Papers

New on arXiv

Sat 22 Aug 2026

205 papers

Improving Generalization and Robustness in Offline Reinforcement Learning via Boundary-Aware Data AugmentationOffline RL and imitation learning from data → Offline RL algorithms
FRAC-MAS: A Safe and Explainable Multi-Agent System for Fracture DiagnosisLLM agents and tool use → Medical and scientific domain agents
LLM Agents Perform Controlled Experiments Using Simulation ModelsLLM agents and tool use → Medical and scientific domain agents
Rebuild Dossier: Mechanically-Enforced Specs for Agentic App Rebuilds, and What Model-Tier Failures RevealLLM agents and tool use → Software engineering coding agents
Lexical Perturbations Disrupt LLM Reasoning: An Empirical Study of Attention DiversionLLM reasoning → Reasoning benchmarks and evaluation
SSE-Bio: A Structured Self-Evolving Agent with Agentic Retrieval Policy for Multi-Hop Biomedical ReasoningLLM agents and tool use → Medical and scientific domain agents
PropUQ-MAS: Propagation-Aware Uncertainty Quantification for LLM Multi-Agent SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
Task-Driven 3D Printability Assistance via Geometry- and Knowledge-Grounded LLM ReasoningLLM reasoning → Spatial and embodied reasoning
Decoupled Physical Modeling and Execution for Physics ReasoningLLM reasoning → Spatial and embodied reasoning
FlowSep 2: Self-Supervised Flow Matching for Language-Queried Audio Source SeparationDiffusion models → Audio, speech and music
How Weight Encoding Affects Language Model Placement and Performance on the Apple Neural EngineEfficient LLM inference and serving → Serving systems and request scheduling
Hack-Verifiable Terminal Bench: Evaluating Reward Hacking in Terminal TasksLLM alignment / RLHF / preference optimisation → Reward hacking and misspecification
Semantic Reasoning Denoising: Correcting Language Model Reasoning with Semantic OperatorsLLM reasoning → Diffusion reasoning and test-time scaling
W-RAG: Source-Aware Retrieval for Enterprise Document Generation from Heterogeneous Knowledge BasesRetrieval-augmented generation → RAG pipeline design and evaluation
Spine-Branch Coordination for Multi-agent Computer UseLLM agents and tool use → General LLM agent frameworks and benchmarks
DELE-w0.5: Inferring Action from Future Latent State for Robotic ManipulationWorld models → Embodied and robot world models
From SQL Generation to Tool Selection: A Domain-Oriented Pattern for MCP ServersLLM agents and tool use → Tool learning and function calling
MEMORY Wins All: Indirect Bias Injection Attacks via Social Media FeedsLLM safety and security → Prompt injection and agent security
More Accurate or More Efficient? Evaluating Locally Deployed Compact Open-Weight Language Models for Mathematical ReasoningLLM reasoning → Mathematical reasoning
Ludiscriptscriptstyle 0.1: An Agentic System for Socially Intelligent RobotsLLM agents and tool use → General LLM agent frameworks and benchmarks
Align, Unify, Suppress, Route: A Coherentist View of Transformer ComputationMechanistic interpretability → Circuits and mechanistic analysis
DynaContext: Self-Improving Dynamic Contextualization of Optimized Prompts for Heterogeneous Parameter ExtractionSynthetic data and self-improvement for LLMs → Self-rewarding alignment and self-training
The Communication Map of a TransformerMechanistic interpretability → Circuits and mechanistic analysis
Close Shortcut Wins Long: Seeking Diverse and Stable Generators for Data-Free Knowledge DistillationKnowledge distillation → General knowledge distillation methods
Gated Decoupled Compositional Bandits: A Unified Theory of Contextual Bandits with Supervised-Calibrated Action Scaling and Pre-Execution GatingLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Redteaming Leading Arabic LLMs with ASASLLM safety and security → Safety alignment, refusal and guardrails
Closed-loop AI achieves certifiable engineering designLLM agents and tool use → Multimodal and embodied agents
Improved denoising diffusion probabilistic models with efficient non-diagonal covariance modelingDiffusion models → Diffusion foundations and sampling
Retrieval-Augmented Visual Prompting: Guiding Foundation Models in Two-Photon ImagingRetrieval-augmented generation → Multimodal RAG
ToSCA: Leveraging Hierarchical Reinforcement Learning on Temporal and Strategic Abstractions of Conversational AgentsDeep reinforcement learning methods → RL training of language and agent models
SSDi8: Accurate and Efficient 8-bit Quantization for State Space DualityState-space sequence models (S4, Mamba) → Core SSM architectures for language
Sparse Multi-Stage Expert-Agent Routing for Complex Clinical ReasoningLLM reasoning → Clinical and biomedical applications +1
EDGE: Experience-Distillation for Guided Exploration in Agentic Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning +1
Enhancing Group Recommendation with Memory-Augmented Reasoning in LLM AgentLLM reasoning → LLM reasoning for recommendation +1
SkillBloat: Token Amplification Attacks via Skill Injection in LLM Coding AgentsLLM agents and tool use → Agent threats, guardrails and governance
ESCRAG-R1: Retrieval-Augmented Reinforcement Learning for Emotional Support ConversationRetrieval-augmented generation → Medical and biomedical RAG +1
Beyond Fixed Directions: Adaptive Representation Analysis of Reasoning and Memorization in LLMsLLM reasoning → Reasoning benchmarks and evaluation
OptiMAS: Automatically Optimize Multi-Agent SystemLLM agents and tool use → General LLM agent frameworks and benchmarks
CIDER: Continual Interactive Distillation for Embodied Reinforcement LearningDeep reinforcement learning methods → RL for robot locomotion and manipulation
Training Needs Trustworthy Worlds: Verified Synthetic Web Environments for Agent LearningLLM agents and tool use → Web, search, and research agents
From Solver Feedback to Faithful Plans: Multi-Role Reinforcement Learning for Symbolic PlanningDeep reinforcement learning methods → RL training of language and agent models
Breaking the Assumptions: Auditing Input-Side Jailbreak Defenses Against Semantic AttacksLLM safety and security → Jailbreaking
Pixel-Space Diffusion via Observation OperatorsDiffusion models → Diffusion foundations and sampling
Loop Engineering: Building Blocks, Adoption, and ImpactLLM agents and tool use → Software engineering coding agents
VIG: Visual Information Gain as a Reward Signal for Multimodal Chain-of-Thought CompressionLLM reasoning → Multimodal reasoning +1
BanglaVeilGuard: Cross-Script Safety Benchmarking and Lightweight Guardrails for Bangla Large Language ModelsLLM safety and security → Safety alignment, refusal and guardrails
ViSMoE: Visual-Aware Sparse Mixture-of-Experts for Embodied Referring Expression GroundingMixture-of-experts → Vision mixture-of-experts
The Chase Is the Curriculum, the Capture Anchors the Credit: Pursuit-Evasion Self-Play for Zero-Data LLM ReasoningLLM reasoning → Reinforcement learning for reasoning +1
GuardPaint:SpeculativeSafetyDecodingforText-to-ImageGenerationDiffusion models → Robustness, privacy and adversarial aspects +1
MemGuard: Persisting Verifier Signals for LLM-Agent Memory GovernanceLLM agents and tool use → Agent memory systems
BioMed-Agent-RL: A Meta Learning, All You Need for Biomedical ApplicationsLLM agents and tool use → Medical and scientific domain agents +1
HiDiffTIR: Hierarchical Difficulty-Aware Policy Optimization for Multi-Turn Tool-Integrated ReasoningLLM reasoning → Reinforcement learning for reasoning +1
ChainPrune: Evaluating and Reducing Redundancy in Long Chain-of-Thought ReasoningLLM reasoning → Prompting and chain-of-thought reasoning
GaussVid: Sparse-View Gaussian Splatting with 3D-Aware Video Diffusion PriorsDiffusion models → 3D, point cloud and scene generation
More Experts, Worse Dynamics: Inverse Scaling and Spectral Bias in Mixture-of-Experts State-Space ModelsMixture-of-experts → MoE theory and gating functions
FIRM-Video: Check Before You Score for Reliable Text-to-Video Reward ModelingLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
LLM4LLM: Bridging Kernel Benchmarks and Real Deployment via Closed-Loop Agentic OptimizationLLM agents and tool use → Software engineering coding agents
GameXpert-Bench: How Far Are Coding Agents from Expert Game Development?LLM agents and tool use → Software engineering coding agents
GUI-Primitives: Diagnosing Spatial Reasoning Failures in Vision-Language GUI GroundingLLM reasoning → Multimodal reasoning +1
Beyond Success and Failure: Length-Aware Contrastive Learning for GUI AgentsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
Training a Knowledge Base: Supervised Structure Learning for Agent-Curated Document StoresRetrieval-augmented generation → Agentic RAG, agents and memory
PatchGate: Narrowing the Verbalization Gap with Intrinsic Object Inventories in Frozen Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Hints, Critics, and Teachers: Prior Injection for Sparse-Reward RL in Vision-Language Math ReasoningLLM reasoning → Multimodal reasoning
MSM-Mem: A Universal Medical Structured Multimodal Memory Framework for Medical AI AgentsLLM agents and tool use → Medical and scientific domain agents
MCite-RL: Towards Reliable Multimodal RAG via Citation-enhanced Agentic Reinforcement LearningRetrieval-augmented generation → Agentic RAG, agents and memory +1
HIRA: A Human-in-the-Loop Retrieval-Augmented Cascade for Document Classification in Regulated IndustriesRetrieval-augmented generation → Generation and context integration
No One Model Catches Every Harm: Benchmarking Content Moderation Across Safety ScenariosLLM safety and security → Safety alignment, refusal and guardrails
Learning to Look Again: Loss-Gap Supervision for Free-form Crop Routing in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
FCPRAG: Fusion-Controller Parametric Retrieval-Augmented Generation for Stable Multi-Passage LoRA InjectionRetrieval-augmented generation → RAG robustness and adaptive retrieval
Architecture as Capability Equalizer for Coding AgentsLLM agents and tool use → Software engineering coding agents
CounterAlign: Counterfactual Supervision for Vision-Language-Action ModelsVision-language-action models → Failure recovery and robustness +1
Reinforcement Learning on Benign Facts Amplifies Leakage of Memorized Private DataDeep reinforcement learning methods → RL training of language and agent models
PowerSlider: Exploiting Phase Asymmetry for LLM Serving under Demand ResponseEfficient LLM inference and serving → Serving systems and request scheduling
ATHENA: Knowledge-guided agentic neural architecture search for AutoFormer-based electronic health record modelingLLM agents and tool use → Medical and scientific domain agents
Posterior Information Dynamics of Diffusion Models for Linear Inverse ProblemsDiffusion models → Inverse problems and image restoration
From Association to Causation: Improving Retrieval Precision of Retrieval-Augmented Generation via Causal Relations and an Attention MechanismRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Scalable quantum simulation of continuous-time generative models via tensor networksDiffusion models → Diffusion foundations and sampling
Benchmarking Composable Compression Techniques in Mixture-of-Experts LLMsMixture-of-experts → MoE large language models +1
Structured Neural Modeling of Daily Arctic Sea-Ice Concentration Evolution: Physical-Trajectory-Driven Learning and Forecast-Domain Adaptation
Quality over Quantity: Semi-Supervised Detection of Illicit Bitcoin Flows via Feature Engineering
Binary Quantized Neural Network Training Is W[1]-Hard Parameterized by Input and Output Dimensions
Dirichlet Process Mixtures of Trees with Gaussian Process Splits: A Bayesian Nonparametric Framework with Posterior Contraction Rate
Learning When Not to Listen: Selective Anti-Interference Pretraining for Language Models
GaussianDS: Depth-supervised Semantic Gaussian Splatting for Scene Understanding
Beyond Accuracy and Surface Fluency: Risk-Sensitive Evaluation of LLMs for Legal Clause Generation
A Hybrid Gaze-Motor Imagery BCI Framework for Effective Decision Communication
Open-Set Cattle Muzzle Identification: A Leakage-Controlled Benchmark and Evaluation Protocol
Effectiveness of IoT and Deep Learning for Detection and Severity Assessment of Postelectrotermes militaris in Tea Plantations
When the Canonical Completion Is Wrong: Formalizing and Measuring the Jump in Large Language Models
Investigating the Influence of Prompt and Response Languages on LLM Content Generation
Can You Say This for Me? Speaking Up by Proxy in Co-Located Discussion
TutorTrace: A Dataset and Taxonomy for Classifying Learner Behavioral States during AI-Assisted Programming Education
When May an Agent Stop? Evidence-Carrying Termination for Tool-Using LLMs
Identifying Latent Declarative Representations of Code for Assisting Repository Migration
Learning Reduced-Order Dynamics with Singularity via Latent-Augmented Neural Ordinary Differential Equations
MEMONDEMAND: A Memory Management System for Large-Scale Enterprise Data
Measuring Stability and Failure Behavior in Language Models Under Structured Perturbations
MegaMem: A Retrieval Solution for Ultra-Large Context Windows
TRACE: Artifact-Robust Statistical Shape Modeling from Imperfect Surface Scans - A Case Study in Craniosynostosis 3D Photography
Blockwise Stabilized Adaptive Cubic Regularization with Subsolvers via Recurrence

All 205