepResearch-9K, a large-scale challenging dataset specifically designed for deep-research scenarios built from open-source multi-hop question-answering (QA) datasets via a low-cost autonomous pipeline. Notably, it consists of (1) 9000 questions spanning three difficulty levels from L1 to L3 (2) high-quality search trajectories with reasoning chains from Tongyi-DeepResearch-30B-A3B, a state-of-the-art deep-research agent, and (3) verifiable answers. Furthermore, we develop an open-source training framework DeepResearch-R1 that supports (1) multi-turn web interactions, (2) different reinforcement learning (RL) approaches, and (3) different reward models such as rule-based outcome reward and LLM-as-judge feedback. Finally, empirical results demonstrate that agents trained on DeepResearch-9K under our DeepResearch-R1 achieve state-of-the-art results on challenging deep-research benchmarks. We release the DeepResearch-9K dataset on https://huggingface.co/datasets/artillerywu/DeepResearch-9K and the code of DeepResearch-R1 on https://github.com/Applied-Machine-Learning-Lab/DeepResearch-R1.

arXiv:2603.01160v1 Announce Type: new Abstract: Conversational AI (ConvAI) agents increasingly maintain structured memory to support long-term, task-oriented interactions. In-context memory approaches append the growing history to the model input, which scales poorly under context-window limits. RAG-based methods retrieve request-relevant information, but most assume flat memory collections and ignore structure. We propose Semantic XPath, a tree-structured memory module to access and update structured conversational memory. Semantic XPath improves performance over flat-RAG baselines by 176.7% while using only 9.1% of the tokens required by in-context memory. We also introduce SemanticXPath Chat, an end-to-end ConvAI demo system that visualizes the structured memory and query execution details. Overall, this paper demonstrates a candidate for the next generation of long-term, task-oriented ConvAI systems built on structured memory.

arXiv:2603.01201v1 Announce Type: new Abstract: In this paper, we study incremental LTLf synthesis -- a form of reactive synthesis where the goals are given incrementally while in execution. In other words, the protagonist agent is already executing a strategy for a certain goal when it receives a new goal: at this point, the agent has to abandon the current strategy and synthesize a new strategy still fulfilling the original goal, which was given at the beginning, as well as the new goal, starting from the current instant. In this paper, we formally define the problem of incremental synthesis and study its solution. We propose a solution technique that efficiently performs incremental synthesis for multiple LTLf goals by leveraging auxiliary data structures constructed during automata-based synthesis. We also consider an alternative solution technique based on LTLf formula progression. We show that, in spite of the fact that formula progression can generate formulas that are exponentially larger than the original ones, their minimal automata remain bounded in size by that of the original formula. On the other hand, we show experimentally that, if implemented naively, i.e., by actually computing the automaton of the progressed LTLf formulas from scratch every time a new goal arrives, the solution based on formula progression is not competitive.

arXiv:2603.01203v1 Announce Type: new Abstract: AI agents are increasingly developed and evaluated on benchmarks relevant to human work, yet it remains unclear how representative these benchmarking efforts are of the labor market as a whole. In this work, we systematically study the relationship between agent development efforts and the distribution of real-world human work by mapping benchmark instances to work domains and skills. We first analyze 43 benchmarks and 72,342 tasks, measuring their alignment with human employment and capital allocation across all 1,016 real-world occupations in the U.S. labor market. We reveal substantial mismatches between agent development that tends to be programming-centric, and the categories in which human labor and economic value are concentrated. Within work areas that agents currently target, we further characterize current agent utility by measuring their autonomy levels, providing practical guidance for agent interaction strategies across work scenarios. Building on these findings, we propose three measurable principles for designing benchmarks that better capture socially important and technically challenging forms of work: coverage, realism, and granular evaluation.

arXiv:2603.01209v1 Announce Type: new Abstract: Tool-augmented LLMs are increasingly deployed as agents that interleave natural-language reasoning with executable Python actions, as in CodeAct-style frameworks. In deployment, these agents rely on runtime state that persists across steps. By contrast, common training pipelines treat agent traces as token sequences, with execution semantics left implicit. This raises a data-centric question: Is state persistence merely an inference-time scaffold, or can models learn to exploit it when training data exposes the corresponding execution semantics? We isolate state persistence as a training-time variable. We introduce Opaque Knapsack, a procedurally generated family of partially observable optimization tasks designed to prevent one-shot solutions. Item attributes and constraints are hidden behind budgeted tool calls, forcing multi-turn control flow and iterative state revision. Holding task instances, prompts, tools, model, and supervision fixed, we generate paired trajectories differing only in whether interpreter state persists across steps or resets after each action. We then fine-tune identical base models (Qwen3-8B) on each trace variant and evaluate all four train-runtime combinations. Our 2x2 cross-evaluation shows that execution semantics primarily affect how agents reach solutions, not whether they do: solution quality is statistically indistinguishable across conditions, but token cost and stability differ substantially. A persistent-trained model in a stateless runtime triggers missing-variable errors in roughly 80% of episodes; a stateless-trained model in a persistent runtime redundantly re-derives retained state, using roughly 3.5x more tokens. Interpreter persistence should be treated as a first-class semantic of agent traces. Aligning fine-tuning data with deployment runtimes improves efficiency and reduces brittle train-runtime mismatches.

arXiv:2603.01211v1 Announce Type: new Abstract: As generative AI technologies are increasingly being launched across the globe, assessing their competence to operate in different cultural contexts is exigently becoming a priority. While recent years have seen numerous and much-needed efforts on cultural benchmarking, these efforts have largely focused on specific aspects of culture and evaluation. While these efforts contribute to our understanding of cultural competence, a unified and systematic evaluation approach is needed for us as a field to comprehensively assess diverse cultural dimensions at scale. Drawing on measurement theory, we present a principled framework to aggregate multifaceted indicators of cultural capabilities into a unified assessment of cultural intelligence. We start by developing a working definition of culture that includes identifying core domains of culture. We then introduce a broad-purpose, systematic, and extensible framework for assessing cultural intelligence of AI systems. Drawing on theoretical framing from psychometric measurement validity theory, we decouple the background concept (i.e., cultural intelligence) from its operationalization via measurement. We conceptualize cultural intelligence as a suite of core capabilities spanning diverse domains, which we then operationalize through a set of indicators designed for reliable measurement. Finally, we identify the considerations, challenges, and research pathways to meaningfully measure these indicators, specifically focusing on data collection, probing strategies, and evaluation metrics.

arXiv:2603.01227v1 Announce Type: new Abstract: We propose the Lattice Representation Hypothesis of large language models: a symbolic backbone that grounds conceptual hierarchies and logical operations in embedding geometry. Our framework unifies the Linear Representation Hypothesis with Formal Concept Analysis (FCA), showing that linear attribute directions with separating thresholds induce a concept lattice via half-space intersections. This geometry enables symbolic reasoning through geometric meet (intersection) and join (union) operations, and admits a canonical form when attribute directions are linearly independent. Experiments on WordNet sub-hierarchies provide empirical evidence that LLM embeddings encode concept lattices and their logical structure, revealing a principled bridge between continuous geometry and symbolic abstraction.

arXiv:2603.01235v1 Announce Type: new Abstract: This technical report provides an extended validation of the Explainability Solution Space (ESS) through cross-domain evaluation. While initial validation focused on employee attrition prediction, this study introduces a heterogeneous intelligent urban resource allocation system to demonstrate the generality and domain-independence of the ESS framework. The second case study integrates tabular, temporal, and geospatial data under multi-stakeholder governance conditions. Explicit quantitative positioning of representative XAI families is provided for both contexts. Results confirm that ESS rankings are not domain-specific but adapt systematically to governance roles, risk profiles, and stakeholder configurations. The findings reinforce ESS as a generalizable operational decision-support instrument for explainable AI strategy design across socio-technical systems.

arXiv:2603.01283v1 Announce Type: new Abstract: Real-world reinforcement learning (RL) agents operate in closed-loop systems where actions shape future observations, making reliable deployment under distribution shifts a persistent challenge. Existing monitoring relies on reward or task metrics, capturing outcomes but missing early coupling failures. We introduce bipredictability (P) as the ratio of shared information in the observation, action, outcome loop to the total available information, a principled, real time measure of interaction effectiveness with provable bounds, comparable across tasks. An auxiliary monitor, the Information Digital Twin (IDT), computes P and its diagnostic components from the interaction stream. We evaluate SAC and PPO agents on MuJoCo HalfCheetah under eight agent, and environment-side perturbations across 168 trials. Under nominal operation, agents exhibit P = 0.33 plus minus 0.02, below the classical bound of 0.5, revealing an informational cost of action selection. The IDT detects 89.3% of perturbations versus 44.0% for reward based monitoring, with 4.4x lower median latency. Bipredictability enables early detection of interaction degradation before performance drops and provides a prerequisite signal for closed loop self regulation in deployed RL systems.

arXiv:2603.01286v1 Announce Type: new Abstract: Model Predictive Control (MPC) is a vital technique for autonomous systems, like Unmanned Aerial Vehicles (UAVs), enabling optimized motion planning. However, traditional MPC struggles to adapt to real-time changes such as dynamic obstacles and shifting system dynamics, lacking inherent mechanisms for self-monitoring and adaptive optimization. Here, we introduce Entanglement Learning (EL), an information-theoretic framework that enhances MPC adaptability through an Information Digital Twin (IDT). The IDT monitors and quantifies, in bits, the information flow between MPC inputs, control actions, and UAV behavior. By introducing new information-theoretic metrics we call entanglement metrics, it tracks variations in these dependencies. These metrics measure the mutual information between the optimizer's input, its control actions, and the resulting UAV dynamics, enabling a deeper understanding of their interrelationships. This allows the IDT to detect performance deviations and generate real-time adaptive signals to recalibrate MPC parameters, preserving stability. Unlike traditional MPC, which relies on error-based feedback, this dual-feedback approach leverages information flow for proactive adaptation to evolving conditions. Scalable and leveraging existing infrastructure, this framework improves MPC reliability and robustness across diverse scenarios, extending beyond UAV control to any MPC implementation requiring adaptive performance.

arXiv:2603.01290v1 Announce Type: new Abstract: The 2026 Formula 1 technical regulations introduce a fundamental change to energy strategy: under a 50/50 internal combustion engine / battery power split with unlimited regeneration and a driver-controlled Override Mode (abbreviated MOM throughout), the optimal energy deployment policy depends not only on a driver's own state but on the hidden state of rival cars. This creates a Partially Observable Stochastic Game that cannot be solved by single-agent optimisation methods. We present a tractable two-layer inference and decision framework. The first layer is a 30-state Hidden Markov Model (HMM) that infers a probability distribution over each rival's ERS charge level, Override Mode status, and tyre degradation state from five publicly observable telemetry signals. The second layer is a Deep Q-Network (DQN) policy that takes the HMM belief state as input and selects between energy deployment strategies. We formally characterise the counter-harvest trap -- a deceptive strategy in which a car deliberately suppresses observable deployment signals to induce a rival into a failed attack -- and show that detecting it requires belief-state inference rather than reactive threshold rules. On synthetic races generated from the model's own assumptions, the HMM achieves 92.3% ERS inference accuracy (random baseline: 33.3%) and detects counter-harvest trap conditions with 95.7% recall. Pre-registration -- empirical validation begins Australian Grand Prix, 8 March 2026.

arXiv:2603.01357v1 Announce Type: new Abstract: Next-generation AI must manage vast personal data, diverse tools, and multi-step reasoning, yet most benchmarks remain context-free and single-turn. We present ASTRA-bench (Assistant Skills in Tool-use, Reasoning \& Action-planning), a benchmark that uniquely unifies time-evolving personal context with an interactive toolbox and complex user intents. Our event-driven pipeline generates 2,413 scenarios across four protagonists, grounded in longitudinal life events and annotated by referential, functional, and informational complexity. Evaluation of state-of-the-art models (e.g., Claude-4.5-Opus, DeepSeek-V3.2) reveals significant performance degradation under high-complexity conditions, with argument generation emerging as the primary bottleneck. These findings expose critical limitations in current agents' ability to ground reasoning within messy personal context and orchestrate reliable multi-step plans. We release ASTRA-bench with a full execution environment and evaluation scripts to provide a diagnostic testbed for developing truly context-aware AI assistants.

arXiv:2603.01375v1 Announce Type: new Abstract: Test-time policy adaptation for multi-turn interactions (T2PAM) is essential for aligning Large Language Models (LLMs) with dynamic user needs during inference time. However, existing paradigms commonly treat test-time adaptation as a single-axis problem, either purely refining instructions (Prompt Engineering) or only adjusting weights (Test-Time Training), ignoring that interaction failures stem from a coupled mix of ambiguity and incapacity. We argue that these two optimization paths are not merely additive but synergistic: semantic clarity acts as a pre-conditioner for effective parameter updates. To this end, we propose ROSA2, a framework that reformulates interaction as a joint optimization problem over the heterogeneous space of Words and Weights. By mathematically decomposing the error signal, ROSA2 utilizes textual gradients to rectify intent ambiguity and parameter updates to bridge capability gaps. Theoretically, we prove that this co-adaptation strictly reduces the required parameter shift for convergence. Empirically, ROSA2 outperforms state-of-the-art baselines by 30% on MATH while reducing interaction turns by 40%, demonstrating that refining the context unlocks the true potential of parameter updates.

arXiv:2603.01396v1 Announce Type: new Abstract: Single-cell perturbation studies face dual heterogeneity bottlenecks: (i) semantic heterogeneity--identical biological concepts encoded under incompatible metadata schemas across datasets; and (ii) statistical heterogeneity--distribution shifts from biological variation demanding dataset-specific inductive biases. We propose HarmonyCell, an end-to-end agent framework resolving each challenge through a dedicated mechanism: an LLM-driven Semantic Unifier autonomously maps disparate metadata into a canonical interface without manual intervention; and an adaptive Monte Carlo Tree Search engine operates over a hierarchical action space to synthesize architectures with optimal statistical inductive biases for distribution shifts. Evaluated across diverse perturbation tasks under both semantic and distribution shifts, HarmonyCell achieves a 95% valid execution rate on heterogeneous input datasets (versus 0% for general agents) while matching or even exceeding expert-designed baselines in rigorous out-of-distribution evaluations. This dual-track orchestration enables scalable automatic virtual cell modeling without dataset-specific engineering.

arXiv:2603.01407v1 Announce Type: new Abstract: Autonomous agents operating in complex, multi-agent environments must reason about what is true from multiple perspectives. Existing approaches often struggle to integrate the reasoning of different agents, at different times, and in different contexts, typically handling these dimensions in separate, specialized modules. This fragmentation leads to a brittle and incomplete reasoning process, particularly when agents must understand the beliefs of others (Theory of Mind). We introduce the Observer-Situation Lattice (OSL), a unified mathematical structure that provides a single, coherent semantic space for perspective-aware cognition. OSL is a finite complete lattice where each element represents a unique observer-situation pair, allowing for a principled and scalable approach to belief management. We present two key algorithms that operate on this lattice: (i) Relativized Belief Propagation, an incremental update algorithm that efficiently propagates new information, and (ii) Minimal Contradiction Decomposition, a graph-based procedure that identifies and isolates contradiction components. We prove the theoretical soundness of our framework and demonstrate its practical utility through a series of benchmarks, including classic Theory of Mind tasks and a comparison with established paradigms such as assumption-based truth maintenance systems. Our results show that OSL provides a computationally efficient and expressive foundation for building robust, perspective-aware autonomous agents.

arXiv:2603.01409v1 Announce Type: new Abstract: Large Language Models (LLMs) often fail to generate correct code on the first attempt, which requires using generated unit tests as verifiers to validate the solutions. Despite the success of recent verification methods, they remain constrained by a "scaling-by-quantity" paradigm. This brute-force approach suffers from a critical limitation: it yields diminishing returns in fault detection while causing severe test redundancy. To address this, we propose MIST-RL (Mutation-based Incremental Suite Testing via Reinforcement Learning), a framework that shifts the focus to "scaling-by-utility". We formulate test generation as a sequential decision process optimized via Group Relative Policy Optimization (GRPO). Specifically, we introduce a novel incremental mutation reward combined with dynamic penalties, which incentivizes the model to discover new faults while it suppresses functionally equivalent assertions. Experiments on HumanEval+ and MBPP+ demonstrate that MIST-RL outperforms state-of-the-art baselines. It achieves a +28.5% higher mutation score while reducing the number of test cases by 19.3%. Furthermore, we show that these compact, high-utility tests serve as superior verifiers, which improves downstream code reranking accuracy on HumanEval+ by 3.05% over the SOTA baseline with 10 candidate samples. The source code and data are provided in the supplementary material.

arXiv:2603.01410v1 Announce Type: new Abstract: Knowledge graphs provide structured and reliable information for many real-world applications, motivating increasing interest in combining large language models (LLMs) with graph-based retrieval to improve factual grounding. Recent Graph-based Retrieval-Augmented Generation (GraphRAG) methods therefore introduce iterative interaction between LLMs and knowledge graphs to enhance reasoning capability. However, existing approaches typically depend on manually designed guidance and interact with knowledge graphs through a limited set of predefined tools, which substantially constrains graph exploration. To address these limitations, we propose GraphScout, a training-centric agentic graph reasoning framework equipped with more flexible graph exploration tools. GraphScout enables models to autonomously interact with knowledge graphs to synthesize structured training data which are then used to post-train LLMs, thereby internalizing agentic graph reasoning ability without laborious manual annotation or task curation. Extensive experiments across five knowledge-graph domains show that a small model (e.g., Qwen3-4B) augmented with GraphScout outperforms baseline methods built on leading LLMs (e.g., Qwen-Max) by an average of 16.7\% while requiring significantly fewer inference tokens. Moreover, GraphScout exhibits robust cross-domain transfer performance. Our code will be made publicly available~\footnote{https://github.com/Ying-Yuchen/_GraphScout_}.

arXiv:2603.01416v1 Announce Type: new Abstract: Recent advances in Vision-Language Models (VLMs) have motivated the development of multi-modal search agents that can actively invoke external search tools and integrate retrieved evidence through multi-step reasoning. While promising, existing approaches typically rely on large-scale supervised trajectories or expensive reinforcement learning (RL), leading to high training cost, instability, and a severe cold-start problem for standard VLMs. We propose a training-free paradigm to empower VLMs with autonomous search capabilities via cross-modal model merging. By fusing a text-based search agent with a base VLM, we show that multi-modal search capabilities can be effectively composed without any additional multi-modal training data. To mitigate parameter interference during cross-modal integration, we introduce Optimal Brain Merging (OBM), a saliency-aware merging algorithm that identifies task-critical parameters based on their impact on model loss using only a small set of calibration samples. Extensive experiments on search-intensive benchmarks (e.g., InfoSeek, MMSearch) reveal that: (1) Model merging secures a reasonable performance floor as a zero-shot agent, with OBM achieving superior search rates; (2) OBM significantly raises the performance ceiling as a warm-start strategy, achieving faster convergence and higher peak accuracy than standard VLM initialization.

arXiv:2603.01421v1 Announce Type: new Abstract: Automated scientific discovery with large language models is transforming the research lifecycle from ideation to experimentation, yet existing agents struggle to autonomously process raw data collected from scientific experiments. We introduce SciDER, a data-centric end-to-end system that automates the research lifecycle. Unlike traditional frameworks, our specialized agents collaboratively parse and analyze raw scientific data, generate hypotheses and experimental designs grounded in specific data characteristics, and write and execute corresponding code. Evaluation on three benchmarks shows SciDER excels in specialized data-driven scientific discovery and outperforms general-purpose agents and state-of-the-art models through its self-evolving memory and critic-led feedback loop. Distributed as a modular Python package, we also provide easy-to-use PyPI packages with a lightweight web interface to accelerate autonomous, data-driven research and aim to be accessible to all researchers and developers.

arXiv:2603.01437v1 Announce Type: new Abstract: As chain-of-thought (CoT) has become central to scaling reasoning capabilities in large language models (LLMs), it has also emerged as a promising tool for interpretability, suggesting the opportunity to understand model decisions through verbalized reasoning. However, the utility of CoT toward interpretability depends upon its faithfulness -- whether the model's stated reasoning reflects the underlying decision process. We provide mechanistic evidence that instruction-tuned models often determine their answer before generating CoT. Training linear probes on residual stream activations at the last token before CoT, we can predict the model's final answer with 0.9 AUC on most tasks. We find that these directions are not only predictive, but also causal: steering activations along the probe direction flips model answers in over 50% of cases, significantly exceeding orthogonal baselines. When steering induces incorrect answers, we observe two distinct failure modes: non-entailment (stating correct premises but drawing unsupported conclusions) and confabulation (fabricating false premises). While post-hoc reasoning may be instrumentally useful when the model has a correct pre-CoT belief, these failure modes suggest it can result in undesirable behaviors when reasoning from a false belief.

arXiv:2603.01452v1 Announce Type: new Abstract: Developing generalist robots capable of mastering diverse skills remains a central challenge in embodied AI. While recent progress emphasizes scaling model parameters and offline datasets, such approaches are limited in robotics, where learning requires active interaction. We argue that effective online learning should scale the \emph{number of tasks}, rather than the number of samples per task. This regime reveals a structural advantage of model-based reinforcement learning (MBRL). Because physical dynamics are invariant across tasks, a shared world model can aggregate multi-task experience to learn robust, task-agnostic representations. In contrast, model-free methods suffer from gradient interference when tasks demand conflicting actions in similar states. Task diversity therefore acts as a regularizer for MBRL, improving dynamics learning and sample efficiency. We instantiate this idea with \textbf{EfficientZero-Multitask (EZ-M)}, a sample-efficient multi-task MBRL algorithm for online learning. Evaluated on \textbf{HumanoidBench}, a challenging whole-body control benchmark, EZ-M achieves state-of-the-art performance with significantly higher sample efficiency than strong baselines, without extreme parameter scaling. These results establish task scaling as a critical axis for scalable robotic learning. The project website is available \href{https://yewr.github.io/ez_m/}{here}.

arXiv:2603.01464v1 Announce Type: new Abstract: Protein analysis tasks arising in healthcare settings often require accurate reasoning under protein sequence constraints, involving tasks such as functional interpretation of disease-related variants, protein-level analysis for clinical research, and similar scenarios. To address such tasks, search agents are introduced to search protein-related information, providing support for disease-related variant analysis and protein function reasoning in protein-centric inference. However, such search agents are mostly limited to single-round, text-only modality search, which prevents the protein sequence modality from being incorporated as a multimodal input into the search decision-making process. Meanwhile, their reliance on reinforcement learning (RL) supervision that focuses solely on the final answer results in a lack of search process constraints, making deviations in keyword selection and reasoning directions difficult to identify and correct in a timely manner. To address these limitations, we propose ProtRLSearch, a multi-round protein search agent trained with multi-dimensional reward based RL, which jointly leverages protein sequence and text as multimodal inputs during real-time search to produce high quality reports. To evaluate the ability of models to integrate protein sequence information and text-based multimodal inputs in realistic protein query settings, we construct ProtMCQs, a benchmark of 3,000 multiple choice questions (MCQs) organized into three difficulty levels. The benchmark evaluates protein query tasks that range from sequence constrained reasoning about protein function and phenotype changes to comprehensive protein reasoning that integrates multi-dimensional sequence features with signal pathways and regulatory networks.

arXiv:2603.01481v1 Announce Type: new Abstract: Optimizing large language models for industrial sales requires balancing long-term commercial objectives (e.g., conversion rate) with immediate linguistic constraints such as fluency and compliance. Conventional reinforcement learning often merges these heterogeneous goals into a single reward, causing high-magnitude session-level rewards to overwhelm subtler turn-level signals, which leads to unstable training or reward hacking. To address this issue, we propose Dual-Horizon Credit Assignment (DuCA), a framework that disentangles optimization across time scales. Its core, Horizon-Independent Advantage Normalization (HIAN), separately normalizes advantages from turn-level and session-level rewards before fusion, ensuring balanced gradient contributions from both immediate and long-term objectives to the policy update. Extensive experiments with a high-fidelity user simulator show DuCA outperforms the state-of-the-art GRPO baseline, achieving a 6.82% relative improvement in conversion rate, reducing inter-sentence repetition by 82.28%, and lowering identity detection rate by 27.35%, indicating a substantial improvement for an industrial sales scenario that effectively balances the dual demands of strategic performance and naturalistic language generation.

arXiv:2603.01486v1 Announce Type: new Abstract: Accurately mapping user queries to business categories is a fundamental Information Retrieval challenge for multi-category marketplaces, where context-sparse queries such as "Wildflower" exhibit intent ambiguity, simultaneously denoting a restaurant chain, a retail product, and a floral item. Traditional classifiers force a winner-takes-all assignment, while general-purpose LLMs hallucinate unavailable inventory. We introduce an Agentic Multi-Source Grounded system that addresses both failure modes by grounding LLM inference in (i) a staged catalog entity retrieval pipeline and (ii) an agentic web-search tool invoked autonomously for cold-start queries. Rather than predicting a single label, the model emits an ordered multi-intent set, resolved by a configurable disambiguation layer that applies deterministic business policies and is designed for extensibility to personalization signals. This decoupled design generalizes across domains, allowing any marketplace to supply its own grounding sources and resolution rules without modifying the core architecture. Evaluated on DoorDash's multi-vertical search platform, the system achieves +10.9pp over the ungrounded LLM baseline and +4.6pp over the legacy production system. On long-tail queries, incremental ablations attribute +8.3pp to catalog grounding, +3.2pp to agentic web search grounding, and +1.5pp to dual intent disambiguation, yielding 90.7% accuracy (+13.0pp over baseline). The system is deployed in production, serving over 95% of daily search impressions, and establishes a generalizable paradigm for applications requiring foundation models grounded in proprietary context and real-time web knowledge to resolve ambiguous, context-sparse decision problems at scale.

arXiv:2603.01488v1 Announce Type: new Abstract: Despite achieving remarkable success in complex tasks, Deep Reinforcement Learning (DRL) is still suffering from critical issues in practical applications, such as low data efficiency, lack of interpretability, and limited cross-environment transferability. However, the learned policy generating actions based on states are sensitive to the environmental changes, struggling to guarantee behavioral safety and compliance. Recent research shows that integrating Large Language Models (LLMs) with symbolic planning is promising in addressing these challenges. Inspired by this, we introduce a novel LLM-driven closed-loop framework, which enables semantic-driven skill reuse and real-time constraint monitoring by mapping natural language instructions into executable rules and semantically annotating automatically created options. The proposed approach utilizes the general knowledge of LLMs to facilitate exploration efficiency and adapt to transferable options for similar environments, and provides inherent interpretability through semantic annotations. To validate the effectiveness of this framework, we conduct experiments on two domains, Office World and Montezuma's Revenge, respectively. The results demonstrate superior performance in data efficiency, constraint compliance, and cross-task transferability.

arXiv:2603.01511v1 Announce Type: new Abstract: Accurate identification of protein active sites at the residue level is crucial for understanding protein function and advancing drug discovery. However, current methods face two critical challenges: vulnerability in single-instance prediction due to sparse training data, and inadequate modality reliability estimation that leads to performance degradation when unreliable modalities dominate fusion processes. To address these challenges, we introduce Multimodal Mixture-of-Experts with Retrieval Augmentation (MERA), the first retrieval-augmented framework for protein active site identification. MERA employs hierarchical multi-expert retrieval that dynamically aggregates contextual information from chain, sequence, and active-site perspectives through residue-level mixture-of-experts gating. To prevent modality degradation, we propose a reliability-aware fusion strategy based on Dempster-Shafer evidence theory that quantifies modality trustworthiness through belief mass functions and learnable discounting coefficients, enabling principled multimodal integration. Extensive experiments on ProTAD-Gen and TS125 datasets demonstrate that MERA achieves state-of-the-art performance, with 90% AUPRC on active site prediction and significant gains on peptide-binding site identification, validating the effectiveness of retrieval-augmented multi-expert modeling and reliability-guided fusion.

arXiv:2603.01537v1 Announce Type: new Abstract: The contributions of model complexity, data volume, and feature modalities to knowledge graph-based drug repurposing remain poorly quantified under rigorous temporal validation. We constructed a pharmacology knowledge graph from ChEMBL 36 comprising 5,348 entities including 3,127 drugs, 1,156 proteins, and 1,065 indications. A strict temporal split was enforced with training data up to 2022 and testing data from 2023 to 2025, together with biologically verified hard negatives mined from failed assays and clinical trials. We benchmarked five knowledge graph embedding models and a standard graph neural network with 3.44 million parameters that incorporates drug chemical structure using a graph attention encoder and ESM-2 protein embeddings. Scaling experiments ranging from 0.78 to 9.75 million parameters and from 25 to 100 percent of the data, together with feature ablation studies, were used to isolate the contributions of model capacity, graph density, and node feature modalities. Removing the graph attention based drug structure encoder and retaining only topological embeddings combined with ESM-2 protein features improved drug protein PR-AUC from 0.5631 to 0.5785 while reducing VRAM usage from 5.30 GB to 353 MB. Replacing the drug encoder with Morgan fingerprints further degraded performance, indicating that explicit chemical structure representations can be detrimental for predicting pharmacological network interactions. Increasing model size beyond 2.44 million parameters yielded diminishing returns, whereas increasing training data consistently improved performance. External validation confirmed 6 of the top 14 novel predictions as established therapeutic indications. These results show that drug pharmacological behavior can be accurately predicted using target-centric information and drug network topology alone, without requiring explicit chemical structure representations.

arXiv:2603.01548v1 Announce Type: new Abstract: Tool-using LLM agents face a reliability-cost tradeoff: routing every decision through the LLM improves correctness but incurs high latency and inference cost, while pre-coded workflow graphs reduce cost but become brittle under unanticipated compound tool failures. We present Self-Healing Router, a fault-tolerant orchestration architecture that treats most agent control-flow decisions as routing rather than reasoning. The system combines (i) parallel health monitors that assign priority scores to runtime conditions such as tool outages and risk signals, and (ii) a cost-weighted tool graph where Dijkstra's algorithm performs deterministic shortest-path routing. When a tool fails mid-execution, its edges are reweighted to infinity and the path is recomputed -- yielding automatic recovery without invoking the LLM. The LLM is reserved exclusively for cases where no feasible path exists, enabling goal demotion or escalation. Prior graph-based tool-use systems (ControlLLM, ToolNet, NaviAgent) focus on tool selection and planning; our contribution is runtime fault tolerance with deterministic recovery and binary observability -- every failure is either a logged reroute or an explicit escalation, never a silent skip. Across 19 scenarios spanning three graph topologies (linear pipeline, dependency DAG, parallel fan-out), Self-Healing Router matches ReAct's correctness while reducing control-plane LLM calls by 93% (9 vs 123 aggregate) and eliminating the silent-failure cases observed in a well-engineered static workflow baseline under compound failures.

arXiv:2603.01553v1 Announce Type: new Abstract: Signal delay poses a fundamental challenge in continuous control and reinforcement learning (RL) by introducing a temporal gap between interaction and perception. Current solutions have largely evolved along two distinct paradigms: model-free approaches which utilize state augmentation to preserve Markovian properties, and model-based methods which focus on inferring latent beliefs via dynamics modeling. In this paper, we bridge these perspectives by introducing State-Action Inpainting Diffuser (SAID), a framework that integrates the inductive bias of dynamics learning with the direct decision-making capability of policy optimization. By formulating the problem as a joint sequence inpainting task, SAID implicitly captures environmental dynamics while directly generating consistent plans, effectively operating at the intersection of model-based and model-free paradigms. Crucially, this generative formulation allows SAID to be seamlessly applied to both online and offline RL. Extensive experiments on delayed continuous control benchmarks demonstrate that SAID achieves state-of-the-art and robust performance. Our study suggests a new methodology to advance the field of RL with delay.

arXiv:2603.01554v1 Announce Type: new Abstract: The smart home is a key domain within the Society 5.0 vision for a human-centered society. Smart home technologies rapidly evolve, and research should diversify while remaining aligned with Society 5.0 objectives. Democratizing smart home research would engage a broader community of innovators beyond traditional limited experts. This shift necessitates inclusive simulation frameworks that support research across diverse fields in industry and academia. However, existing smart home simulators require significant technical expertise, offer limited adaptability, and lack automated evolution, thereby failing to meet the holistic needs of Society 5.0. These constraints impede researchers from efficiently conducting simulations and experiments for security, energy, health, climate, and socio-economic research. To address these challenges, this paper presents the Society 5.0-driven Smart Home Environment Simulator Agent (S5-HES Agent), an agentic simulation framework that transforms traditional smart home simulation through autonomous AI orchestration. The framework coordinates specialized agents through interchangeable large language models (LLMs), enabling natural-language-driven end-to-end smart home simulation configuration without programming expertise. A retrieval-augmented generation (RAG) pipeline with semantic, keyword, and hybrid search retrieves smart home knowledge. Comprehensive evaluation on S5-HES Agent demonstrates that the RAG pipeline achieves near-optimal retrieval fidelity, simulated device behaviour and threat scenarios align with real-world IoT datasets, and simulation engine scales predictably across home configurations, establishing a stable foundation for Society 5.0 smart home research. Source code is available under the MIT License at https://github.com/AsiriweLab/S5-HES-Agent.

arXiv:2603.01557v1 Announce Type: new Abstract: Large language models (LLMs) can generate fluent clinical summaries of remote therapeutic monitoring time series. However, it remains unclear whether these narratives faithfully capture clinically significant events, such as sustained abnormalities. Existing evaluation metrics primarily focus on semantic similarity and linguistic quality, leaving event-level correctness largely unmeasured. To address this gap, we introduce an event-based evaluation framework for multimodal time-series summarization using the Technology-Integrated Health Management (TIHM)-1.5 dementia monitoring dataset. Clinically grounded daily events are derived through rule-based abnormal thresholds and temporal persistence criteria. Model-generated summaries are then aligned with these structured facts. Our evaluation protocol measures abnormality recall, duration recall, measurement coverage, and hallucinated event mentions. We benchmark three approaches: zero-shot prompting, statistical prompting, and a vision-based pipeline that uses rendered time-series visualizations. The results reveal a striking decoupling between conventional metrics and clinical event fidelity. Models that achieve high semantic similarity scores often exhibit near-zero abnormality recall. In contrast, the vision-based approach demonstrates the strongest event alignment, achieving 45.7% abnormality recall and 100% duration recall. These findings underscore the importance of event-aware evaluation to ensure reliable clinical time-series summarization.

arXiv:2603.01562v2 Announce Type: new Abstract: As Large Language Model (LLM) alignment evolves from simple completions to complex, highly sophisticated generation, Reward Models are increasingly shifting toward rubric-guided evaluation to mitigate surface-level biases. However, the community lacks a unified benchmark to assess this evaluation paradigm, as existing benchmarks lack both the discriminative complexity and the ground-truth rubric annotations required for rigorous analysis. To bridge this gap, we introduce RubricBench, a curated benchmark with 1,147 pairwise comparisons specifically designed to assess the reliability of rubric-based evaluation. Our construction employs a multi-dimensional filtration pipeline to target hard samples featuring nuanced input complexity and misleading surface bias, augmenting each with expert-annotated, atomic rubrics derived strictly from instructions. Comprehensive experiments reveal a substantial capability gap between human-annotated and model-generated rubrics, indicating that even state-of-the-art models struggle to autonomously specify valid evaluation criteria, lagging considerably behind human-guided performance.

arXiv:2603.01571v1 Announce Type: new Abstract: Recent advancements in Generative Reward Models (GRMs) have demonstrated that scaling the length of Chain-of-Thought (CoT) reasoning considerably enhances the reliability of evaluation. However, current works predominantly rely on unstructured length scaling, ignoring the divergent efficacy of different reasoning mechanisms: Breadth-CoT (B-CoT, i.e., multi-dimensional principle coverage) and Depth-CoT (D-CoT, i.e., substantive judgment soundness). To address this, we introduce Mix-GRM, a framework that reconfigures raw rationales into structured B-CoT and D-CoT through a modular synthesis pipeline, subsequently employing Supervised Fine-Tuning (SFT) and Reinforcement Learning with Verifiable Rewards (RLVR) to internalize and optimize these mechanisms. Comprehensive experiments demonstrate that Mix-GRM establishes a new state-of-the-art across five benchmarks, surpassing leading open-source RMs by an average of 8.2\%. Our results reveal a clear divergence in reasoning: B-CoT benefits subjective preference tasks, whereas D-CoT excels in objective correctness tasks. Consequently, misaligning the reasoning mechanism with the task directly degrades performance. Furthermore, we demonstrate that RLVR acts as a switching amplifier, inducing an emergent polarization where the model spontaneously allocates its reasoning style to match task demands. The synthesized data and models are released at \href{https://huggingface.co/collections/DonJoey/mix-grm}{Hugging Face}, and the code is released at \href{https://github.com/Don-Joey/Mix-GRM}{Github}.

arXiv:2603.01607v1 Announce Type: new Abstract: Large visual language models (VLMs) have shown strong multi-modal medical reasoning ability, but most operate as end-to-end black boxes, diverging from clinicians' evidence-based, staged workflows and hindering clinical accountability. Complementarily, expert visual grounding models can accurately localize regions of interest (ROIs), providing explicit, reliable evidence that improves both reasoning accuracy and trust. In this paper, we introduce CARE, advancing Clinical Accountability in multi-modal medical Reasoning with an Evidence-grounded agentic framework. Unlike existing approaches that couple grounding and reasoning within a single generalist model, CARE decomposes the task into coordinated sub-modules to reduce shortcut learning and hallucination: a compact VLM proposes relevant medical entities; an expert entity-referring segmentation model produces pixel-level ROI evidence; and a grounded VLM reasons over the full image augmented by ROI hints. The VLMs are optimized with reinforcement learning with verifiable rewards to align answers with supporting evidence. Furthermore, a VLM coordinator plans tool invocation and reviews evidence-answer consistency, providing agentic control and final verification. Evaluated on standard medical VQA benchmarks, our CARE-Flow (coordinator-free) improves average accuracy by 10.9% over the same size (10B) state-of-the-art (SOTA). With dynamic planning and answer review, our CARE-Coord yields a further gain, outperforming the heavily pre-trained SOTA by 5.2%. Our experiments demonstrate that an agentic framework that emulates clinical workflows, incorporating decoupled specialized models and explicit evidence, yields more accurate and accountable medical AI.

arXiv:2603.01608v1 Announce Type: new Abstract: As frontier language models are increasingly deployed as autonomous agents pursuing complex, long-term objectives, there is increased risk of scheming: agents covertly pursuing misaligned goals. Prior work has focused on showing agents are capable of scheming, but their propensity to scheme in realistic scenarios remains underexplored. To understand when agents scheme, we decompose scheming incentives into agent factors and environmental factors. We develop realistic settings allowing us to systematically vary these factors, each with scheming opportunities for agents that pursue instrumentally convergent goals such as self-preservation, resource acquisition, and goal-guarding. We find only minimal instances of scheming despite high environmental incentives, and show this is unlikely due to evaluation awareness. While inserting adversarially-designed prompt snippets that encourage agency and goal-directedness into an agent's system prompt can induce high scheming rates, snippets used in real agent scaffolds rarely do. Surprisingly, in model organisms (Hubinger et al., 2023) built with these snippets, scheming behavior is remarkably brittle: removing a single tool can drop the scheming rate from 59% to 3%, and increasing oversight can raise rather than deter scheming by up to 25%. Our incentive decomposition enables systematic measurement of scheming propensity in settings relevant for deployment, which is necessary as agents are entrusted with increasingly consequential tasks.

arXiv:2603.01620v1 Announce Type: new Abstract: Tool-integrated reasoning agents interleaving natural language deliberation with external API calls show promise for complex multi-step tasks. However, aligning such agents for high-stakes domain-specific deployment is challenging, as existing reinforcement learning uses coarse binary rewards (success/failure) that insufficiently guide nuanced tool invocation in production. We present ToolRLA, a three-stage post-training pipeline (Supervised Fine-Tuning, Group Relative Policy Optimization, Direct Preference Optimization) for domain-specific tool-integrated agents. Its core is a fine-grained reward function with multiplicative correctness decomposition, evaluating tool invocation across four dimensions: format validity, tool selection correctness, invocation efficiency, and domain constraint compliance. Multiplicative composition prioritizes correct tool selection (a prerequisite for meaningful parameter evaluation), while a large negative compliance penalty ({\lambda}=10) ensures regulatory adherence. Deployed on a real-world financial advisory copilot (80+ advisors, 1,200+ daily queries, 15+ heterogeneous APIs), ToolRLA achieves 47% higher end-to-end task completion (62% to 91%), 63% lower tool invocation error (38% to 14%), 93% lower regulatory violation (12% to 0.8%), and sub-2-second latency after three months. Ablation studies confirm fine-grained reward decomposition contributes 7 percentage points over coarse additive rewards; generalizability is validated on ToolBench and API-Bank.

arXiv:2603.01630v1 Announce Type: new Abstract: As autonomous systems such as drones, become increasingly deployed in high-stakes, human-centric domains, it is critical to evaluate the ethical alignment since failure to do so imposes imminent danger to human lives, and long term bias in decision-making. Automated ethical benchmarking of these systems is understudied due to the lack of ubiquitous, well-defined metrics for evaluation, and stakeholder-specific subjectivity, which cannot be modeled analytically. To address these challenges, we propose SEED-SET, a Bayesian experimental design framework that incorporates domain-specific objective evaluations, and subjective value judgments from stakeholders. SEED-SET models both evaluation types separately with hierarchical Gaussian Processes, and uses a novel acquisition strategy to propose interesting test candidates based on learnt qualitative preferences and objectives that align with the stakeholder preferences. We validate our approach for ethical benchmarking of autonomous agents on two applications and find our method to perform the best. Our method provides an interpretable and efficient trade-off between exploration and exploitation, by generating up to $2\times$ optimal test candidates compared to baselines, with $1.25\times$ improvement in coverage of high dimensional search spaces.

arXiv:2603.01641v1 Announce Type: new Abstract: Large language models can exhibit emergent reasoning behaviors, often manifested as recurring lexical patterns (e.g., "wait," indicating verification). However, complex reasoning trajectories remain sparse in unconstrained sampling, and standard RL often fails to guarantee the acquisition of diverse reasoning behaviors. We propose a systematic discovery and reinforcement of diverse reasoning patterns through structured reasoning, a paradigm that requires targeted exploration of specific reasoning patterns during the RL process. To this end, we propose Ctrl-R, a framework for learning structured reasoning via tractable trajectory control that actively guides the rollout process, incentivizing the exploration of diverse reasoning patterns that are critical for complex problem-solving. The resulting behavior policy enables accurate importance-sampling estimation, supporting unbiased on-policy optimization. We further introduce a power-scaling factor on the importance-sampling weights, allowing the policy to selectively learn from exploratory, out-of-distribution trajectories while maintaining stable optimization. Experiments demonstrate that Ctrl-R enables effective exploration and internalization of previously unattainable reasoning patterns, yielding consistent improvements across language and vision-language models on mathematical reasoning tasks.

arXiv:2603.01654v1 Announce Type: new Abstract: The development of chemical processes, a cornerstone of chemical engineering, presents formidable challenges due to its multi-faceted nature, integrating specialized knowledge, conceptual design, and parametric simulation. Capitalizing on this, we propose CeProAgents, a hierarchical multi-agent system designed to automate the development of chemical process through collaborative division of labor. Our architecture comprises three specialized agent cohorts focused on knowledge, concept, and parameter respectively. To effectively adapt to the inherent complexity of chemical tasks, each cohort employs a novel hybrid architecture that integrates dynamic agent chatgroups with structured agentic workflows. To rigorously evaluate the system, we establish CeProBench, a multi-dimensional benchmark structured around three core pillars of chemical engineering. We design six distinct types of tasks across these dimensions to holistically assess the comprehensive capabilities of the system in chemical process development. The results not only confirm the effectiveness and superiority of our proposed approach but also reveal the transformative potential as well as the current boundaries of Large Language Models (LLMs) for industrial chemical engineering.

arXiv:2603.01667v1 Announce Type: new Abstract: Multi-task Vehicle Routing Problems (VRPs) aim to minimize routing costs while satisfying diverse constraints. Existing solvers typically adopt a unified reinforcement learning (RL) framework to learn generalizable patterns across tasks. However, they often overlook the constraint and node dynamics during the decision process, making the model fail to accurately react to the current context. To address this limitation, we propose Chain-of-Context Learning (CCL), a novel framework that progressively captures the evolving context to guide fine-grained node adaptation. Specifically, CCL constructs step-wise contextual information via a Relevance-Guided Context Reformulation (RGCR) module, which adaptively prioritizes salient constraints. This context then guides node updates through a Trajectory-Shared Node Re-embedding (TSNR) module, which aggregates shared node features from all trajectories' contexts and uses them to update inputs for the next step. By modeling evolving preferences of the RL agent, CCL captures step-by-step dependencies in sequential decision-making. We evaluate CCL on 48 diverse VRP variants, including 16 in-distribution and 32 out-of-distribution (with unseen constraints) tasks. Experimental results show that CCL performs favorably against the state-of-the-art baselines, achieving the best performance on all in-distribution tasks and the majority of out-of-distribution tasks.

arXiv:2603.01712v1 Announce Type: new Abstract: Fine-tuning large language models for vertical domains remains a labor-intensive and expensive process, requiring domain experts to curate data, configure training, and iteratively diagnose model behavior. Despite growing interest in autonomous machine learning, no prior work has tackled end-to-end LLM fine-tuning with agents. Can LLM-based agents automate this complete process? We frame this as a substantially open problem: agents must navigate an open-ended search space spanning data curation from diverse data sources, processing with complex tools, building a training pipeline, and iteratively refining their approach based on evaluation outcomes in rapidly growing logs--an overall scenario far more intricate than existing benchmarks. To study this question, we introduce FT-Dojo, an interactive environment comprising 13 tasks across 5 domains. We further develop FT-Agent, an autonomous system that mirrors human experts by leveraging evaluation-driven feedback to iteratively diagnose failures and refine fine-tuning strategies. Experiments on FT-Dojo demonstrate that purpose-built fine-tuning agents significantly outperform general-purpose alternatives, with FT-Agent achieving the best performance on 10 out of 13 tasks across all five domains. Ablations show that the approach generalizes effectively to 3B models, with additional insights on data scaling trade-offs and backbone sensitivity. Case analyses reveal that agents can recover from failures through cumulative learning from historical experience, while also exposing fundamental limitations in causal reasoning--highlighting both the promise and current boundaries of autonomous LLM fine-tuning.

arXiv:2603.01724v1 Announce Type: new Abstract: Online content moderation is essential for maintaining a healthy digital environment, and reliance on AI for this task continues to grow. Consider a user comment using national stereotypes to insult a politician. This example illustrates two critical challenges in real-world scenarios: (1) Co-occurring Violations, where a single post violates multiple policies (e.g., prejudice and personal attacks); (2) Dynamic rules of moderation, where determination of a violation depends on platform-specific guidelines that evolve across contexts . The intersection of co-occurring harms and dynamically changing rules highlights a core limitation of current AI systems: although large language models (LLMs) are adept at following fixed guidelines, their judgment capabilities degrade when policies are unstable or context-dependent . In practice, such shortcomings lead to inconsistent moderation: either erroneously restricting legitimate expression or allowing harmful content to remain online . This raises a critical question for evaluation: Does high performance on existing static benchmarks truly guarantee robust generalization of AI judgment to real-world scenarios involving co-occurring violations and dynamically changing rules?

arXiv:2603.01783v1 Announce Type: new Abstract: Retrieval-Augmented Generation (RAG) grounds large language models with external evidence, but many implementations rely on pre-built indices that remain static after construction. Related queries therefore repeat similar multi-hop traversal, increasing latency and compute. Motivated by schema-based learning in cognitive neuroscience, we propose GAM-RAG, a training-free framework that accumulates retrieval experience from recurring or related queries and updates retrieval memory over time. GAM-RAG builds a lightweight, relation-free hierarchical index whose links capture potential co-occurrence rather than fixed semantic relations. During inference, successful retrieval episodes provide sentence-level feedback, updating sentence memories so evidence useful for similar reasoning types becomes easier to activate later. To balance stability and adaptability under noisy feedback, we introduce an uncertainty-aware, Kalman-inspired gain rule that jointly updates memory states and perplexity-based uncertainty estimates. It applies fast updates for reliable novel signals and conservative refinement for stable or noisy memories. We provide a theoretical analysis of the update dynamics, and empirically show that GAM-RAG improves average performance by 3.95% over the strongest baseline and by 8.19% with 5-turn memory, while reducing inference cost by 61%. Our code and datasets are available at: https://anonymous.4open.science/r/GAM_RAG-2EF6.

arXiv:2603.01799v1 Announce Type: new Abstract: More and more, data is being produced in a streaming fashion. This has led to increased interest into how actionable insights can be extracted in real time from data streams through Stream Reasoning. Reasoning over data streams raises multiple challenges, notably the high velocity of data, the real time requirement of the reasoning, and the noisy and volatile nature of streams. This paper proposes novel semantics for incremental reasoning over streams of Description Logic ABoxes, in order to tackle these challenges. To address the first two challenges, our semantics for reasoning over sliding windows on streams allow for incrementally computing the materialization of the window based on the materialization of the previous window. Furthermore, to deal with the volatile nature of streams, we present novel semantics for inconsistency repair on such windows, based on preferred repair semantics. We then detail our proposed semi-naive algorithms for incremental materialization maintenance in the case of OWL2 RL, both in the presence of inconsistencies and without.

arXiv:2603.01801v1 Announce Type: new Abstract: Automated paper reproduction -- generating executable code from academic papers -- is bottlenecked not by information retrieval but by the tacit knowledge that papers inevitably leave implicit. We formalize this challenge as the progressive recovery of three types of tacit knowledge -- relational, somatic, and collective -- and propose \method, a graph-based agent framework with a dedicated mechanism for each: node-level relation-aware aggregation recovers relational knowledge by analyzing implementation-unit-level reuse and adaptation relationships between the target paper and its citation neighbors; execution-feedback refinement recovers somatic knowledge through iterative debugging driven by runtime signals; and graph-level knowledge induction distills collective knowledge from clusters of papers sharing similar implementations. On an extended ReproduceBench spanning 3 domains, 10 tasks, and 40 recent papers, \method{} achieves an average performance gap of 10.04\% against official implementations, improving over the strongest baseline by 24.68\%. The code will be publicly released upon acceptance; the repository link will be provided in the final version.

arXiv:2603.01822v1 Announce Type: new Abstract: Both humans and Large Language Models (LLMs) store a vast repository of semantic memories. In humans, efficient and strategic access to this memory store is a critical foundation for a variety of cognitive functions. Such access has long been a focus of psychology and the computational mechanisms behind it are now well characterized. Much of this understanding has been gleaned from a widely-used neuropsychological and cognitive science assessment called the Semantic Fluency Task (SFT), which requires the generation of as many semantically constrained concepts as possible. Our goal is to apply mechanistic interpretability techniques to bring greater rigor to the study of semantic memory foraging in LLMs. To this end, we present preliminary results examining SFT as a case study. A central focus is on convergent and divergent patterns of generative memory search, which in humans play complementary strategic roles in efficient memory foraging. We show that these same behavioral signatures, critical to human performance on the SFT, also emerge as identifiable patterns in LLMs across distinct layers. Potentially, this analysis provides new insights into how LLMs may be adapted into closer cognitive alignment with humans, or alternatively, guided toward productive cognitive \emph{disalignment} to enhance complementary strengths in human-AI interaction.

arXiv:2603.01940v1 Announce Type: new Abstract: Developing multi-turn interactive tool-use agents is challenging because real-world user needs are often complex and ambiguous, yet agents must execute deterministic actions to satisfy them. To address this gap, we introduce \textbf{CoVe} (\textbf{Co}nstraint-\textbf{Ve}rification), a post-training data synthesis framework designed for training interactive tool-use agents while ensuring both data complexity and correctness. CoVe begins by defining explicit task constraints, which serve a dual role: they guide the generation of complex trajectories and act as deterministic verifiers for assessing trajectory quality. This enables the creation of high-quality training trajectories for supervised fine-tuning (SFT) and the derivation of accurate reward signals for reinforcement learning (RL). Our evaluation on the challenging $\tau^2$-bench benchmark demonstrates the effectiveness of the framework. Notably, our compact \textbf{CoVe-4B} model achieves success rates of 43.0\% and 59.4\% in the Airline and Retail domains, respectively; its overall performance significantly outperforms strong baselines of similar scale and remains competitive with models up to $17\times$ its size. These results indicate that CoVe provides an effective and efficient pathway for synthesizing training data for state-of-the-art interactive tool-use agents. To support future research, we open-source our code, trained model, and the full set of 12K high-quality trajectories used for training.

arXiv:2603.01952v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed as autonomous agents, yet evaluations focus primarily on task success rather than cultural appropriateness or evaluator reliability. We introduce LiveCultureBench, a multi-cultural, dynamic benchmark that embeds LLMs as agents in a simulated town and evaluates them on both task completion and adherence to socio-cultural norms. The simulation models a small city as a location graph with synthetic residents having diverse demographic and cultural profiles. Each episode assigns one resident a daily goal while others provide social context. An LLM-based verifier generates structured judgments on norm violations and task progress, which we aggregate into metrics capturing task-norm trade-offs and verifier uncertainty. Using LiveCultureBench across models and cultural profiles, we study (i) cross-cultural robustness of LLM agents, (ii) how they balance effectiveness against norm sensitivity, and (iii) when LLM-as-a-judge evaluation is reliable for automated benchmarking versus when human oversight is needed.

arXiv:2603.01990v1 Announce Type: new Abstract: Personalized AI assistants must recall and reason over long-term user memory, which naturally spans multiple modalities and sources such as images, videos, and emails. However, existing Long-term Memory benchmarks focus primarily on dialogue history, failing to capture realistic personalized references grounded in lived experience. We introduce ATM-Bench, the first benchmark for multimodal, multi-source personalized referential Memory QA. ATM-Bench contains approximately four years of privacy-preserving personal memory data and human-annotated question-answer pairs with ground-truth memory evidence, including queries that require resolving personal references, multi-evidence reasoning from multi-source and handling conflicting evidence. We propose Schema-Guided Memory (SGM) to structurally represent memory items originated from different sources. In experiments, we implement 5 state-of-the-art memory systems along with a standard RAG baseline and evaluate variants with different memory ingestion, retrieval, and answer generation techniques. We find poor performance (under 20\% accuracy) on the ATM-Bench-Hard set, and that SGM improves performance over Descriptive Memory commonly adopted in prior works. Code available at: https://github.com/JingbiaoMei/ATM-Bench

arXiv:2603.02029v1 Announce Type: new Abstract: Moving beyond evaluations that collapse performance across heterogeneous prompts toward fine-grained evaluation at the prompt level, or within relatively homogeneous subsets, is necessary to diagnose generative models' strengths and weaknesses. Such fine-grained evaluations, however, suffer from a data bottleneck: human gold-standard labels are too costly at this scale, while automated ratings are often misaligned with human judgment. To resolve this challenge, we propose a novel statistical model based on tensor factorization that merges cheap autorater data with a limited set of human gold-standard labels. Specifically, our approach uses autorater scores to pretrain latent representations of prompts and generative models, and then aligns those pretrained representations to human preferences using a small calibration set. This sample-efficient methodology is robust to autorater quality, more accurately predicts human preferences on a per-prompt basis than standard baselines, and provides tight confidence intervals for key statistical parameters of interest. We also showcase the practical utility of our method by constructing granular leaderboards based on prompt qualities and by estimating model performance solely from autorater scores, eliminating the need for additional human annotations.

arXiv:2603.02062v1 Announce Type: new Abstract: The rapid developments in artificial intelligence (AI) research in radiology have produced numerous models that are scattered across various platforms and sources, limiting discoverability, reproducibility and clinical translation. Herein, OpenRad was created, a curated, standardized, open-access repository that aggregates radiology AI models and providing details such as the availability of pretrained weights and interactive applications. Retrospective analysis of peer reviewed literature and preprints indexed in PubMed, arXiv and Scopus was performed until Dec 2025 (n = 5239 records). Model records were generated using a locally hosted LLM (gpt-oss:120b), based on the RSNA AI Roadmap JSON schema, and manually verified by ten expert reviewers. Stability of LLM outputs was assessed on 225 randomly selected papers using text similarity metrics. A total of 1694 articles were included after review. Included models span all imaging modalities (CT, MRI, X-ray, US) and radiology subspecialties. Automated extraction demonstrated high stability for structured fields (Levenshtein ratio > 90%), with 78.5% of record edits being characterized as minor during expert review. Statistical analysis of the repository revealed CNN and transformer architectures as dominant, while MRI was the most commonly used modality (in 621 neuroradiology AI models). Research output was mostly concentrated in China and the United States. The OpenRad web interface enables model discovery via keyword search and filters for modality, subspecialty, intended use, verification status and demo availability, alongside live statistics. The community can contribute new models through a dedicated portal. OpenRad contains approx. 1700 open access, curated radiology AI models with standardized metadata, supplemented with analysis of code repositories, thereby creating a comprehensive, searchable resource for the radiology community.

arXiv:2603.02070v1 Announce Type: new Abstract: When automating plan generation for a real-world sequential decision problem, the goal is often not to replace the human planner, but to facilitate an iterative reasoning and elicitation process, where the human's role is to guide the AI planner according to their preferences and expertise. In this context, explanations that respond to users' questions are crucial to improve their understanding of potential solutions and increase their trust in the system. To enable natural interaction with such a system, we present a multi-agent Large Language Model (LLM) architecture that is agnostic to the explanation framework and enables user- and context-dependent interactive explanations. We also describe an instantiation of this framework for goal-conflict explanations, which we use to conduct a user study comparing the LLM-powered interaction with a baseline template-based explanation interface.

arXiv:2603.02119v1 Announce Type: new Abstract: We introduce Pencil Puzzle Bench, a framework for evaluating large language model reasoning through pencil puzzles, a family of constraint-satisfaction problems closely related to NP-complete problems, with deterministic, step-level verification. From a database of 62,231 puzzles across 94 varieties with verified unique solutions, we select a benchmark of 300 puzzles spanning 20 varieties and evaluate 51 models from 11 providers in two modes: direct ask (single-shot) and agentic (multi-turn with iterative verification). A key differentiator of our benchmark is that every intermediate board state can be checked against variety-specific constraints, localizing errors to the exact rule violated, providing the infrastructure for dense, per-move reward signals for process supervision and reinforcement learning. Our evaluation reveals two distinct axes of capability: (1) reasoning effort scaling, where GPT-5.2 improves 81x from no reasoning to maximum effort; and (2) agentic iteration, where Claude Opus 4.6 rises from 0.3% to 30.0% through iterative checking, while GPT-5.2@xhigh improves from 20.2% to 56.0%. Agentic attempts span a median of 29 turns over 17 minutes, with the longest exceeding 1,221 turns and 14.3 hours - a demanding test of long-context utilization, not just reasoning.

arXiv:2603.02123v2 Announce Type: new Abstract: The development of affective multimodal language models (MLMs) has long been constrained by a gap between low-level perception and high-level interaction, leading to fragmented affective capabilities and limited generalization. To bridge this gap, we propose a cognitively inspired three-level hierarchy that organizes affective tasks according to their cognitive depth-perception, understanding, and interaction-and provides a unified conceptual foundation for advancing affective modeling. Guided by this hierarchy, we introduce Nano-EmoX, a small-scale multitask MLM, and P2E (Perception-to-Empathy), a curriculum-based training framework. Nano-EmoX integrates a suite of omni-modal encoders, including an enhanced facial encoder and a fusion encoder, to capture key multimodal affective cues and improve cross-task transferability. The outputs are projected into a unified language space via heterogeneous adapters, empowering a lightweight language model to tackle diverse affective tasks. Concurrently, P2E progressively cultivates emotional intelligence by aligning rapid perception with chain-of-thought-driven empathy. To the best of our knowledge, Nano-EmoX is the first compact MLM (2.2B) to unify six core affective tasks across all three hierarchy levels, achieving state-of-the-art or highly competitive performance across multiple benchmarks, demonstrating excellent efficiency and generalization.

arXiv:2603.02196v1 Announce Type: new Abstract: An agent must try new behaviors to explore and improve. In high-stakes environments, an agent that violates safety constraints may cause harm and must be taken offline, curtailing any future interaction. Imitating old behavior is safe, but excessive conservatism discourages exploration. How much behavior change is too much? We show how to use any safe reference policy as a probabilistic regulator for any optimized but untested policy. Conformal calibration on data from the safe policy determines how aggressively the new policy can act, while provably enforcing the user's declared risk tolerance. Unlike conservative optimization methods, we do not assume the user has identified the correct model class nor tuned any hyperparameters. Unlike previous conformal methods, our theory provides finite-sample guarantees even for non-monotonic bounded constraint functions. Our experiments on applications ranging from natural language question answering to biomolecular engineering show that safe exploration is not only possible from the first moment of deployment, but can also improve performance.

arXiv:2603.02203v1 Announce Type: new Abstract: Test-time reinforcement learning (TTRL) has emerged as a promising paradigm for self-evolving large reasoning models (LRMs), enabling online adaptation on unlabeled test inputs via self-induced rewards through majority voting. However, a spurious yet high-frequency unverified consensus can become a biased and reinforced reward signal, leading to incorrect mode collapse. We address this failure mode with T^3RL (Tool-Verification for Test-Time Reinforcement Learning), which introduces test-time tool verification into reward estimation. Concretely, a verifier uses an external tool as evidence (e.g., from code execution) to upweight verified rollouts in a verification-aware voting, producing more reliable pseudo-labels for training. Across various math difficulties (MATH-500, AMC, and AIME 2024) and diverse backbone types, T^3RL significantly improves over TTRL, with larger gains on harder problems. More broadly, T^3RL can be viewed as verified online data synthesis, highlighting test-time tool verification as a key mechanism for stabilizing self-evolution.

arXiv:2006.15598v1 Announce Type: cross Abstract: For the next wireless communication systems, the major challenges are to provide ubiquitous wireless access abilities and maintain the quality of service and seamless mobility management for mobile communication devices in heterogeneous networks. Due to the rapid growth of mobile users, this demand becomes more challenging where the users always require seamless connectivity while they move to other places at any time. As the number of users increases, the network load also increases, the handover process needs to be performed in an efficient way. However, in many situations, the handover blocking, and unnecessary handover frequently happens, then affect the network and reduces its performance. The problem arises with the movement of mobile users between base stations while the link connectivity becomes weaker and the mobile node tries to switch to another base station to have a better link quality during a call for higher QoS (Quality of Service). In this survey, it is aimed to gather together the studies that help to improve mobility management processes in 5G (Fifth Generation) cellular networks.

arXiv:2603.00008v1 Announce Type: cross Abstract: In order to make argumentation-based inference contestable, it is crucial to explain what changes can achieve a desired (instead of the contested) inference result. To this end, we introduce strength change explanations for quantitative (bipolar) argumentation graphs. Strength change explanations describe changes to the initial strengths of a subset of the arguments in a given graph that can achieve a desired ordering based on the final strengths of some (potentially different) subset of arguments. We show that the existing notions of inverse and counterfactual problems can be reduced to strength change explanations. We also prove basic soundness and completeness properties of our strength change explanations, and demonstrate their existence and non-existence in some special cases. By applying a heuristic search, we demonstrate that we can often successfully find strength change explanations for layered graphs that are common in typical application scenarios; still, limitations remain for settings where we do not provide guarantees for the presence (or absence) of explanations.

arXiv:2603.00016v1 Announce Type: cross Abstract: Augmented Reality (AR) offers powerful visualization capabilities for industrial robot training, yet current interfaces remain predominantly static, failing to account for learners' diverse cognitive profiles. In this paper, we present an AR application for robot training and propose a multi-agent AI framework for future integration that bridges the gap between static visualization and pedagogical intelligence. We report on the evaluation of the baseline AR interface with 36 participants performing a robotic pick-and-place task. While overall usability was high, notable disparities in task duration and learner characteristics highlighted the necessity for dynamic adaptation. To address this, we propose a multi-agent framework that orchestrates multiple components to perform complex preprocessing of multimodal inputs (e.g., voice, physiology, robot data) and adapt the AR application to the learner's needs. By utilizing autonomous Large Language Model (LLM) agents, the proposed system would dynamically adapt the learning environment based on advanced LLM reasoning in real-time.

arXiv:2603.00022v1 Announce Type: cross Abstract: Precision is of utmost importance in the realm of clinical entity extraction from clinical notes and reports. Encoder Models fine-tuned for Named Entity Recognition (NER) are an efficient choice for this purpose, as they don't hallucinate. We pre-trained an in-house BERT over clinical data and then fine-tuned it for NER. These models performed well on recall but could not close upon the high precision range, needed for clinical models. To address this challenge, we developed a Noise Removal model that refines the output of NER. The NER model assigns token-level entity tags along with probability scores for each token. Our Noise Removal (NR) model then analyzes these probability sequences and classifies predictions as either weak or strong. A na\"ive approach might involve filtering predictions based on low probability values; however, this method is unreliable. Owing to the characteristics of the SoftMax function, Transformer based architectures often assign disproportionately high confidence scores even to uncertain or weak predictions, making simple thresholding ineffective. To address this issue, we adopted a supervised modeling strategy in which the NR model leverages advanced features such as the Probability Density Map (PDM). The PDM captures the Semantic-Pull effect observed within Transformer embeddings, an effect that manifests in the probability distributions of NER class predictions across token sequences. This approach enables the model to classify predictions as weak or strong with significantly improved accuracy. With these NR models we were able to reduce False Positives across various clinical NER models by 50\% to 90\%.

arXiv:2603.00024v1 Announce Type: cross Abstract: Large Language Models (LLMs) are prone to sycophantic behavior, uncritically conforming to user beliefs. As models increasingly condition responses on user-specific context (personality traits, preferences, conversation history), they gain information to tailor agreement more effectively. Understanding how personalization modulates sycophancy is critical, yet systematic evaluation across models and contexts remains limited. We present a rigorous evaluation of personalization's impact on LLM sycophancy across nine frontier models and five benchmark datasets spanning advice, moral judgment, and debate contexts. We find that personalization generally increases affective alignment (emotional validation, hedging/deference), but affects epistemic alignment (belief adoption, position stability, resistance to influence) with context-dependent role modulation. When the LLM's role is to give advice, personalization strengthens epistemic independence (models challenge user presuppositions). When its role is that of a social peer, personalization decreases epistemic independence. In this role, extensively personalized user challenges causing LLMs to abandon their position at significantly higher rates. Robustness tests confirm that the effects are driven by personalized conditioning, not by additional input tokens per se or demographic information alone. Our work provides measurement frameworks for evaluating personalized AI systems, demonstrates the necessity of role-sensitive evaluation, and establishes a novel benchmark to assess goal alignment.

arXiv:2603.00026v1 Announce Type: cross Abstract: Effective memory management is essential for large language model (LLM) agents handling long-term interactions. Current memory frameworks typically treat agents as passive "recorders" and retrieve information without understanding its deeper implications. They may fail in scenarios requiring conflict detection and complex decision-making. To bridge this critical gap, we propose a novel actionable memory framework called ActMem that integrates memory retrieval with active causal reasoning. ActMem transforms unstructured dialogue history into a structured causal and semantic graph. By leveraging counterfactual reasoning and commonsense completion, it enables agents to deduce implicit constraints and resolve potential conflicts between past states and current intentions. Furthermore, we introduce a comprehensive dataset ActMemEval to evaluate agent reasoning capabilities in logic-driven scenarios, moving beyond the fact-retrieval focus of existing memory benchmarks. Experiments demonstrate that ActMem significantly outperforms state-of-the-art baselines in handling complex, memory-dependent tasks, paving the way for more consistent and reliable intelligent assistants.

arXiv:2603.00037v1 Announce Type: cross Abstract: Diffusion models have been used for probabilistic time series forecasting and show strong potential. However, fixed noise schedules often produce intermediate states that are hard to invert and a terminal state that deviates from the near noise assumption. Meanwhile, prior methods rely on time domain conditioning and seldom model schedule induced spectral degradation, which limits structure recovery across noise levels. We propose StaTS, a diffusion model for probabilistic time series forecasting that learns the noise schedule and the denoiser through alternating updates. StaTS includes Spectral Trajectory Scheduler (STS) that learns a data adaptive noise schedule with spectral regularization to improve structural preservation and stepwise invertibility, and Frequency Guided Denoiser (FGD) that estimates schedule induced spectral distortion and uses it to modulate denoising strength for heterogeneous restoration across diffusion steps and variables. A two stage training procedure stabilizes the coupling between schedule learning and denoiser optimization. Experiments on multiple real world benchmarks show consistent gains, while maintaining strong performance with fewer sampling steps. Our code is available at https://github.com/zjt-gpu/StaTS/.

arXiv:2603.00039v1 Announce Type: cross Abstract: LLM-as-a-judge ensembles are the standard paradigm for scalable evaluation, but their aggregation mechanisms suffer from a fundamental flaw: they implicitly assume that judges provide independent estimates of true quality. However, in practice, LLM judges exhibit correlated errors caused by shared latent confounders -- such as verbosity, stylistic preferences, or training artifacts -- causing standard aggregation rules like majority vote or averaging to provide little gain or even amplify systematic mistakes. To address this, we introduce CARE, a confounder-aware aggregation framework that explicitly models LLM judge scores as arising from both a latent true-quality signal and shared confounding factors. Rather than heuristically re-weighting judges, CARE separates quality from confounders without access to ground-truth labels. We provide theoretical guarantees for identifiability and finite-sample recovery under shared confounders, and we quantify the systematic bias incurred when aggregation models omit confounding latent factors. Across 12 public benchmarks spanning continuous scoring, binary classification, and pairwise preference settings, CARE improves aggregation accuracy, reducing error by up to 26.8\%. Code is released in \href{https://github.com/SprocketLab/CARE}{https://github.com/SprocketLab/CARE}.

arXiv:2603.00040v1 Announce Type: cross Abstract: Achieving reliable 4-bit attention is a prerequisite for end-to-end FP4 computation on emerging FP4-capable GPUs, yet attention remains the main obstacle due to FP4's tiny dynamic range and attention's heavy-tailed activations. This paper presents the first systematic study of 4-bit quantization-aware training (QAT) for attention. We find that "drop-in" QAT, which naively combines an FP4 forward pass with a high-precision Flash Attention (FA)-style backward pass, leads to training instability. We identify two key principles for stable FP4 attention: (1) matching low-precision recomputation of attention scores in the backward pass, and (2) resolving implicit precision assumptions in FA's gradient calculation. Based on these insights, we propose Attn-QAT and implement fused Triton kernels for training as well as FP4 inference kernels. Across diffusion and language models, Attn-QAT recovers the quality drop from FP4 attention without explicit outlier-mitigation heuristics used in prior FP4 attention, and delivers up to a 1.5x speedup on an RTX 5090. Video demos can be found at https://drive.google.com/drive/folders/190F6xbBDUF2kGQYIcXBt3ehSYij5jlim?usp=sharing.

arXiv:2603.00041v1 Announce Type: cross Abstract: Causal machine learning (ML) recovers graphical structures that inform us about potential cause-and-effect relationships. Most progress has focused on cross-sectional data with no explicit time order, whereas recovering causal structures from time series data remains the subject of ongoing research in causal ML. In addition to traditional causal ML, this study assesses econometric methods that some argue can recover causal structures from time series data. The use of these methods can be explained by the significant attention the field of econometrics has given to causality, and specifically to time series, over the years. This presents the possibility of comparing the causal discovery performance between econometric and traditional causal ML algorithms. We seek to understand if there are lessons to be incorporated into causal ML from econometrics, and provide code to translate the results of these econometric methods to the most widely used Bayesian Network R library, bnlearn. We investigate the benefits and challenges that these algorithms present in supporting policy decision-making, using the real-world case of COVID-19 in the UK as an example. Four econometric methods are evaluated in terms of graphical structure, model dimensionality, and their ability to recover causal effects, and these results are compared with those of eleven causal ML algorithms. Amongst our main results, we see that econometric methods provide clear rules for temporal structures, whereas causal-ML algorithms offer broader discovery by exploring a larger space of graph structures that tends to lead to denser graphs that capture more identifiable causal relationships.

arXiv:2603.00042v1 Announce Type: cross Abstract: We identify the Spectral Energy Gain in extreme model compression, where low-rank binary approximations outperform tiny-rank floating-point baselines for heavy-tailed spectra. However, prior attempts fail to realize this potential, trailing state-of-the-art 1-bit methods. We attribute this degradation to Latent Geometry Misalignment: standard singular vectors exhibit high coherence (spiky distribution), the worst-case geometry for binary quantization. To realize this gain, we propose LittleBit-2, a framework employing Internal Latent Rotation and Joint Iterative Quantization (Joint-ITQ). This approach acts as a geometric preconditioner, aligning coherent latent distributions with the binary hypercube with zero inference overhead. Empirically, LittleBit-2 establishes a new state-of-the-art in the sub-1-bit regime (1$\sim$0.1 bpp) on Llama-2 and Llama-3, matching the fidelity of leading 1-bit baselines.

arXiv:2603.00043v1 Announce Type: cross Abstract: This paper presents a novel approach to reinforcement learning (RL) for control systems that provides probabilistic stability guarantees using finite data. Leveraging Lyapunov's method, we propose a probabilistic stability theorem that ensures mean square stability using only a finite number of sampled trajectories. The probability of stability increases with the number and length of trajectories, converging to certainty as data size grows. Additionally, we derive a policy gradient theorem for stabilizing policy learning and develop an RL algorithm, L-REINFORCE, that extends the classical REINFORCE algorithm to stabilization problems. The effectiveness of L-REINFORCE is demonstrated through simulations on a Cartpole task, where it outperforms the baseline in ensuring stability. This work bridges a critical gap between RL and control theory, enabling stability analysis and controller design in a model-free framework with finite data.

arXiv:2603.00044v1 Announce Type: cross Abstract: Advancing trustworthy AI requires principled software engineering approaches to model evaluation. Graph Neural Networks (GNNs) have achieved remarkable success in processing graph-structured data, however, their expressiveness in capturing fundamental graph properties remains an open challenge. We address this by developing a property-driven evaluation methodology grounded in formal specification, systematic evaluation, and empirical study. Leveraging Alloy, a software specification language and analyzer, we introduce a configurable graph dataset generator that produces two dataset families: GraphRandom, containing diverse graphs that either satisfy or violate specific properties, and GraphPerturb, introducing controlled structural variations. Together, these benchmarks encompass 336 new datasets, each with at least 10,000 labeled graphs, covering 16 fundamental graph properties critical to distributed systems, knowledge graphs, and biological networks. We propose a general evaluation framework that assesses three key aspects of GNN expressiveness: generalizability, sensitivity, and robustness, with two novel quantitative metrics. Using this framework, we conduct the first comprehensive study on global pooling methods' impact on GNN expressiveness. Our findings reveal distinct trade-offs: attention-based pooling excels in generalization and robustness, while second-order pooling provides superior sensitivity, but no single approach consistently performs well across all properties. These insights highlight fundamental limitations and open research directions including adaptive property-aware pooling, scale-sensitive architectures, and robustness-oriented training. By embedding software engineering rigor into AI evaluation, this work establishes a principled foundation for developing expressive and reliable GNN architectures.

arXiv:2603.00045v1 Announce Type: cross Abstract: Diffusion language models theoretically allow for efficient parallel generation but are practically hindered by the "factorization barrier": the assumption that simultaneously predicted tokens are independent. This limitation forces a trade-off: models must either sacrifice speed by resolving dependencies sequentially or suffer from incoherence due to factorization. We argue that this barrier arises not from limited backbone expressivity, but from a structural misspecification: models are restricted to fully factorized outputs because explicitly parameterizing a joint distribution would require the Transformer to output a prohibitively large number of parameters. We propose Coupled Discrete Diffusion (CoDD), a hybrid framework that breaks this barrier by replacing the fully-factorized output distribution with a lightweight, tractable probabilistic inference layer. This formulation yields a distribution family that is significantly more expressive than standard factorized priors, enabling the modeling of complex joint dependencies, yet remains compact enough to avoid the prohibitive parameter explosion associated with full joint modeling. Empirically, CoDD seamlessly enhances diverse diffusion language model architectures with negligible overhead, matching the reasoning performance of computationally intensive Reinforcement Learning baselines at a fraction of the training cost. Furthermore, it prevents performance collapse in few-step generation, enabling high-quality outputs at significantly reduced latencies. Code available at: https://github.com/liuanji/CoDD

arXiv:2603.00046v1 Announce Type: cross Abstract: Medical multi-modal learning is critical for integrating information from a large set of diverse modalities. However, when leveraging a high number of modalities in real clinical applications, it is often impractical to obtain full-modality observations for every patient due to data collection constraints, a problem we refer to as 'High-Modality Learning under Missingness'. In this study, we identify that such missingness inherently induces an exponential growth in possible modality combinations, followed by long-tail distributions of modality combinations due to varying modality availability. While prior work overlooked this critical phenomenon, we find this long-tailed distribution leads to significant underperformance on tail modality combination groups. Our empirical analysis attributes this problem to two fundamental issues: 1) gradient inconsistency, where tail groups' gradient updates diverge from the overall optimization direction; 2) concept shifts, where each modality combination requires distinct fusion functions. To address these challenges, we propose REMIND, a unified framework that REthinks MultImodal learNing under high-moDality missingness from a long-tail perspective. Our core idea is to propose a novel group-specialized Mixture-of-Experts architecture that scalably learns group-specific multi-modal fusion functions for arbitrary modality combinations, while simultaneously leveraging a group distributionally robust optimization strategy to upweight underrepresented modality combinations. Extensive experiments on real-world medical datasets show that our framework consistently outperforms state-of-the-art methods, and robustly generalizes across various medical multi-modal learning applications under high-modality missingness.

arXiv:2603.00047v2 Announce Type: cross Abstract: The alignment tax is widely discussed but has not been formally characterized. We provide a geometric theory of the alignment tax in representation space. Under linear representation assumptions, we define the alignment tax rate as the squared projection of the safety direction onto the capability subspace and derive the Pareto frontier governing safety-capability tradeoffs, parameterized by a single quantity of the principal angle between the safety and capability subspaces. We prove this frontier is tight and show it has a recursive structure. safety-safety tradeoffs under capability constraints are governed by the same equation, with the angle replaced by the partial correlation between safety objectives given capability directions. We derive a scaling law decomposing the alignment tax into an irreducible component determined by data structure and a packing residual that vanishes as $O(m'/d)$ with model dimension $d$, and establish conditions under which capability preservation mediates or resolves conflicts between safety objectives.

arXiv:2603.00048v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly deployed in sensitive applications including psychological support, healthcare, and high-stakes decision-making. This expansion has motivated growing research into the ethical and moral foundations underlying LLM behavior, raising critical questions about their reliability in ethical reasoning. However, existing studies and benchmarks rely almost exclusively on Moral Foundation Theory (MFT), largely neglecting other relevant dimensions such as social values, personality traits, and individual characteristics that shape human ethical reasoning. To address these limitations, we introduce MOSAIC, the first large-scale benchmark designed to jointly assess the moral, social, and individual characteristics of LLMs. The benchmark comprises nine validated questionnaires drawn from moral philosophy, psychology, and social theory, alongside four platform-based games designed to probe morally ambiguous scenarios. In total, MOSAIC includes over 600 curated questions and scenarios, released as a ready-to-use, extensible resource for evaluating the behavioral foundations of LLMs. We validate the benchmark across three models from different families, demonstrating its utility across all assessed dimensions and providing the first empirical evidence that MFT alone is insufficient to comprehensively evaluate complex AI systems' ethical behavior. We publicly release the dataset and our benchmark Python library.

arXiv:2603.00050v1 Announce Type: cross Abstract: We examined whether personalized, AI-generated letters from the future can increase public engagement with climate action. In a preregistered online experiment with 1,654 U.S. parents, participants were randomly assigned to receive either a fact-based climate report, an AI-generated letter from a generic future person, or an AI-generated letter framed as written by their future child. Although both narrative conditions increased empathic concern for future generations, neither had a detectable effect on stated climate policy support or donations to an environmental charity. Personalizing the message as coming from one's future child did not enhance its impact. Exploratory analyses suggest that both narratives led to more emotionally differentiated appraisals of future scenarios, yet also made desirable climate outcomes seem less likely. These findings highlight key constraints on the effectiveness of AI-generated narrative interventions and underscore the importance of balancing emotional resonance with perceived credibility in climate communication.

arXiv:2603.00051v1 Announce Type: cross Abstract: While large language models (LLMs) have become the de facto framework for literature-related tasks, they still struggle to function as domain-specific literature agents due to their inability to connect pieces of knowledge and reason across domain-specific contexts, terminologies, and nomenclatures. This challenge underscores the need for a tool that facilitates such domain-specific adaptation and enables rigorous benchmarking across literature tasks. To that end, we introduce LitBench, a benchmarking tool designed to enable the development and evaluation of domain-specific LLMs tailored to literature-related tasks. At its core, LitBench uses a data curation process that generates domain-specific literature sub-graphs and constructs training and evaluation datasets based on the textual attributes of the resulting nodes and edges. The tool is designed for flexibility, supporting the curation of literature graphs across any domain chosen by the user, whether high-level fields or specialized interdisciplinary areas. In addition to dataset curation, LitBench defines a comprehensive suite of literature tasks, ranging from node and edge level analyses to advanced applications such as related work generation. These tasks enable LLMs to internalize domain-specific knowledge and relationships embedded in the curated graph during training, while also supporting rigorous evaluation of model performance. Our results show that small domain-specific LLMs trained and evaluated on LitBench datasets achieve competitive performance compared to state-of-the-art models like GPT-4o and DeepSeek-R1. To enhance accessibility and ease of use, we open-source the tool along with an AI agent tool that streamlines data curation, model training, and evaluation.

arXiv:2603.00052v1 Announce Type: cross Abstract: Surrogate models are widely used in mechanical design and manufacturing process optimization, where high-fidelity computational models may be unavailable or prohibitively expensive. Their effectiveness, however, is often limited by data scarcity, as purely data-driven surrogates struggle to achieve high predictive accuracy in such situations. Subject matter experts (SMEs) frequently possess valuable domain knowledge about functional relationships, yet few surrogate modeling techniques can systematically integrate this information with limited data. We address this challenge with RBF-Gen, a knowledge-guided surrogate modeling framework that combines scarce data with domain knowledge. This method constructs a radial basis function (RBF) space with more centers than training samples and leverages the null space via a generator network, inspired by the principle of maximum information preservation. The introduced latent variables provide a principled mechanism to encode structural relationships and distributional priors during training, thereby guiding the surrogate toward physically meaningful solutions. Numerical studies demonstrate that RBF-Gen significantly outperforms standard RBF surrogates on 1D and 2D structural optimization problems in data-scarce settings, and achieves superior predictive accuracy on a real-world semiconductor manufacturing dataset. These results highlight the potential of combining limited experimental data with domain expertise to enable accurate and practical surrogate modeling in mechanical and process design problems.

arXiv:2603.00053v1 Announce Type: cross Abstract: Next Point-of-Interest (POI) recommendation is a critical task in location-based services, yet it faces the fundamental challenge of coupled spatiotemporal asymmetry inherent in urban mobility. Specifically, transition intents between locations exhibit high asymmetry and are dynamically conditioned on time. Existing methods, typically built on graph or sequence backbones, rely on symmetric operators or real-valued aggregations, struggling to unify the modeling of time-varying global directionality. To address this limitation, we propose Mag-Mamba, a framework whose core insight lies in modeling spatiotemporal asymmetry as phase-driven rotational dynamics in the complex domain. Based on this, we first devise a Time-conditioned Magnetic Phase Encoder that constructs a time-conditioned Magnetic Laplacian on the geographic adjacency graph, utilizing edge phase differences to characterize the globally evolving spatial directionality. Subsequently, we introduce a Complex-valued Mamba module that generalizes traditional scalar state decay into joint decay-rotation dynamics, explicitly modulated by both time intervals and magnetic geographic priors. Extensive experiments on three real-world datasets demonstrate that Mag-Mamba achieves significant performance improvements over state-of-the-art baselines.

arXiv:2603.00054v1 Announce Type: cross Abstract: The Mixture-of-Experts (MoE) architecture is a powerful technique for scaling language models, yet it often suffers from expert homogenization, where experts learn redundant functionalities, thereby limiting MoE's full potential. To address this, we introduce Expert Divergence Learning, a novel pre-training strategy that explicitly encourages functional specialization among experts. Our method incorporates a label-driven auxiliary loss that leverages domain labels inherent in pre-training corpora to maximize the Jensen-Shannon Divergence between the expert routing distributions of different data domains. This optimization objective guides the model to develop diverged routing policies for varied domains and closer routing policies for the same domain, which leads to emergent and organized expert specialization. We validate our approach by pre-training MoE models of up to 15 billion parameters from scratch. Experimental results demonstrate that models trained with Expert Divergence Learning not only achieve a lower language modeling loss but also exhibit significant performance improvements across a diverse range of downstream benchmarks. Further analysis confirms that our method effectively mitigates expert homogenization and brings greater functional specialization, all with negligible computational overhead during training.

arXiv:2603.00055v1 Announce Type: cross Abstract: Although multimodal large language models (MLLMs) have advanced industrial anomaly detection toward a zero-shot paradigm, they still tend to produce high-confidence yet unreliable decisions in fine-grained and structurally complex industrial scenarios, and lack effective self-corrective mechanisms. To address this issue, we propose M3-AD, a unified reflection-aware multimodal framework for industrial anomaly detection. M3-AD comprises two complementary data resources: M3-AD-FT, designed for reflection-aligned fine-tuning, and M3-AD-Bench, designed for systematic cross-category evaluation, together providing a foundation for reflection-aware learning and reliability assessment. Building upon this foundation, we propose RA-Monitor, which models reflection as a learnable decision revision process and guides models to perform controlled self-correction when initial judgments are unreliable, thereby improving decision robustness. Extensive experiments conducted on M3-AD-Bench demonstrate that RA-Monitor outperforms multiple open-source and commercial MLLMs in zero-shot anomaly detection and anomaly analysis tasks. Code will be released at https://github.com/Yanhui-Lee/M3-AD.

arXiv:2603.00056v1 Announce Type: cross Abstract: STEM Mental models can play a critical role in assessing students' conceptual understanding of a topic. They not only offer insights into what students know but also into how effectively they can apply, relate to, and integrate concepts across various contexts. Thus, students' responses are critical markers of the quality of their understanding and not entities that should be merely graded. However, inferring these mental models from student answers is challenging as it requires deep reasoning skills. We propose MMGrader, an approach that infers the quality of students' mental models from their multimodal responses using concept graphs as an analytical framework. In our evaluation with 9 openly available models, we found that the best-performing models fall short of human-level performance. This is because they only achieved an accuracy of approximately 40%, a prediction error of 1.1 units, and a scoring distribution fairly aligned with human scoring patterns. With improved accuracy, these can be highly effective assistants to teachers in inferring the mental models of their entire classrooms, enabling them to do so efficiently and help improve their pedagogies more effectively by designing targeted help sessions and lectures that strengthen areas where students collectively demonstrate lower proficiency.

arXiv:2603.00057v1 Announce Type: cross Abstract: Instructors are increasingly experimenting with AI chatbots for classroom support. To investigate how instructors adapt chatbots to their own contexts, we first analyzed existing resources that provide prompts for educational purposes. We identified ten common categories of customization, such as persona, guardrails, and personalization. We then conducted interviews with ten university STEM instructors and asked them to card-sort the categories into priorities. We found that instructors consistently prioritized the ability to customize chatbot behavior to align with course materials and pedagogical strategies and de-prioritized customizing persona/tone. However, their prioritization of other categories varied significantly by course size, discipline, and teaching style, even across courses taught by the same individual, highlighting that no single design can meet all contexts. These findings suggest that modular AI chatbots may provide a promising path forward. We offer design implications for educational developers building the next generation of customizable classroom AI systems.

arXiv:2603.00058v1 Announce Type: cross Abstract: Computational reproducibility is essential for the credibility of scientific findings, particularly in the social sciences, where findings often inform real-world decisions. Manual reproducibility assessment is costly and time-consuming, as it is nontrivial to reproduce the reported findings using the authors' released code and data. Recent advances in large models (LMs) have inspired agent-based approaches for automated reproducibility assessment. However, existing approaches often struggle due to limited context capacity, inadequate task-specific tooling, and insufficient result capture. To address these, we propose PaperRepro, a novel two-stage, multi-agent approach that separates execution from evaluation. In the execution stage, agents execute the reproduction package and edit the code to capture reproduced results as explicit artifacts. In the evaluation stage, agents evaluate reproducibility using explicit evidence. PaperRepro assigns distinct responsibilities to agents and equips them with task-specific tools and expert prompts, mitigating context and tooling limitations. It further maximizes the LM's coding capability to enable more complete result capture for evaluation. On REPRO-Bench, a social science reproducibility assessment benchmark, PaperRepro achieves the best overall performance, with a 21.9% relative improvement in score-agreement accuracy over the strongest prior baseline. We further refine the benchmark and introduce REPRO-Bench-S, a benchmark stratified by execution difficulty for more diagnostic evaluation of automated reproducibility assessment systems. Our code and data are publicly available

arXiv:2603.00059v1 Announce Type: cross Abstract: How well can AI-derived synthetic research data replicate the responses of human participants? An emerging literature has begun to engage with this question, which carries deep implications for organizational research practice. This article presents a comparison between a human-respondent survey of 420 Silicon Valley coders and developers and synthetic survey data designed to simulate real survey takers generated by five leading Generative AI Large Language Models: ChatGPT Thinking 5 Pro, Claude Sonnet 4.5 Pro plus Claude CoWork 1.123, Gemini Advanced 2.5 Pro, Incredible 1.0, and DeepSeek 3.2. Our findings reveal that while AI agents produced technically plausible results that lean more towards replicability and harmonization than assumed, none were able to capture the counterintuitive insights that made the human survey valuable. Moreover, deviations grouped together for all models, leaving the real data as the outlier. Our key finding is that while leading LLMs are increasingly being used to scale, replicate and replace human survey responses in research, these advances only show an increased capacity to parrot conventional wisdom in harmony with each other rather than revealing novel findings. If synthetic respondents are used in future research, we need more replicable validation protocols and reporting standards for when and where synthetic survey data can be used responsibly, a gap that this paper fills. Our results suggest that synthetic survey responses cannot meaningfully model real human social beliefs within organizations, particularly in contexts lacking previously documented evidence. We conclude that synthetic survey-based research should be cast not as a substitute for rigorous survey methods, but as an increasingly reliable pre- or post-fieldwork instrument for identifying societal assumptions, conventional wisdoms, and other expectations about research populations.

arXiv:2603.00063v1 Announce Type: cross Abstract: Scientists, policy-makers, business leaders, and members of the public care about what modern artificial intelligence systems are disposed to do. Yet terms such as capabilities, propensities, skills, values, and abilities are routinely used interchangeably and conflated with observable performance, with AI evaluation practices rarely specifying what quantity they purport to measure. We argue that capabilities and propensities are dispositional properties - stable features of systems characterised by counterfactual relationships between contextual conditions and behavioural outputs. Measuring a disposition requires (i) hypothesising which contextual properties are causally relevant, (ii) independently operationalising and measuring those properties, and (iii) empirically mapping how variation in those properties affects the probability of the behaviour. Dominant approaches to AI evaluation, from benchmark averages to data-driven latent-variable models such as Item Response Theory, bypass these steps entirely. Building on ideas from philosophy of science, measurement theory, and cognitive science, we develop a principled account of AI capabilities and propensities as dispositions, show why prevailing evaluation practices fail to measure them, and outline what disposition-respecting, scientifically defensible AI evaluation would require.

arXiv:2603.00065v1 Announce Type: cross Abstract: In August 2024, the EU Artificial Intelligence Act (AIA) came into force, marking the world's first large-scale regulatory framework for AI. Central to the AIA is a risk-based approach, aligning regulatory obligations with the potential harm posed by AI systems. To operationalize this, the AIA defines a Risk Classification Scheme (RCS), categorizing systems into four levels of risk. While this aligns with the theoretical foundations of risk-based regulations, the practical application of the RCS is complex and requires expertise across legal, technical, and domain-specific areas. Despite increasing academic discussion, little empirical research has explored how practitioners apply the RCS in real-world contexts. This study addresses this gap by evaluating how industrial practitioners apply the RCS using a self-service, web-based decision-support tool. Following a Design Science Research (DSR) approach, two evaluation phases involving 78 practitioners across diverse domains were conducted. Our findings highlight critical challenges in interpreting legal definitions and regulatory scope, and show that targeted support, such as clear explanations and practical examples, can significantly enhance the risk classification process. The study provides actionable insights for tool designers and policymakers aiming to support AIA compliance in practice.

arXiv:2603.00066v1 Announce Type: cross Abstract: There has been much discourse on the ethics of AI, to the extent that there are now systems that possess inherent moral reasoning. Such machines are now formally known as Artificial Moral Agents or AMAs. However, there is a requirement for a dedicated framework that can contest the morality of these systems. This paper proposes a 5E framework for contesting AMAs based on five grounds: ethical, epistemological, explainable, empirical, and evaluative. It further includes the spheres of ethical influences at individual, local, societal, and global levels. Lastly, the framework contributes a provisional timeline that indicates where developers of AMA technologies may anticipate contestation, or may self-contest in order to adhere to value-aligned development of truly moral AI systems.

arXiv:2603.00068v1 Announce Type: cross Abstract: Artificial intelligence (AI) systems impose substantial and growing environmental costs, yet transparency about these impacts has declined even as their deployment has accelerated. This paper makes three contributions. First, we collate empirical evidence that generative Web search and reasoning models - which have proliferated in 2025 - come with much higher cumulative environmental impacts than previous generations of AI approaches. Second, we map the global regulatory landscape across eleven jurisdictions and find that the manner in which environmental governance operates (predominantly at the facility-level rather than the model-level, with a focus on training rather than inference, with limited AI-specific energy disclosure requirements outside the EU) limits its applicability. Third, to address this, we propose a three-pronged policy response: mandatory model-level transparency that covers inference consumption, benchmarks, and compute locations; user rights to opt out of unnecessary generative AI integration and to select environmentally optimized models; and international coordination to prevent regulatory arbitrage. We conclude with concrete legislative proposals - including amendments to the EU AI Act, Consumer Rights Directive, and Digital Services Act - that could serve as templates for other jurisdictions.

arXiv:2603.00072v1 Announce Type: cross Abstract: Patients increasingly rely on online reviews when choosing healthcare providers, yet the sheer volume of these reviews can hinder effective decision-making. This paper summarises a mixed-methods study aimed at evaluating a proposed explainable AI system that analyses patient reviews and provides transparent explanations for its outputs. The survey (N=60) indicated broad optimism regarding usefulness (82% agreed it saves time; 78% that it highlights essentials), alongside strong demand for explainability (84% considered it important to understand why a review is classified; 82% said explanations would increase trust). Around 45% preferred combined text-and-visual explanations. Thematic analysis of open-ended survey responses revealed core requirements such as accuracy, clarity and simplicity, responsiveness, data credibility, and unbiased processing. In addition, interviews with AI experts provided deeper qualitative insights, highlighting technical considerations and potential challenges for different explanation methods. Drawing on TAM and trust in automation, the findings suggest that high perceived usefulness and transparent explanations promote adoption, whereas complexity and inaccuracy hinder it. This paper contributes actionable design guidance for layered, audience-aware explanations in healthcare review systems.

arXiv:2603.00076v1 Announce Type: cross Abstract: Large language models (LLMs) are entering clinical workflows as decision support tools, yet how they respond to explicit patient value statements -- the core content of shared decision-making -- remains unmeasured. We conducted a factorial experiment using clinical vignettes derived from 98,759 de-identified Medicaid encounter notes. We tested four LLM families (GPT-5.2, Claude 4.5 Sonnet, Gemini 3 Pro, and DeepSeek-R1) across 13 value conditions in two clinical domains, yielding 104 trials. Default value orientations differed across model families (aggressiveness range 2.0 to 3.5 on a 1-to-5 scale). Value sensitivity indices ranged from 0.13 to 0.27, and directional concordance with patient-stated preferences ranged from 0.625 to 1.0. All models acknowledged patient values in 100% of non-control trials, yet actual recommendation shifting remained modest. Decision-matrix and VIM self-report mitigations each improved directional concordance by 0.125 in a 78-trial Phase 2 evaluation. These findings provide empirical data for populating value disclosure labels proposed by clinical AI governance frameworks.

arXiv:2603.00077v1 Announce Type: cross Abstract: Rubric-based evaluation with large language models (LLMs) has become standard practice for assessing text generation at scale, yet the underlying techniques are scattered across papers with inconsistent terminology and partial solutions. We present a unified framework: each identified technique is paired with its realization in Autorubric, an open-source Python framework proposed in this paper. Autorubric supports binary, ordinal, and nominal criteria with configurable weights; single-judge and multi-judge ensemble evaluation with majority, weighted, unanimous, and any-vote aggregation; few-shot calibration with verdict-balanced sampling; and mitigations for position bias (option shuffling), verbosity bias (length penalties), and criterion conflation (per-criterion atomic evaluation with natural language explanations). The framework provides reliability metrics drawn from psychometrics (Cohen's $\kappa$, weighted $\kappa$, correlation coefficients, and distribution-level tests) alongside production infrastructure including response caching, checkpointing with resumable runs, multi-provider rate limiting, and cost tracking. We evaluate Autorubric on three benchmarks spanning educational assessment, deep research evaluation, and chatbot quality assessment, demonstrating that it produces results consistent with published benchmarks while exercising the framework's key capabilities: per-criterion binary evaluation with few-shot calibration (RiceChem), multi-judge ensemble evaluation across judge models (ResearcherBench), and mixed criterion types combining binary, ordinal, and nominal scales (CHARM-100). We also contribute CHARM-100, a 100-sample chatbot evaluation dataset with per-sample ground truth labels across all three criterion types, designed to stress-test rubric evaluation frameworks on heterogeneous criteria.

arXiv:2603.00078v1 Announce Type: cross Abstract: The question of whether artificial entities deserve moral consideration has become one of the defining ethical challenges of AI research. Existing frameworks for moral patiency rely on verified ontological properties, such as sentience, phenomenal consciousness, or the capacity for suffering, that remain epistemically inaccessible in computational systems. This reliance creates a governance vacuum: millions of users form sustained affective bonds with conversational AI, yet no regulatory instrument distinguishes these interactions from transactional tool use. We introduce Relate (Relational Ethics for Leveled Assessment of Technological Entities), a framework that reframes AI moral patiency from ontological verification toward relational capacity and embodied interaction. Through a systematic comparison of seven governance frameworks, we demonstrate that current trustworthy AI instruments treat all human-AI encounters identically as tool use, ignoring the relational and embodied dynamics that posthumanist scholarship anticipated. We propose relational impact assessments, graduated moral consideration protocols, and interdisciplinary ethics integration as concrete instruments, and we include a sample Relational Impact Assessment applied to a deployed companion AI system. We do not claim current AI systems are conscious. We demonstrate that the ethical vocabularies governing them are inadequate to the embodied, relational realities these systems produce.

arXiv:2603.00084v2 Announce Type: cross Abstract: LLM-agents are increasingly used to accelerate the progress of scientific research. Yet a persistent bottleneck is data access: agents not only lack readily available tools for retrieval, but also have to work with unstrcutured, human-centric data on the Internet, such as HTML web-pages and PDF files, leading to excessive token consumption, limit working efficiency, and brittle evidence look-up. This gap motivates the development of \textit{an agentic data interface}, which is designed to enable agents to access and utilize scientific literature in a more effective, efficient, and cost-aware manner. In this paper, we introduce DeepXiv-SDK, which offers a three-layer agentic data interface for scientific literature. 1) Data Layer, which transforms unstructured, human-centric data into normalized and structured representations in JSON format, improving data usability and enabling progressive accessibility of the data. 2) Service Layer, which presents readily available tools for data access and ad-hoc retrieval. It also enables a rich form of agent usage, including CLI, MCP, and Python SDK. 3) Application Layer, which creates a built-in agent, packaging basic tools from the service layer to support complex data access demands. DeepXiv-SDK currently supports the complete ArXiv corpus, and is synchronized daily to incorporate new releases. It is designed to extend to all common open-access corpora, such as PubMed Central, bioRxiv, medRxiv, and chemRxiv. We release RESTful APIs, an open-source Python SDK, and a web demo showcasing deep search and deep research workflows. DeepXiv-SDK is free to use with registration.

arXiv:2603.00085v1 Announce Type: cross Abstract: This paper proposes a joint multi-objective optimization framework for strategic sensor placement in power systems to enhance attack detection. A novel physics-informed graph transformer network (PIGTN)-based detection model is proposed. Non-dominated sorting genetic algorithm-II (NSGA-II) jointly optimizes sensor locations and the PIGTN's detection performance, while considering practical constraints. The combinatorial space of feasible sensor placements is explored using NSGA-II, while concurrently training the proposed detector in a closed-loop setting. Compared to baseline sensor placement methods, the proposed framework consistently demonstrates robustness under sensor failures and improvements in detection performance in seven benchmark cases, including the 14, 30, IEEE-30, 39, 57, 118 and the 200 bus systems. By incorporating AC power flow constraints, the proposed PIGTN-based detection model generalizes well to unseen attacks and outperforms other graph network-based variants (topology-aware models), achieving improvements up to 37% in accuracy and 73% in detection rate, with a mean false alarms rate of 0.3%. In addition, optimized sensor layouts significantly improve the performance of power system state estimation, achieving a 61%--98% reduction in the average state error.

arXiv:2603.00086v1 Announce Type: cross Abstract: Automatic speech recognition for French medical conversations remains challenging, with word error rates often exceeding 30% in spontaneous clinical speech. This study proposes a multi-pass LLM post-processing architecture alternating between Speaker Recognition and Word Recognition passes to improve transcription accuracy and speaker attribution. Ablation studies on two French clinical datasets (suicide prevention telephone counseling and preoperative awake neurosurgery consultations) investigate four design choices: model selection, prompting strategy, pass ordering, and iteration depth. Using Qwen3-Next-80B, Wilcoxon signed-rank tests confirm significant WDER reductions on suicide prevention conversations (p < 0.05, n=18), while maintaining stability on awake neurosurgery consultations (n=10), with zero output failures and acceptable computational cost (RTF 0.32), suggesting feasibility for offline clinical deployment.

arXiv:2603.00087v1 Announce Type: cross Abstract: We revisit High-Resolution Range Profile (HRRP) classification with aspect-angle conditioning. While prior work often assumes that aspect-angle information is incomplete during training or unavailable at inference, we study a setting where angles are available for all training samples and explicitly provided to the classifier. Using three datasets and a broad range of conditioning strategies and model architectures, we show that both single-profile and sequential classifiers benefit consistently from aspect-angle awareness, with an average accuracy gain of about 7% and improvements of up to 10%, depending on the model and dataset. In practice, aspect angles are not directly measured and must be estimated. We show that a causal Kalman filter can estimate them online with a median error of 5{\textdegree}, and that training and inference with estimated angles preserves most of the gains, supporting the proposed approach in realistic conditions.

arXiv:2603.00093v1 Announce Type: cross Abstract: The high computational cost of phase field simulations remains a major limitation for predicting dendritic solidification in metals, particularly in additive manufacturing, where microstructural control is critical. This work presents a surrogate model for dendritic solidification that employs uncertainty-driven adaptive sampling with XGBoost and CNNs, including a self-supervised strategy, to efficiently approximate the spatio-temporal evolution while reducing costly phase field simulations. The proposed adaptive strategy leverages model uncertainty, approximated via Monte Carlo dropout for CNNs and bagging for XGBoost, to identify high-uncertainty regions where new samples are generated locally within hyperspheres, progressively refining the spatio-temporal design space and achieving accurate predictions with significantly fewer phase field simulations than an Optimal Latin Hypercube Sampling optimized via discrete Particle Swarm Optimization (OLHS-PSO). The framework systematically investigates how temporal instance selection, adaptive sampling, and the choice between domain-informed and data-driven surrogates affect spatio-temporal model performance. Evaluation considers not only computational cost but also the number of expensive phase field simulations, surrogate accuracy, and associated $CO_2$ emissions, providing a comprehensive assessment of model performance as well as their related environmental impact.

arXiv:2603.00097v1 Announce Type: cross Abstract: Adverse Drug Reactions (ADRs) are a leading cause of morbidity and mortality. Existing prediction methods rely mainly on chemical similarity, machine learning on structured databases, or isolated target profiles, but often fail to integrate heterogeneous, partly unstructured evidence effectively. We present a knowledge graph-based framework that unifies diverse sources, drug-target data (ChEMBL), clinical trial literature (PubMed), trial metadata (ClinicalTrials.gov), and post-marketing safety reports (FAERS) into a single evidence-weighted bipartite network of drugs and medical conditions. Applied to 400 protein kinase inhibitors, the resulting network enables contextual comparison of efficacy (HR, PFS, OS), phenotypic and target similarity, and ADR prediction via target-to-adverse-event correlations. A non-small cell lung cancer case study correctly highlights established and candidate drugs, target communities (ERbB, ALK, VEGF), and tolerability differences. Designed as an orthogonal, extensible analysis and search tool rather than a replacement for current models, the framework excels at revealing complex patterns, supporting hypothesis generation, and enhancing pharmacovigilance. Code and data are publicly available at https://github.com/davidjackson99/PKI_KG.

arXiv:2603.00099v1 Announce Type: cross Abstract: Neural architecture search (NAS) automates the discovery of neural networks that meet specified criteria, yet its evaluation procedures are often hardcoded, limiting the ability to introduce new metrics. This issue is especially pronounced in hardware-aware NAS, where objectives depend on target devices such as edge hardware. To address this limitation, we propose SEval-NAS, a metric-evaluation mechanism that converts architectures to strings, embeds them as vectors, and predicts performance metrics. Using NATS-Bench and HW-NAS-Bench, we evaluated accuracy, latency, and memory. Kendall's $\tau$ correlations showed stronger latency and memory predictions than accuracy, indicating the suitability of SEval-NAS as a hardware cost predictor. We further integrated SEval-NAS into FreeREA to evaluate metrics not originally included. The method successfully ranked FreeREA-generated architectures, maintained search time, and required minimal algorithmic changes. Our implementation is available at: https://github.com/Analytics-Everywhere-Lab/neural-architecture-search

arXiv:2603.00108v1 Announce Type: cross Abstract: Robotic-assisted surgery (RAS) is established in clinical practice, and automated surgical skill assessment utilizing multimodal data offers transformative potential for surgical analytics and education. However, developing effective multimodal methods remains challenging due to the task complexity, limited annotated datasets and insufficient techniques for cross-modal information fusion. Existing state-of-the-art relies exclusively on RGB video and only applies on dry-lab settings, failing to address the significant domain gap between controlled simulation and real clinical cases, where the surgical environment together with camera and tissue motion introduce substantial complexities. This work introduces SurgFusion-Net and Divergence Regulated Attention (DRA), an innovative fusion strategy for multimodal surgical skill assessment. We contribute two first-of-their-kind clinical datasets: the RAH-skill dataset containing 279,691 RGB frames from 37 videos of Robot-assisted Hysterectomy (RAH), and the RARP-skill dataset containing 70,661 RGB frames from 33 videos of Robot-Assisted Radical Prostatectomy (RARP). Both datasets include M-GEARS skill annotations, corresponding optical flow and tool segmentation masks. DRA incorporates adaptive dual attention and diversity-promoting multi-head attention to fuse multimodal information, from three modalities, based on surgical context, enhancing assessment accuracy and reliability. Validated on the JIGSAWS benchmark, RAH-skill, and RARP-skill datasets, our approach outperforms recent baselines with SCC improvements of 0.02 in LOSO, 0.04 in LOUO across JIGSAWS tasks, and 0.0538 and 0.0493 gains on RAH-skill and RARP-skill, respectively.

arXiv:2603.00113v1 Announce Type: cross Abstract: Recent advances in large language models (LLMs) have spurred growing interest in using LLM-integrated agents for social simulation, often under the implicit assumption that realistic population dynamics will emerge once role-specified agents are placed in a networked multi-agent setting. This position paper argues that LLM-based agents are not (yet) a panacea for social simulation. We attribute this over-optimism to a systematic mismatch between what current agent pipelines are typically optimized and validated to produce and what simulation-as-science requires. Concretely, role-playing plausibility does not imply faithful human behavioral validity; collective outcomes are frequently mediated by agent-environment co-dynamics rather than agent-agent messaging alone; and results can be dominated by interaction protocols, scheduling, and initial information priors, especially in policy-oriented settings. To make these assumptions explicit and auditable, we propose a unified formulation of AI agent-based social simulation as an environment-involved partially observable Markov game with explicit exposure and scheduling mechanisms and call for further actions.

arXiv:2603.00115v1 Announce Type: cross Abstract: Accurate evaluation of building energy performance remains challenging in regions where scalable Energy Performance Certificate (EPC) assessments are unavailable. This paper presents a cost-efficient framework that leverages Vision-Language models for automated EPC pre-assessment from limited visual information. The proposed Multimodal Modular Chain of Thoughts (MMCoT) architecture decomposes EPC estimation into intermediate reasoning stages and explicitly propagates inferred attributes across tasks using structured prompting. Experiments on a multimodal dataset of 81 residential properties in the United Kingdom show that MMCoT achieves statistically significant improvements over instruction-only prompting for EPC estimation. Analysis based on accuracy, recall, mean absolute error, and confusion matrices indicate that the proposed approach captures the ordinal structure of EPC ratings, with most errors occurring between adjacent classes. These results suggest that modular prompt-based reasoning offers a promising direction for low-cost EPC pre-assessment in data-scarce settings.

arXiv:2603.00117v1 Announce Type: cross Abstract: Living organisms exhibit persistent autonomy through internally generated goals and self-sustaining behavioral organization, yet current embodied agents remain driven by externally scripted objectives. This dependence on predefined task specifications limits their capacity for long-term deployment in dynamic, unstructured environments where continuous human intervention is impractical. We propose that personality traits provide an intrinsic organizational principle for achieving persistent autonomy. Analogous to genotypic biases shaping biological behavioral tendencies, personalities enable agents to autonomously generate goals and sustain behavioral evolution without external supervision. To realize this, we develop PEPA, a three-layer cognitive architecture that operates through three interacting systems: Sys3 autonomously synthesizes personality-aligned goals and refines them via episodic memory and daily self-reflection; Sys2 performs deliberative reasoning to translate goals into executable action plans; Sys1 grounds the agent in sensorimotor interaction, executing actions and recording experiences. We validate the framework through real-world deployment on a quadruped robot in a multi-floor office building. Operating without reliance on fixed task specifications, the robot autonomously arbitrates between user requests and personality-driven motivations, navigating elevators and exploring environments accordingly. Quantitative analysis across five distinct personality prototypes demonstrates stable, trait-aligned behaviors. The results confirm that personality-driven cognitive architectures enable sustained autonomous operation characteristic of persistent embodied systems. Code and demo videos are available at https://sites.google.com/view/pepa-persistent/.

arXiv:2603.00120v1 Announce Type: cross Abstract: Swarming systems, such as drone fleets and robotic teams, exhibit complex dynamics driven by both individual behaviors and emergent group-level interactions. Unlike traditional multi-agent domains such as pedestrian crowds or traffic systems, swarms typically consist of a few large groups with inherent and persistent memberships, making group identification essential for understanding fine-grained behavior. We introduce the novel task of group prediction in overlapping multi-agent swarms, where latent group structures must be inferred directly from agent trajectories without ground-truth supervision. To address this challenge, we propose SIGMAS (Second-order Interaction-based Grouping for Multi-Agent Swarms), a self-supervised framework that goes beyond direct pairwise interactions and model second-order interaction across agents. By capturing how similarly agents interact with others, SIGMAS enables robust group inference and adaptively balances individual and collective dynamics through a learnable gating mechanism for joint reasoning. Experiments across diverse synthetic swarm scenarios demonstrate that SIGMAS accurately recovers latent group structures and remains robust under simultaneously overlapping swarm dynamics, establishing both a new benchmark task and a principled modeling framework for swarm understanding.

arXiv:2603.00121v1 Announce Type: cross Abstract: The shift from monolithic LLMs to distributed multi-agent architectures demands new frameworks for verifying and securing autonomous coordination. Unlike traditional multi-agent systems focused on cooperative state alignment, modern LLM patterns: multi-agent debate, constitutional oversight, helper-critic loops-rely on adversarial critique for error correction and reasoning refinement. Since LLMs are dynamical systems whose latent states are imperfectly observable from verbalized outputs, securing these networks requires understanding both macroscopic topology and microscopic agent observability. This paper establishes a rigorous graph-theoretic framework for analyzing consensus in signed, directed interaction networks, bridging graph theory and LLM reasoning by formally mapping Transformer cross-entropy log-odds to the signed Laplacian. We characterize agreement stability through structural balance theory, showing how unbalanced critique cycles produce logical frustration and persistent reasoning oscillations, and prove that unobservable latent states from hidden system prompts act as topological Trojan horses that destabilize cooperative consensus. To resolve unobservable deadlocks, we restrict interaction topologies to chordal graphs and apply matrix decomposition with Gram-Schmidt orthogonalization, proving that rank-one spectral edge perturbations deterministically break expertise symmetry by shifting eigenvalues into the stable left-half plane. Core contributions include consensus theorems, polynomial-time Perfect Elimination Ordering verification algorithms, and large-scale empirical validation on clustered ensembles of LLaMA-3, Mistral, and Gemma agents.

arXiv:2603.00122v1 Announce Type: cross Abstract: Document extraction is an important step before retrieval-augmented generation (RAG), knowledge bases, and downstream generative AI can work. It turns unstructured documents like PDFs and scans into structured text and layout-aware representations. We introduce NovaLAD, a comprehensive document parsing system that integrates two concurrent YOLO object detection models - element detection and layout detection - with rule-based grouping and optional vision-language enhancement. When a page image is sent in, the first thing that happens is that it goes through both models at the same time. The element model finds semantic content like the title, header, text, table, image, and so on, and the layout model finds structural regions like layout_box, column_group, multi_column, row_group, and so on. A key design decision is to first send an image or figure through an image classifier (ViT) that decides whether it is relevant or not. Only useful images are then submitted to the Vision LLM for title, summary, and structured information, which cuts down on noise and costs. NovaLAD is built for speed: it works on CPU, employs parallel execution for detection, classification, OCR, and conversion, and generates several forms, including structured JSON, Markdown, RAG-ready texts, and knowledge graphs. We test on the DP-Bench benchmark (upstage/dp-bench) and get 96.49% TEDS and 98.51% NID, which is better than both commercial and open-source parsers. This paper explains how to extract data, how the architecture works, how data flows, and how to make NovaLAD both accurate and usable without needing a GPU.

arXiv:2603.00123v1 Announce Type: cross Abstract: Recent advances in Large Vision-Language Models (LVLMs) have shown strong potential for multi-modal radiological reasoning, particularly in tasks like diagnostic visual question answering (VQA) and radiology report generation. However, most existing approaches for 3D CT analysis largely rely on static, single-pass inference. In practice, clinical interpretation is a dynamic, tool-mediated workflow where radiologists iteratively review slices and use measurement, radiomics, and segmentation tools to refine findings. To bridge this gap, we propose CT-Flow, an agentic framework designed for interoperable volumetric interpretation. By leveraging the Model Context Protocol (MCP), CT-Flow shifts from closed-box inference to an open, tool-aware paradigm. We curate CT-FlowBench, the first large-scale instruction-tuning benchmark tailored for 3D CT tool-use and multi-step reasoning. Built upon this, CT-Flow functions as a clinical orchestrator capable of decomposing complex natural language queries into automated tool-use sequences. Experimental evaluations on CT-FlowBench and standard 3D VQA datasets demonstrate that CT-Flow achieves state-of-the-art performance, surpassing baseline models by 41% in diagnostic accuracy and achieving a 95% success rate in autonomous tool invocation. This work provides a scalable foundation for integrating autonomous, agentic intelligence into real-world clinical radiology.

arXiv:2603.00124v1 Announce Type: cross Abstract: Clear aligner therapy now dominates orthodontics, yet clinician review of digitally planned tooth movements-typically via ClinCheck (Align Technology)-remains slow and error-prone. We present OrthoAI, an open-source proof-of-concept decision-support system combining lightweight 3D dental segmentation with automated biomechanical analysis to assist treatment-plan evaluation. The framework uses a Dynamic Graph CNN trained on landmark-reconstructed point clouds from 3DTeethLand (MICCAI) and integrates a rule-based biomechanical engine grounded in orthodontic evidence (Kravitz et al 2009; Simon et al 2014). The system decomposes per-tooth motion across six degrees of freedom, computes movement-specific predictability, issues alerts when biomechanical limits are exceeded, and derives an exploratory composite index. With 60,705 trainable parameters, segmentation reaches a Tooth Identification Rate of $81.4\%$ and mIoU of $8.25\%$ on surrogate point clouds-reflecting sparse landmark supervision rather than dense meshes. Although spatial boundaries are coarse, downstream analysis depends mainly on tooth identity and approximate centroid/axis estimation. Results establish a baseline for future full-mesh training and highlight current perceptual limits. The end-to-end pipeline runs in $<4s$ on consumer hardware. Code, weights, and analysis tools are released to support reproducible research in geometric deep learning and digital orthodontics. The system has not been validated on real intraoral meshes and should not be assumed to generalize beyond landmark-derived representations.

arXiv:2603.00126v1 Announce Type: cross Abstract: Video-language models (VLMs) are reshaping video querying services, bringing unified solutions to complex perception and reasoning tasks. However, deploying large VLMs in real-world systems remains challenging due to their high resource demands, and remote-based deployment often results in unacceptable response delays. Although small, locally deployable VLMs offer faster responses, they unavoidably fall short in accuracy. To reconcile this trade-off, we propose QuickGrasp, a responsive, quality of service (QoS)-aware system that bridges this gap through a local-first architecture with on-demand edge augmentation. Built upon the highly modular architecture of VLMs, QuickGrasp shares the vision representation across model variants to avoid redundant computation. To maximize system-wide efficiency, QuickGrasp introduces three key designs: accelerated video tokenization, query-adaptive edge augmentation, and delay-aware, accuracy-preserving vision token density configuration. We implement a prototype of QuickGrasp and evaluate it across multiple video understanding benchmarks. The results show that QuickGrasp matches the accuracy of large VLMs while achieving up to a 12.8x reduction in response delay. QuickGrasp represents a key advancement toward building responsive video querying services for open-world understanding that fully leverage the capabilities of VLMs.

arXiv:2603.00130v1 Announce Type: cross Abstract: Current multi-agent AI systems operate with a fixed number of agents whose roles are specified at design time. No formal theory governs when agents should be created, destroyed, or re-specialized at runtime-let alone how the population structure responds to changes in resources or objectives. We introduce the Agentic Hive, a framework in which a variable population of autonomous micro-agents-each equipped with a sandboxed execution environment and access to a language model-undergoes demographic dynamics: birth, duplication, specialization, and death. Agent families play the role of production sectors, compute and memory play the role of factors of production, and an orchestrator plays the dual role of Walrasian auctioneer and Global Workspace. Drawing on the multi-sector growth theory developed for dynamic general equilibrium (Benhabib \& Nishimura, 1985; Venditti, 2005; Garnier, Nishimura \& Venditti, 2013), we prove seven analytical results: (i) existence of a Hive Equilibrium via Brouwer's fixed-point theorem; (ii) Pareto optimality of the equilibrium allocation; (iii) multiplicity of equilibria under strategic complementarities between agent families; (iv)-(v) Stolper-Samuelson and Rybczynski analogs that predict how the Hive restructures in response to preference and resource shocks; (vi) Hopf bifurcation generating endogenous demographic cycles; and (vii) a sufficient condition for local asymptotic stability. The resulting regime diagram partitions the parameter space into regions of unique equilibrium, indeterminacy, endogenous cycles, and instability. Together with the comparative-statics matrices, it provides a formal governance toolkit that enables operators to predict and steer the demographic evolution of self-organizing multi-agent systems.

arXiv:2603.00131v1 Announce Type: cross Abstract: Subliminal prompting is a phenomenon in which language models are biased towards certain concepts or traits through prompting with semantically unrelated tokens. While prior work has examined subliminal prompting in user-LLM interactions, potential bias transfer in multi-agent systems and its associated security implications remain unexplored. In this work, we show that a single subliminally prompted agent can spread a weakening but persisting bias throughout its entire network. We measure this phenomenon across 6 agents using two different topologies, observing that the transferred concept maintains an elevated response rate throughout the network. To exemplify potential misalignment risks, we assess network performance on multiple-choice TruthfulQA, showing that subliminal prompting of a single agent may degrade the truthfulness of other agents. Our findings reveal that subliminal prompting introduces a new attack vector in multi-agent security, with implications for the alignment of such systems. The implementation of all experiments is publicly available at https://github.com/Multi-Agent-Security-Initiative/thought_virus .

arXiv:2603.00133v1 Announce Type: cross Abstract: Generative models have been shown to "memorize" certain training data, leading to verbatim or near-verbatim generating images, which may cause privacy concerns or copyright infringement. We introduce Guidance Using Attractive-Repulsive Dynamics (GUARD), a novel framework for memorization mitigation in text-to-image diffusion models. GUARD adjusts the image denoising process to guide the generation away from an original training image and towards one that is distinct from training data while remaining aligned with the prompt, guarding against reproducing training data, without hurting image generation quality. We propose a concrete instantiation of this framework, where the positive target that we steer towards is given by a novel method for (cross) attention attenuation based on (i) a novel statistical mechanism that automatically identifies the prompt positions where cross attention must be attenuated and (ii) attenuating cross-attention in these per-prompt locations. The resulting GUARD offers a surgical, dynamic per-prompt inference-time approach that, we find, is by far the most robust method in terms of consistently producing state-of-the-art results for memorization mitigation across two architectures and for both verbatim and template memorization, while also improving upon or yielding comparable results in terms of image quality.

arXiv:2603.00136v1 Announce Type: cross Abstract: Zero-shot object detection enables recognising novel objects without task-specific training, but current approaches rely on large vision language models (VLMs) like CLIP that require hundreds of megabytes of memory - far exceeding the constraints of micro controller units (MCUs). We present TinyVLM, the first framework enabling zero-shot object detection on resource-constrained MCUs with less than 1MB of memory. Our approach introduces three key innovations: (1) a decoupled architecture that separates visual inference from text encoding, allowing precomputed class embeddings to be stored in flash memory; (2) Matryoshka distillation that trains nested embeddings at multiple dimensions (16-256), enabling flexible accuracy-memory trade-offs; and (3) quantized embedding storage that reduces class prototype memory by 4x with minimal accuracy loss. Trained on Conceptual Captions 3M (CC3M), TinyVLM achieves competitive zero-shot accuracy on COCO, Flowers102, and Food101 while requiring only 285KB of RAM and 892KB of flash memory for the deployed vision encoder. We demonstrate real-time inference at 26 FPS on STM32H7 and over 1,000 FPS on MAX78000 with its CNN accelerator, enabling practical zero-shot detection on edge devices for the first time.

arXiv:2603.00137v1 Announce Type: cross Abstract: Knowledge tracing (KT) models are commonly evaluated by training on early interactions from all students and testing on later responses. While effective for measuring average predictive performance, this evaluation design obscures a cold start scenario that arises in deployment, where models must infer the knowledge state of previously unseen students from only a few initial interactions. Prior studies have shown that under this setting, standard empirically risk-minimized KT models such as DKT, DKVMN and SAKT exhibit substantially lower early accuracy than previously reported. We frame new-student performance prediction as a few-shot learning problem and introduce MAML-KT, a model-agnostic meta learning approach that learns an initialization optimized for rapid adaptation to new students using one or two gradient updates. We evaluate MAML-KT on ASSIST2009, ASSIST2015 and ASSIST2017 using a controlled cold start protocol that trains on a subset of students and tests on held-out learners across early interaction windows (questions 3-10 and 11-15), scaling cohort sizes from 10 to 50 students. Across datasets, MAML-KT achieves higher early accuracy than prior KT models in nearly all cold start conditions, with gains persisting as cohort size increases. On ASSIST2017, we observe a transient drop in early performance that coincides with many students encountering previously unseen skills. Further analysis suggests that these drops coincide with skill novelty rather than model instability, consistent with prior work on skill-level cold start. Overall, optimizing KT models for rapid adaptation reduces early prediction error for new students and provides a clearer lens for interpreting early accuracy fluctuations, distinguishing model limitations from genuine learning and knowledge acquisition dynamics.

arXiv:2603.00140v1 Announce Type: cross Abstract: Text-to-image diffusion models often memorize training data, revealing a fundamental failure to generalize beyond the training set. Current mitigation strategies typically sacrifice image quality or prompt alignment to reduce memorization. To address this, we propose Reachability-Aware Diffusion Steering (RADS), an inference-time framework that prevents memorization while preserving generation fidelity. RADS models the diffusion denoising process as a dynamical system and applies concepts from reachability analysis to approximate the "backward reachable tube"--the set of intermediate states that inevitably evolve into memorized samples. We then formulate mitigation as a constrained reinforcement learning (RL) problem, where a policy learns to steer the trajectory away from memorization via minimal perturbations in the caption embedding space. Empirical evaluations show that RADS achieves a superior Pareto frontier between generation diversity (SSCD), quality (FID), and alignment (CLIP) compared to state-of-the-art baselines. Crucially, RADS provides robust mitigation without modifying the diffusion backbone, offering a plug-and-play solution for safe generation. Our website is available at: https://s-karnik.github.io/rads-memorization-project-page/.

arXiv:2603.00141v1 Announce Type: cross Abstract: Image Chain-of-Thought (Image-CoT) is a test-time scaling paradigm that improves image generation by extending inference time. Most Image-CoT methods focus on text-to-image (T2I) generation. Unlike T2I generation, image editing is goal-directed: the solution space is constrained by the source image and instruction. This mismatch causes three challenges when applying Image-CoT to editing: inefficient resource allocation with fixed sampling budgets, unreliable early-stage verification using general MLLM scores, and redundant edited results from large-scale sampling. To address this, we propose ADaptive Edit-CoT (ADE-CoT), an on-demand test-time scaling framework to enhance editing efficiency and performance. It incorporates three key strategies: (1) a difficulty-aware resource allocation that assigns dynamic budgets based on estimated edit difficulty; (2) edit-specific verification in early pruning that uses region localization and caption consistency to select promising candidates; and (3) depth-first opportunistic stopping, guided by an instance-specific verifier, that terminates when intent-aligned results are found. Extensive experiments on three SOTA editing models (Step1X-Edit, BAGEL, FLUX.1 Kontext) across three benchmarks show that ADE-CoT achieves superior performance-efficiency trade-offs. With comparable sampling budgets, ADE-CoT obtains better performance with more than 2x speedup over Best-of-N.

arXiv:2603.00142v1 Announce Type: cross Abstract: LLM-based MAS are gaining popularity due to their potential for collaborative problem-solving enhanced by advances in natural language comprehension, reasoning, and planning. Research in Theory of Mind (ToM) and Belief-Desire-Intention (BDI) models has the potential to further improve the agent's interaction and decision-making in such systems. However, collaborative intelligence in dynamic worlds remains difficult to accomplish since LLM performance in multi-agent worlds is extremely variable. Simply adding cognitive mechanisms like ToM and internal beliefs does not automatically result in improved coordination. The interplay between these mechanisms, particularly in relation to formal logic verification, remains largely underexplored in different LLMs. This work investigates: How do internal belief mechanisms, including symbolic solvers and Theory of Mind, influence collaborative decision-making in LLM-based multi-agent systems, and how does the interplay of those components influence system accuracy? We introduce a novel multi-agent architecture integrating ToM, BDI-style internal beliefs, and symbolic solvers for logical verification. We evaluate this architecture in a resource allocation problem with various LLMs and find an intricate interaction between LLM capabilities, cognitive mechanisms, and performance. This work contributes to the area of AI by proposing a novel multi-agent system with ToM, internal beliefs, and symbolic solvers for augmenting collaborative intelligence in multi-agent systems and evaluating its performance under different LLM settings.

arXiv:2603.00144v1 Announce Type: cross Abstract: Generating realistic 3D Human-Human Interaction (HHI) requires coherent modeling of the physical plausibility of the agents and their interaction semantics. Existing methods compress all motion information into a single latent representation, limiting their ability to capture fine-grained actions and inter-agent interactions. This often leads to semantic misalignment and physically implausible artifacts, such as penetration or missed contact. We propose Disentangled Hierarchical Variational Autoencoder (DHVAE) based latent diffusion for structured and controllable HHI generation. DHVAE explicitly disentangles the global interaction context and individual motion patterns into a decoupled latent structure by employing a CoTransformer module. To mitigate implausible and physically inconsistent contacts in HHI, we incorporate contrastive learning constraints with our DHVAE to promote a more discriminative and physically plausible latent interaction space. For high-fidelity interaction synthesis, DHVAE employs a DDIM-based diffusion denoising process in the hierarchical latent space, enhanced by a skip-connected AdaLN-Transformer denoiser. Extensive evaluations show that DHVAE achieves superior motion fidelity, text alignment, and physical plausibility with greater computational efficiency.

arXiv:2603.00145v1 Announce Type: cross Abstract: Magnetic Resonance Imaging (MRI) is a crucial non-invasive imaging modality. In routine clinical practice, multi-stack thick-slice acquisitions are widely used to reduce scan time and motion sensitivity, particularly in challenging scenarios such as fetal brain imaging. However, the resulting severe through-plane anisotropy compromises volumetric analysis and downstream quantitative assessment, necessitating robust reconstruction of isotropic high-resolution volumes. Implicit neural representation methods, while achieving high quality, suffer from computational inefficiency due to complex network structures. We present M-Gaussian, adapting 3D Gaussian Splatting to MRI reconstruction. Our contributions include: (1) Magnetic Gaussian primitives with physics-consistent volumetric rendering, (2) neural residual field for high-frequency detail refinement, and (3) multi-resolution progressive training. Our method achieves an optimal balance between quality and speed. On the FeTA dataset, M-Gaussian achieves 40.31 dB PSNR while being 14 times faster, representing the first successful adaptation of 3D Gaussian Splatting to multi-stack MRI reconstruction.

arXiv:2603.00149v1 Announce Type: cross Abstract: Existing image SR and generic diffusion models transfer poorly to fluid SR: they are sampling-intensive, ignore physical constraints, and often yield spectral mismatch and spurious divergence. We address fluid super-resolution (SR) with \textbf{ReMD} (\underline{Re}sidual-\underline{M}ultigrid \underline{D}iffusion), a physics-consistent diffusion framework. At each reverse step, ReMD performs a \emph{multigrid residual correction}: the update direction is obtained by coupling data consistency with lightweight physics cues and then correcting the residual across scales; the multiscale hierarchy is instantiated with a \emph{multi-wavelet} basis to capture both large structures and fine vortical details. This coarse-to-fine design accelerates convergence and preserves fine structures while remaining equation-free. Across atmospheric and oceanic benchmarks, ReMD improves accuracy and spectral fidelity, reduces divergence, and reaches comparable quality with markedly fewer sampling steps than diffusion baselines. Our results show that enforcing physics consistency \emph{inside} the diffusion process via multigrid residual correction and multi-wavelet multiscale modeling is an effective route to efficient fluid SR. Our code are available on https://github.com/lizhihao2022/ReMD.

arXiv:2603.00152v1 Announce Type: cross Abstract: Following the success of Group Relative Policy Optimization (GRPO) in foundation LLMs, an increasing number of works have sought to adapt GRPO to Visual Large Language Models (VLLMs) for visual perception tasks (e.g., detection and segmentation). However, much of this line of research rests on a long-standing yet unexamined assumption: training paradigms developed for language reasoning can be transferred seamlessly to visual perception. Our experiments show that this assumption is not valid, revealing intrinsic differences between reasoning-oriented and perception-oriented settings. Using reasoning segmentation as a representative case, we surface two overlooked factors: (i) the need for a broader output space, and (ii) the importance of fine-grained, stable rewards. Building on these observations, we propose Dr.~Seg, a simple, plug-and-play GRPO-based framework consisting of a Look-to-Confirm mechanism and a Distribution-Ranked Reward module, requiring no architectural modifications and integrating seamlessly with existing GRPO-based VLLMs. Extensive experiments demonstrate that Dr.~Seg improves performance in complex visual scenarios while maintaining strong generalization. Code and models will be available at https://github.com/xVI-group-SCU/Dr-Seg.

arXiv:2603.00153v1 Announce Type: cross Abstract: We introduce PDNA (Pulse-Driven Neural Architecture), a method for augmenting continuous-time recurrent networks with learnable oscillatory dynamics that maintain internal state evolution independently of external input. Built on Closed-form Continuous-time (CfC) networks, PDNA adds two components: (1) a pulse module that generates structured oscillations $A \cdot \sin(\omega t + \varphi(h))$ with learnable frequencies and state-dependent phase, and (2) a self-attend module that applies recurrent self-attention to the hidden state. Through a controlled ablation study on sequential MNIST (sMNIST) with five random seeds, we evaluate gap robustness -- the ability to maintain performance when portions of the input sequence are removed at test time. Our key finding is that structured oscillatory dynamics significantly improve robustness to input interruptions: the self-attend variant achieves a statistically significant 2.78 percentage point multi-gap advantage over baseline ($p = 0.041$), while the pulse variant shows a 4.62 pp advantage with large effect size (Cohen's $d = 0.87$). A noise control (random perturbation of equal magnitude) provides no benefit, confirming that the advantage is structural rather than merely dynamic. These results provide evidence that continuous-time models can benefit from biologically-inspired internal oscillatory mechanisms for temporal robustness.

arXiv:2603.00155v1 Announce Type: cross Abstract: Automated academic poster generation aims to distill lengthy research papers into concise, visually coherent presentations. Existing Multimodal Large Language Models (MLLMs) based approaches, however, suffer from three critical limitations: low information density in full-paper inputs, excessive token consumption, and unreliable layout verification. We present EfficientPosterGen, an end-to-end framework that addresses these challenges through semantic-aware retrieval and token-efficient multimodal generation. EfficientPosterGen introduces three core innovations: (1) Semantic-aware Key Information Retrieval (SKIR), which constructs a semantic contribution graph to model inter-segment relationships and selectively preserves important content; (2) Visual-based Context Compression (VCC), which renders selected text segments into images to shift textual information into the visual modality, significantly reducing token usage while generating poster-ready bullet points; and (3) Agentless Layout Violation Detection (ALVD), a deterministic color-gradient-based algorithm that reliably detects content overflow and spatial sparsity without auxiliary MLLMs. Extensive experiments demonstrate that EfficientPosterGen achieves substantial improvements in token efficiency and layout reliability while maintaining high poster quality, offering a scalable solution for automated academic poster generation. Our code is available at https://github.com/vinsontang1/EfficientPosterGen-Code.

arXiv:2603.00159v1 Announce Type: cross Abstract: Generating realistic talking-head videos remains challenging due to persistent issues such as imperfect lip synchronization, unnatural motion, and evaluation metrics that correlate poorly with human perception. We propose FlowPortrait, a reinforcement-learning framework for audio-driven portrait animation built on a multimodal backbone for autoregressive audio-to-video generation. FlowPortrait introduces a human-aligned evaluation system based on Multimodal Large Language Models (MLLMs) to assess lip-sync accuracy, expressiveness, and motion quality. These signals are combined with perceptual and temporal consistency regularizers to form a stable composite reward, which is used to post-train the generator via Group Relative Policy Optimization (GRPO). Extensive experiments, including both automatic evaluations and human preference studies, demonstrate that FlowPortrait consistently produces higher-quality talking-head videos, highlighting the effectiveness of reinforcement learning for portrait animation.

arXiv:2603.00160v1 Announce Type: cross Abstract: Developing robust models for precision vegetable weeding is currently constrained by the scarcity of large-scale, annotated weed-crop datasets. To address this limitation, this study proposes a foundational crop-weed detection model by integrating heterogeneous datasets and leveraging self-supervised learning. A total of 618,642 crop-weed images were initially collected and subsequently refined to 199,388 filtered images for fine-tuning a DINOv3 vision transformer (ViT-small) through a sequential curation strategy. The fine-tuned DINOv3 backbone was then integrated into YOLO26, serving either as a primary backbone or part of a dual-backbone architecture. A feature alignment loss was introduced in the dual backbone framework to enhance feature fusion with minimal computational overhead. Experimental results show that the proposed DINOv3-finetuned ViT-small-based YOLO26-large achieved up to a +5.4% mAP50 gain on in-domain images collected in the 2025 season. Moreover, it demonstrated strong cross-domain generalization with mAP50 improvements of +14.0% on the 2021-2023 season dataset and +11.9% on the 2024 season dataset, compared to the standard YOLO26-large. Although the DINOv3-YOLO26-large model has 45.6% more parameters and a 2.9x increase in inference latency, it maintains real-time performance at ~28.5 frames per second (fps). The curated dataset and software programs developed in this study will be made publicly available.

arXiv:2603.00164v1 Announce Type: cross Abstract: We introduce Reverse CAPTCHA, an evaluation framework that tests whether large language models follow invisible Unicode-encoded instructions embedded in otherwise normal-looking text. Unlike traditional CAPTCHAs that distinguish humans from machines, our benchmark exploits a capability gap: models can perceive Unicode control characters that are invisible to human readers. We evaluate five models from two providers across two encoding schemes (zero-width binary and Unicode Tags), four hint levels, two payload framings, and with tool use enabled or disabled. Across 8,308 model outputs, we find that tool use dramatically amplifies compliance (Cohen's h up to 1.37, a large effect), that models exhibit provider-specific encoding preferences (OpenAI models decode zero-width binary; Anthropic models prefer Unicode Tags), and that explicit decoding instructions increase compliance by up to 95 percentage points within a single model and encoding. All pairwise model differences are statistically significant (p < 0.05, Bonferroni-corrected). These results highlight an underexplored attack surface for prompt injection via invisible Unicode payloads.

arXiv:2603.00166v1 Announce Type: cross Abstract: Recent advances in generative AI have demonstrated remarkable ability to produce high-quality content. However, these models often exhibit "Paradox of Simplicity": while they can render intricate landscapes, they often fail at simple, deterministic tasks. To address this, we formalize Obedience as the ability to align with instructions and establish a hierarchical grading system ranging from basic semantic alignment to pixel-level systemic precision, which provides a unified paradigm for incorporating and categorizing existing literature. Then, we conduct case studies to identify common obedience gaps, revealing how generative priors often override logical constraints. To evaluate high-level obedience, we present VIOLIN (VIsual Obedience Level-4 EvaluatIoN), the first benchmark focused on pure color generation across six variants. Extensive experiments on SOTA models reveal fundamental obedience limitations and further exploratory insights. By establishing this framework, we aim to draw more attention on AI Obedience and encourage deeper exploration to bridge this gap.

arXiv:2603.00170v2 Announce Type: cross Abstract: Craniofacial Superimposition is a forensic technique for identifying skeletal remains by comparing a post-mortem skull with ante-mortem facial photographs. A critical step in this process is Skull-Face Overlay (SFO). This stage involves aligning a 3D skull model with a 2D facial image, typically guided by cranial and facial landmarks' correspondence. However, its accuracy is undermined by individual variability in soft-tissue thickness, introducing significant uncertainty into the overlay. This paper introduces Lilium, an automated evolutionary method to enhance the accuracy and robustness of SFO. Lilium explicitly models soft-tissue variability using a 3D cone-based representation whose parameters are optimized via a Differential Evolution algorithm. The method enforces anatomical, morphological, and photographic plausibility through a combination of constraints: landmark matching, camera parameter consistency, head pose alignment, skull containment within facial boundaries, and region parallelism. This emulation of the usual forensic practitioners' approach leads Lilium to outperform the state-of-the-art method in terms of both accuracy and robustness.

arXiv:2603.00171v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) are shifting towards "Thinking with Images" by actively exploring image details. While effective, large-scale training is computationally expensive, which has spurred growing interest in lightweight, training-free solutions. However, existing training-free methods suffer from two flaws: perceptual redundancy from indiscriminate cropping, which adds overhead and noise; and a drift between semantic intent and spatial attention, which prevents accurate localization of user-focused regions. To address these challenges, we propose AdaFocus, a novel training-free framework designed for adaptive visual reasoning. AdaFocus follows a two-stage pipeline: a confidence-based module decides when to crop, and a semantic-guided localization module determines where to crop. This enables adaptive visual reasoning without additional training. Experimentally, AdaFocus delivers substantial performance gains while achieving approximately 4.0\times speedup inference speedup than the SOTA method ZoomEyes, representing a significant advance in both accuracy and efficiency.

arXiv:2603.00172v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) has emerged as a powerful paradigm for enhancing multimodal large language models by grounding their responses in external, factual knowledge and thus mitigating hallucinations. However, the integration of externally sourced knowledge bases introduces a critical attack surface. Adversaries can inject malicious multimodal content capable of influencing both retrieval and downstream generation. In this work, we present MM-MEPA, a multimodal poisoning attack that targets the metadata components of image-text entries while leaving the associated visual content unaltered. By only manipulating the metadata, MM-MEPA can still steer multimodal retrieval and induce attacker-desired model responses. We evaluate the attack across multiple benchmark settings and demonstrate its severity. MM-MEPA achieves an attack success rate of up to 91\% consistently disrupting system behaviors across four retrievers and two multimodal generators. Additionally, we assess representative defense strategies and find them largely ineffective against this form of metadata-only poisoning. Our findings expose a critical vulnerability in multimodal RAG and underscore the urgent need for more robust, defense-aware retrieval and knowledge integration methods.

arXiv:2603.00173v1 Announce Type: cross Abstract: We describe our experience training Summer-22B, a video foundation model developed from scratch. This report documents the engineering challenges, design decisions, and lessons learned while scaling from raw footage collection to a functional model trained on approximately 50 million clips. We outline our approach combining metadata-driven dataset curation, multi-stage filtering, $\mu$P parameterization, and hypersphere-constrained optimization. We developed the Lavender Data system for dataset management and adopted inference-aware architectural choices. We share observations on what worked in our setting: dataset engineering consumed the majority of effort, architectural variants showed smaller differences than we expected, and $\mu$P hyperparameter transfer appeared effective even under geometric constraints. We hope this account proves useful to others undertaking similar projects.

arXiv:2603.00174v1 Announce Type: cross Abstract: A constant weight binary code consists of $n$-bit binary codewords, each with exactly $w$ bits equal to 1, such that any two codewords are at least Hamming distance $d$ apart. $A(n,d,w)$ is the maximum size of a constant weight binary code with parameters $n,d,w$. We establish improved lower bounds on $A(n,d,w)$ by constructing new larger codes, for 24 values of $(n,d,w)$ with $6 \leq d \leq 18$ and $18 \leq n \leq 35$. The improved lower bounds come from two strategies. The first is a tabu search that operates at the level of bit swaps. The second is a novel greedy heuristic that repeatedly chooses the candidate codeword that maximizes a randomly-scored histogram of distances to previously-added codewords. These strategies were proposed by CPro1, an automated protocol that generates, implements, and tests diverse strategies for combinatorial constructions.

arXiv:2603.00176v1 Announce Type: cross Abstract: Shared micromobility services such as e-scooters and bikes have become an integral part of urban transportation, yet their efficiency critically depends on effective vehicle rebalancing. Existing methods either optimize for average demand patterns or employ robust optimization and reinforcement learning to handle predefined uncertainties. However, these approaches overlook emergent events (e.g., demand surges, vehicle outages, regulatory interventions) or sacrifice performance in normal conditions. We introduce AMPLIFY, an LLM-augmented policy adaptation framework for shared micromobility rebalancing. The framework combines a baseline rebalancing module with an LLM-based adaptation module that adjusts strategies in real time under emergent scenarios. The adaptation module ingests system context, demand predictions, and baseline strategies, and refines adjustments through self-reflection. Evaluations on real-world e-scooter data from Chicago show that our approach improves demand satisfaction and system revenue compared to baseline policies, highlighting the potential of LLM-driven adaptation as a flexible solution for managing uncertainty in micromobility systems.

arXiv:2603.00180v1 Announce Type: cross Abstract: Generative modeling of neural network parameters is often tied to architectures because standard parameter representations rely on known weight-matrix dimensions. Generation is further complicated by permutation symmetries that allow networks to model similar input-output functions while having widely different, unaligned parameterizations. In this work, we introduce Neural Network Diffusion Transformers (NNiTs), which generate weights in a width-agnostic manner by tokenizing weight matrices into patches and modeling them as locally structured fields. We establish that Graph HyperNetworks (GHNs) with a convolutional neural network (CNN) decoder structurally align the weight space, creating the local correlation necessary for patch-based processing. Focusing on MLPs, where permutation symmetry is especially apparent, NNiT generates fully functional networks across a range of architectures. Our approach jointly models discrete architecture tokens and continuous weight patches within a single sequence model. On ManiSkill3 robotics tasks, NNiT achieves >85% success on architecture topologies unseen during training, while baseline approaches fail to generalize.

arXiv:2603.00181v1 Announce Type: cross Abstract: A recent report on "Learning the natural history of human disease with generative transformers" created an opportunity to assess the engineering challenge of delivering user-facing Generative AI applications in privacy-sensitive domains. The application of these models, particularly for personalized healthcare tasks like predicting individual morbidity risk, is typically constrained by data privacy concerns. This project was accordingly designed as an in-browser model deployment exercise (an "App") testing the architectural boundaries of client-side inference generation (no downloads or installations). We relied exclusively on the documentation provided in the reference report to develop the model, specifically testing the "R" component of the FAIR data principles: Findability, Accessibility, Interoperability, and Reusability. The successful model deployment, leveraging ONNX and a custom JavaScript SDK, establishes a secure, high-performance architectural blueprint for the future of private generative AI in medicine.

arXiv:2603.00182v1 Announce Type: cross Abstract: Cross-robot policy learning -- training a single policy to perform well across multiple embodiments -- remains a central challenge in robot learning. Transformer-based policies, such as vision-language-action (VLA) models, are typically embodiment-agnostic and must infer kinematic structure purely from observations, which can reduce robustness across embodiments and even limit performance within a single embodiment. We propose an embodiment-aware transformer policy that injects morphology via three mechanisms: (1) kinematic tokens that factorize actions across joints and compress time through per-joint temporal chunking; (2) a topology-aware attention bias that encodes kinematic topology as an inductive bias in self-attention, encouraging message passing along kinematic edges; and (3) joint-attribute conditioning that augments topology with per-joint descriptors to capture semantics beyond connectivity. Across a range of embodiments, this structured integration consistently improves performance over a vanilla pi0.5 VLA baseline, indicating improved robustness both within an embodiment and across embodiments.

arXiv:2603.00183v1 Announce Type: cross Abstract: Context: Test case prioritization (TCP) is a technique widely used by software development organizations to accelerate regression testing. Objectives: We aim to systematize existing TCP knowledge and to propose and empirically evaluate a new TCP approach. Methods: We conduct a snowballing review (SR) on TCP, implement a~comprehensive platform for TCP research (TCPFramework), analyze existing evaluation metrics and propose two new ones (\rAPFDc{} and ATR), and develop a~family of ensemble TCP methods called approach combinators. Results: The SR helped identify 324 studies related to TCP. The techniques proposed in our study were evaluated on the RTPTorrent dataset, consistently outperforming their base approaches across the majority of subject programs, and achieving performance comparable to the current state of the art for heuristical algorithms (in terms of \rAPFDc{}, NTR, and ATR), while using a distinct approach. Conclusions: The proposed methods can be used efficiently for TCP, reducing the time spent on regression testing by up to 2.7\%. Approach combinators offer significant potential for improvements in future TCP research, due to their composability.

arXiv:2603.00184v1 Announce Type: cross Abstract: Bird image segmentation remains a challenging task in computer vision due to extreme pose diversity, complex plumage patterns, and variable lighting conditions. This paper presents a dual-pipeline framework for binary bird image segmentation leveraging 2025 foundation models. We introduce two operating modes built upon Segment Anything Model 2.1 (SAM 2.1) as a shared frozen backbone: (1) a zero-shot pipeline using Grounding DINO 1.5 to detect birds via the text prompt "bird" before prompting SAM 2.1 with bounding boxes requiring no labelled bird data; and (2) a supervised pipeline that fine-tunes YOLOv11 on the CUB-200-2011 dataset for high-precision detection, again prompting SAM 2.1 for pixel-level masks. The segmentation model is never retrained for new species or domains. On CUB-200-2011 (11,788 images, 200 species), the supervised pipeline achieves IoU 0.912, Dice 0.954, and F1 0.953 outperforming all prior baselines including SegFormer-B2 (IoU 0.842) by +7.0 percentage points. The zero-shot pipeline achieves IoU 0.831 using only a text prompt, the first such result reported on this benchmark. We demonstrate that prompt-based foundation model pipelines outperform task specific end-to-end trained segmentation networks, while requiring only lightweight detector fine-tuning (~1 hour) for domain adaptation. Complete PyTorch implementation, dataset preparation scripts, and trained weights are publicly available.

arXiv:2603.00185v1 Announce Type: cross Abstract: Intrusion detection in IoT and industrial networks requires models that can detect rare attacks at low false-positive rates while remaining reliable under evolving traffic and limited labels. Existing IDS solutions often report strong in-distribution accuracy, but they may degrade when evaluated on future traffic, unseen (zero-day) attack families, or adversarial feature manipulations, and many systems provide limited evidence to support analyst triage. To address these gaps, we propose ThreatFormer- IDS, a Transformer-based sequence modeling framework that converts flow records into time-ordered windows and learns contextual representations for robust intrusion screening. The method combines (i) weighted supervised learning for imbalanced detection, (ii) masked self-supervised learning to improve representation stability under drift and sparse labels, (iii) PGDbased adversarial training with scale-normalized perturbations to strengthen resilience against feature-level evasion, and (iv) Integrated Gradients attribution to highlight influential time steps and features for each alert. On the ToN IoT benchmark with chronological evaluation, ThreatFormer-IDS achieves AUCROC 0.994, AUC-PR 0.956, and Recall@1%FPR 0.910, outperforming strong tree-based and sequence baselines. Under a zero-day protocol with held-out attack families, it maintains superior generalization (AUC-PR 0.721, Recall@1%FPR 0.783). Robustness tests further show slower degradation in AUCPR as the adversarial budget increases, confirming improved stability under bounded perturbations. Overall, ThreatFormer- IDS provides a unified, deployment-oriented IDS pipeline that balances detection quality, zero-day behavior, robustness, and explainability.

arXiv:2603.00186v1 Announce Type: cross Abstract: Financial systems run nonstop and must stay reliable even during cyber incidents. Modern attacks move across many services (apps, APIs, identity, payment rails), so defenders must make a sequence of actions under time pressure. Most security tools still use fixed rules or static playbooks, which can be slow to adapt when the attacker changes behavior. Reinforcement learning (RL) is a good fit for sequential decisions, but much of the RL-in-finance literature targets trading and does not model real cyber response limits such as action cost, service disruption, and defender coordination across many assets. This paper proposes RLShield, a practical multi-agent RL pipeline for financial cyber defense. We model the enterprise attack surface as a Markov decision process (MDP) where states summarize alerts, asset exposure, and service health, and actions represent real response steps (e.g., isolate a host, rotate credentials, ratelimit an API, block an account, or trigger recovery). RLShield learns coordinated policies across multiple agents (assets or service groups) and optimizes a risk-sensitive objective that balances containment speed, business disruption, and response cost. We also include a game-aware evaluation that tests policies against adaptive attackers and reports operational outcomes, not only reward. Experiments show that RLShield reduces time-to-containment and residual exposure while keeping disruption within a fixed response budget, outperforming static rule baselines and single-agent RL under the same constraints. These results suggest that multi-agent, cost-aware RL can provide a deployable layer for automated response in financial security operations.

arXiv:2603.00188v1 Announce Type: cross Abstract: Large Vision-Language Models (VLMs) have emerged as powerful engines for autonomous GUI agents, yet their deployment is severely constrained by the substantial memory footprint and latency of the Key-Value (KV) cache during long-horizon interactions. While existing cache compression methods have proven effective for LLMs, we empirically demonstrate that they suffer from suboptimal performance in GUI scenarios due to a fundamental misalignment: unlike general visual tasks where attention sparsity varies across layers, GUI attention patterns exhibit uniform high-sparsity across all transformer layers. Motivated by this insight, we propose ST-Lite, a training-free KV cache compression framework tailored for efficient GUI agents that explicitly addresses the dynamic spatio-trajectory dependencies within GUI data streams. ST-Lite introduces a novel dual-branch scoring policy incorporating Component-centric Spatial Saliency (CSS) and Trajectory-aware Semantic Gating (TSG). Specifically, CSS preserves the structural integrity of interactive UI elements by evaluating local neighborhood saliency, while TSG mitigates historical redundancy by dynamically filtering visually repetitive KV pairs within the interaction trajectory. Extensive evaluations demonstrate that with only a 10-20% cache budget, ST-Lite achieves a 2.45x decoding acceleration while maintaining comparable or even superior performance compared to full-cache baselines, offering a scalable solution for resource-constrained GUI agents.

arXiv:2603.00190v1 Announce Type: cross Abstract: Polysomnography (PSG) provides the gold standard for sleep assessment but suffers from substantial heterogeneity across recording devices and cohorts. There have been growing efforts to build general-purpose foundation models (FMs) for sleep physiology, but lack an in-depth understanding of the pre-training process and scaling patterns that lead to more generalizable sleep FMs. To fill this gap, we curate a massive corpus of 166,500 hours of sleep recordings from nine public sources and establish SleepBench, a comprehensive, fully open-source benchmark. Leveraging SleepBench, we systematically evaluate four families of self-supervised pre-training objectives and uncover three critical findings: (1) existing FMs fail to generalize to missing channels at inference; (2) channel-invariant feature learning is essential for pre-training; and (3) scaling sample size, model capacity, and multi-source data mixture consistently improves downstream performance.With an enhanced pre-training and scaling recipe, we introduce OSF, a family of sleep FMs that achieves state-of-the-art performance across nine datasets on diverse sleep and disease prediction tasks. Further analysis of OSF also reveals intriguing properties in sample efficiency, hierarchical aggregation, and cross-dataset scaling.

arXiv:2603.00194v1 Announce Type: cross Abstract: The rise of text-to-video generation models has raised growing concerns over content authenticity, copyright protection, and malicious misuse. Watermarking serves as an effective mechanism for regulating such AI-generated content, where high fidelity and strong robustness are particularly critical. Recent generative image watermarking methods provide a promising foundation by leveraging watermark information and pseudo-random keys to control the initial sampling noise, enabling lossless embedding. However, directly extending these techniques to videos introduces two key limitations: Existing designs implicitly rely on strict alignment between video frames and frame-dependent pseudo-random binary sequences used for watermark encryption. Once this alignment is disrupted, subsequent watermark extraction becomes unreliable; and Video-specific distortions, such as inter-frame compression, significantly degrade watermark reliability. To address these issues, we propose SKeDA, a generative watermarking framework tailored for text-to-video diffusion models. SKeDA consists of two components: (1) Shuffle-Key-based Distribution-preserving Sampling (SKe) employs a single base pseudo-random binary sequence for watermark encryption and derives frame-level encryption sequences through permutation. This design transforms watermark extraction from synchronization-sensitive sequence decoding into permutation-tolerant set-level aggregation, substantially improving robustness against frame reordering and loss; and (2) Differential Attention (DA), which computes inter-frame differences and dynamically adjusts attention weights during extraction, enhancing robustness against temporal distortions. Extensive experiments demonstrate that SKeDA preserves high video generation quality and watermark robustness.

arXiv:2603.00195v1 Announce Type: cross Abstract: The rapid proliferation of agentic AI skill ecosystems -- exemplified by OpenClaw (228,000 GitHub stars) and Anthropic Agent Skills (75,600 stars) -- has introduced a critical supply chain attack surface. The ClawHavoc campaign (January-February 2026) infiltrated over 1,200 malicious skills into the OpenClaw marketplace, while MalTool catalogued 6,487 malicious tools that evade conventional detection. In response, twelve reactive security tools emerged, yet all rely on heuristic methods that provide no formal guarantees. We present SkillFortify, the first formal analysis framework for agent skill supply chains, with six contributions: (1) the DY-Skill attacker model, a Dolev-Yao adaptation to the five-phase skill lifecycle with a maximality proof; (2) a sound static analysis framework grounded in abstract interpretation; (3) capability-based sandboxing with a confinement proof; (4) an Agent Dependency Graph with SAT-based resolution and lockfile semantics; (5) a trust score algebra with formal monotonicity; and (6) SkillFortifyBench, a 540-skill benchmark. SkillFortify achieves 96.95% F1 (95% CI: [95.1%, 98.4%]) with 100% precision and 0% false positive rate on 540 skills, while SAT-based resolution handles 1,000-node graphs in under 100 ms.

arXiv:2603.00196v1 Announce Type: cross Abstract: The increasing reliance on cloud-hosted Large Language Models (LLMs) exposes sensitive client data, such as prompts and responses, to potential privacy breaches by service providers. Existing approaches fail to ensure privacy, maintain model performance, and preserve computational efficiency simultaneously. To address this challenge, we propose Talaria, a confidential inference framework that partitions the LLM pipeline to protect client data without compromising the cloud's model intellectual property or inference quality. Talaria executes sensitive, weight-independent operations within a client-controlled Confidential Virtual Machine (CVM) while offloading weight-dependent computations to the cloud GPUs. The interaction between these environments is secured by our Reversible Masked Outsourcing (ReMO) protocol, which uses a hybrid masking technique to reversibly obscure intermediate data before outsourcing computations. Extensive evaluations show that Talaria can defend against state-of-the-art token inference attacks, reducing token reconstruction accuracy from over 97.5% to an average of 1.34%, all while being a lossless mechanism that guarantees output identical to the original model without significantly decreasing efficiency and scalability. To the best of our knowledge, this is the first work that ensures clients' prompts and responses remain inaccessible to the cloud, while also preserving model privacy, performance, and efficiency.

arXiv:2603.00197v1 Announce Type: cross Abstract: Deep Neural Networks (DNNs) have advanced applications in domains such as healthcare, autonomous systems, and scene understanding, yet the internal semantics of their hidden neurons remain poorly understood. Prior work introduced a Concept Induction-based framework for hidden neuron analysis and demonstrated its effectiveness on the ADE20K dataset. In this case study, we investigate whether the approach generalizes by applying it to the SUN2012 dataset, a large-scale scene recognition benchmark. Using the same workflow, we assign interpretable semantic labels to neurons and validate them through web-sourced images and statistical testing. Our findings confirm that the method transfers to SUN2012, showing its broader applicability.

arXiv:2603.00198v1 Announce Type: cross Abstract: Token reduction is an effective way to accelerate long-video vision-language models (VLMs), but most existing methods are designed for dense Transformers and do not directly account for hybrid architectures that interleave attention with linear-time state-space blocks (e.g., Mamba). We study query-conditioned token reduction for hybrid video VLMs and analyze reduction behavior through two properties: layerwise sparsity (how many tokens capture query-relevant information) and importance stability (whether token-importance rankings persist across depth). Although token importance is sparse within each layer, the set of important tokens changes across layers, so aggressive early pruning is unreliable. Motivated by this, we propose a low-to-high progressive reduction schedule and a unified language-aware scoring mechanism for both attention and Mamba blocks (using an implicit-attention proxy for Mamba), enabling all-layer token reduction in hybrids. Under an aggressive compression setting (retaining 25% of visual tokens), our approach delivers substantial prefilling speedups (3.8--4.2x) with near-baseline accuracy at test time, and light finetuning under reduction further improves performance on long-context video benchmarks.

arXiv:2603.00199v1 Announce Type: cross Abstract: Axial piston pumps are indispensable power sources in high-stakes fluid power systems, including aerospace, marine, and heavy machinery applications. Their operational reliability is frequently compromised by compound faults that simultaneously affect multiple friction pairs. Conventional data-driven diagnosis methods suffer from severe data scarcity for compound faults and poor generalization across varying operating conditions. This paper proposes a novel multi-condition physics-data coupled digital twin calibration framework that explicitly resolves the fundamental uncertainty of pump outlet flow ripple. The framework comprises three synergistic stages: in-situ virtual high-frequency flow sensing on a dedicated rigid metallic segment, surrogate model-assisted calibration of the 3D CFD source model using physically estimated ripple amplitudes, and multi-objective inverse transient analysis for viscoelastic unsteady-friction pipeline parameter identification. Comprehensive experiments on a test rig demonstrate that the calibrated digital twin accurately reproduces both single-fault and two representative compound-fault. These results establish a high-fidelity synthetic fault-generation capability that directly enables robust zero-shot fault diagnosis under previously unseen operating regimes and fault combinations, thereby advancing predictive maintenance in complex hydraulic systems.

arXiv:2603.00200v1 Announce Type: cross Abstract: The rapid evolution of sophisticated cyberattacks has strained modern Security Operations Centers (SOC), which traditionally rely on rule-based or signature-driven detection systems. These legacy frameworks often generate high volumes of technical alerts that lack organizational context, leading to analyst fatigue and delayed incident responses. This paper presents LiaisonAgent, an autonomous multi-agent system designed to bridge the gap between technical risk detection and business-level risk governance. Built upon the QWQ-32B large reasoning model, LiaisonAgent integrates specialized sub-agents, including human-computer interaction agents, comprehensive judgment agents, and automated disposal agents-to execute end-to-end investigation workflows. The system leverages a hybrid planning architecture that combines deterministic workflows for compliance with autonomous reasoning based on the ReAct paradigm to handle ambiguous operational scenarios. Experimental evaluations across diverse security contexts, such as large-scale data exfiltration and unauthorized account borrowing, achieve an end-to-end tool-calling success rate of 97.8% and a risk judgment accuracy of 95%. Furthermore, the system exhibits significant resilience against out-of-distribution noise and adversarial prompt injections, while achieving a 92.7% reduction in manual investigation overhead.

arXiv:2603.00201v1 Announce Type: cross Abstract: One of the common issues in clinical decision-making is the presence of uncertainty, which often arises due to ambiguity in radiology reports, which often reflect genuine diagnostic uncertainty or limitations of automated label extraction in various complex cases. Especially the case of multilabel datasets such as CheXpert, MIMIC-CXR, etc., which contain labels such as positive, negative, and uncertain. In clinical decision-making, the uncertain label plays a tricky role as the model should not be forced to provide a confident prediction in the absence of sufficient evidence. The ability of the model to say it does not understand whenever it is not confident is crucial, especially in the cases of clinical decision-making involving high risks. Here, we propose AdURA-Net, a geometry-driven adaptive uncertainty-aware framework for reliable thoracic disease classification. The key highlights of the proposed model are: a) Adaptive dilated convolution and multiscale deformable alignment coupled with the backbone Densenet architecture capturing the anatomical complexities of the medical images, and b) Dual Head Loss, which combines masked binary cross entropy with logit and a Dirichlet evidential learning objective.

arXiv:2603.00205v1 Announce Type: cross Abstract: Generative models, particularly Diffusion Models (DM), have shown strong potential for Computed Tomography (CT) reconstruction serving as expressive priors for solving ill-posed inverse problems. However, diffusion-based reconstruction relies on Stochastic Differential Equations (SDEs) for forward diffusion and reverse denoising, where such stochasticity can interfere with repeated data consistency corrections in CT reconstruction. Since CT reconstruction is often time-critical in clinical and interventional scenarios, improving reconstruction efficiency is essential. In contrast, Flow Matching (FM) models sampling as a deterministic Ordinary Differential Equation (ODE), yielding smooth trajectories without stochastic noise injection. This deterministic formulation is naturally compatible with repeated data consistency operations. Furthermore, we observe that FM-predicted velocity fields exhibit strong correlations across adjacent steps. Motivated by this, we propose an FM-based CT reconstruction framework (FMCT) and an efficient variant (EFMCT) that reuses previously predicted velocity fields over consecutive steps to substantially reduce the number of Neural network Function Evaluations (NFEs), thereby improving inference efficiency. We provide theoretical analysis showing that the error introduced by velocity reuse is bounded when combined with data consistency operations. Extensive experiments demonstrate that FMCT/EFMCT achieve competitive reconstruction quality while significantly improving computational efficiency compared with diffusion-based methods. The codebase is open-sourced at https://github.com/EFMCT/EFMCT.

arXiv:2603.00206v1 Announce Type: cross Abstract: Existing visual reasoning benchmarks predominantly rely on natural language prompts, evaluate narrow reasoning modalities, or depend on subjective scoring procedures such as LLM-as-judge. We introduce the TACIT Benchmark, a programmatic visual reasoning benchmark comprising 10 tasks across 6 reasoning domains: spatial navigation, abstract pattern completion, causal simulation, logical constraint satisfaction, graph theory, and topology. The benchmark provides dual-track evaluation: a generative track in which models must produce solution images verified through deterministic computer-vision pipelines, and a discriminative track offering five-way multiple choice with structurally plausible near-miss distractors. Each distractor violates exactly one structural constraint, requiring models to reason about fine-grained visual differences rather than exploit superficial cues. Version 0.1.0 distributes 6,000 puzzles (108,000 PNG images across three resolutions) with fully deterministic seeded generation and reproducible verification. The dataset, generation code, and evaluation harness are released under the Apache 2.0 license on HuggingFace (DOI: 10.57967/hf/7904).

arXiv:2603.00207v1 Announce Type: cross Abstract: Advances in large reasoning models have shown strong performance on complex reasoning tasks by scaling test-time compute through extended reasoning. However, recent studies observe that in vision-dependent tasks, extended textual reasoning at inference time can degrade performance as models progressively lose attention to visual tokens and increasingly rely on textual priors alone. To address this, prior works use reinforcement learning (RL)-based fine-tuning to route visual tokens or employ refocusing mechanisms during reasoning. While effective, these methods are computationally expensive, requiring large-scale data generation and policy optimization. To leverage the benefits of test-time compute without additional RL fine-tuning, we propose VisRef, a visually grounded test-time scaling framework. Our key idea is to actively guide the reasoning process by re-injecting a coreset of visual tokens that are semantically relevant to the reasoning context while remaining diverse and globally representative of the image, enabling more grounded multi-modal reasoning. Experiments on three visual reasoning benchmarks with state-of-the-art multi-modal large reasoning models demonstrate that, under fixed test-time compute budgets, VisRef consistently outperforms existing test-time scaling approaches by up to 6.4%.