Papers

New on arXiv

Wed 15 Oct 2025

407 papers

When "Correct" Is Not Safe: Can We Trust Functionally Correct Patches Generated by Code Agents?LLM agents and tool use → Agent threats, guardrails and governance
Shortcutting Pre-trained Flow Matching Diffusion Models is Almost Free LunchDiffusion models → Flow matching, time series and science +1
Feature-driven reinforcement learning for photovoltaic in continuous intraday tradingDeep reinforcement learning methods → RL for energy, networks and operations
InfraGPT Smart Infrastructure: An End-to-End VLM-Based Framework for Detecting and Managing Urban DefectsMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Transfer learning strategies for accelerating reinforcement-learning-based flow controlDeep reinforcement learning methods → RL with diffusion and flow models
FinAI Data Assistant: LLM-based Financial Database Query Processing with the OpenAI Function Calling APILLM agents and tool use → Financial trading and forecasting agents
Combining Reinforcement Learning and Behavior Trees for NPCs in Video Games with AMD ScholaDeep reinforcement learning methods → RL for games and combinatorial optimization
CodeEvolve: an open source evolutionary coding agent for algorithmic discovery and optimizationLLM agents and tool use → Software engineering coding agents
Formalizing the Safety, Security, and Functional Properties of Agentic AI SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
Demystifying the Mechanisms Behind Emergent Exploration in Goal-conditioned RLDeep reinforcement learning methods → Deep RL algorithms and representations
ViTacGen: Robotic Pushing with Vision-to-Touch GenerationDeep reinforcement learning methods → RL for robot locomotion and manipulation
Briding Diffusion Posterior Sampling and Monte Carlo methods: a surveyDiffusion models → Inverse problems and image restoration
Unlocking Out-of-Distribution Generalization in Transformers via Recursive Latent Space ReasoningLLM reasoning → Mathematical reasoning
Optimistic Reinforcement Learning-Based Skill Insertions for Task and Motion PlanningDeep reinforcement learning methods → RL for robot locomotion and manipulation
Vgent: Graph-based Retrieval-Reasoning-Augmented Generation For Long Video UnderstandingLLM reasoning → Multimodal reasoning +1
Think Globally, Group Locally: Evaluating LLMs Using Multi-Lingual Word Grouping GamesLLM reasoning → Reasoning for agents, planning and code
Context-Selective State Space Models: Feedback is All You NeedState-space sequence models (S4, Mamba) → Core SSM architectures for language
CRaFT: An Explanation-Based Framework for Evaluating Cultural Reasoning in Multilingual Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Stop Reducing Responsibility in LLM-Powered Multi-Agent Systems to Local AlignmentLLM agents and tool use → General LLM agent frameworks and benchmarks
PIShield: Detecting Prompt Injection Attacks via Intrinsic LLM FeaturesLLM safety and security → Prompt injection and agent security
A Diffusion-Refined Planner with Reinforcement Learning Priors for Confined-Space ParkingDeep reinforcement learning methods → RL for autonomous driving and traffic
REAP the Experts: Why Pruning Prevails for One-Shot MoE compressionMixture-of-experts → MoE large language models
Efficient Few-Shot Learning in Remote Sensing: Fusing Vision and Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Signature in Code Backdoor Detection, how far are we?LLM safety and security → Backdoor and data poisoning attacks
Static Sandboxes Are Inadequate: Modeling Societal Complexity Requires Open-Ended Co-Evolution in LLM-Based Multi-Agent SimulationsLLM agents and tool use → Social simulation and multi-agent debate
Classifying and Addressing the Diversity of Errors in Retrieval-Augmented Generation SystemsRetrieval-augmented generation → RAG pipeline design and evaluation
Less is More: Improving LLM Reasoning with Minimal Test-Time InterventionLLM reasoning → Test-time compute and efficient thinking
FinDeepResearch: Evaluating Deep Research Agents in Rigorous Financial AnalysisLLM agents and tool use → Financial trading and forecasting agents
Big Reasoning with Small Models: Instruction Retrieval at Inference TimeLLM reasoning → Reasoning benchmarks and evaluation
An LLM-Powered AI Agent Framework for Holistic IoT Traffic InterpretationLLM agents and tool use → Agent threats, guardrails and governance
FACTS: Table Summarization via Offline Template Generation with Agentic WorkflowsLLM agents and tool use → Web, search, and research agents
Optimal Aggregation of LLM and PRM Signals for Efficient Test-Time ScalingLLM reasoning → Reinforcement learning for reasoning +1
Readability ne Learnability: Rethinking the Role of Simplicity in Training Small Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Synthesizing Agentic Data for Web Agents with Progressive Difficulty Enhancement MechanismsLLM agents and tool use → Web, search, and research agents
RAGCap-Bench: Benchmarking Capabilities of LLMs in Agentic Retrieval Augmented Generation SystemsRetrieval-augmented generation → Agentic RAG, agents and memory +1
Knowledge Reasoning Language Model: Unifying Knowledge and Language for Inductive Knowledge Graph ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning
PhysMaster: Mastering Physical Representation for Video Generation via Reinforcement LearningWorld models → Latent predictive world models +2
VisCoP: Visual Probing for Video Domain Adaptation of Vision Language ModelsLLM reasoning → Multimodal reasoning +1
Generative Universal Verifier as Multimodal Meta-ReasonerLLM reasoning → Multimodal reasoning +1
Reasoning in Space via Grounding in the WorldLLM reasoning → Spatial and embodied reasoning
BRIEF-Pro: Universal Context Compression with Short-to-Long Synthesis for Fast and Accurate Multi-Hop ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning +2
Breadcrumbs Reasoning: Memory-Efficient Reasoning with Compression BeaconsLLM reasoning → Test-time compute and efficient thinking +1
Bee: A High-Quality Corpus and Full-Stack Suite to Unlock Advanced Fully Open MLLMsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
MimicKit: A Reinforcement Learning Framework for Motion Imitation and ControlDeep reinforcement learning methods → RL for robot locomotion and manipulation
NoisePrints: Distortion-Free Watermarks for Authorship in Private Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects +1
Provably Invincible Adversarial Attacks on Reinforcement Learning Systems: A Rate-Distortion Information-Theoretic ApproachDeep reinforcement learning methods → Adversarial robustness and RL in finance
InternVLA-M1: A Spatially Guided Vision-Language-Action Framework for Generalist Robot PolicyVision-language-action models → 3D and spatial-aware VLA
PriorGuide: Test-Time Prior Adaptation for Simulation-Based InferenceDiffusion models → Diffusion foundations and sampling
RECODE: Reasoning Through Code Generation for Visual Question AnsweringLLM reasoning → Multimodal reasoning +1
Confidence-Based Response Abstinence: Improving LLM Trustworthiness via Activation-Based Uncertainty EstimationRetrieval-augmented generation → RAG robustness and adaptive retrieval
Asymptotically optimal reinforcement learning in Block Markov Decision ProcessesDeep reinforcement learning methods → RL theory and sample complexity
Hard2Verify: A Step-Level Verification Benchmark for Open-Ended Frontier MathLLM reasoning → Mathematical reasoning
From Refusal to Recovery: A Control-Theoretic Approach to Generative AI GuardrailsLLM safety and security → Prompt injection and agent security
NExT-OMNI: Towards Any-to-Any Omnimodal Foundation Models with Discrete Flow MatchingDiffusion models → Image generation, text-to-image and editing
Training LLM Agents to Empower HumansLLM agents and tool use → General LLM agent frameworks and benchmarks
Simplicial Embeddings Improve Sample Efficiency in Actor-Critic AgentsDeep reinforcement learning methods → RL theory and sample complexity
Attention Misses Visual Risk: Risk-Adaptive Steering for Multimodal Safety AlignmentLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
Information-Theoretic Reward Modeling for Stable RLHF: Detecting and Mitigating Reward HackingLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Generating healthy counterfactuals with denoising diffusion bridge modelsDiffusion models → Medical imaging and segmentation
FlashWorld: High-quality 3D Scene Generation within SecondsDiffusion models → 3D, point cloud and scene generation
Rethinking Evaluation in the Era of Time Series Foundation Models: (Un)known Information Leakage ChallengesLLM reasoning → Time series reasoning
What is the objective of reasoning with reinforcement learning?LLM reasoning → Reinforcement learning for reasoning
LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action ModelsVision-language-action models → Failure recovery and robustness
Manifold Decoders: A Framework for Generative Modeling from Nonlinear EmbeddingsDiffusion models → Diffusion foundations and sampling
MemoTime: Memory-Augmented Temporal Knowledge Graph Enhanced Large Language Model ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning
NOSA: Native and Offloadable Sparse AttentionEfficient LLM inference and serving → KV cache optimization
RAG Meets Temporal Graphs: Time-Sensitive Modeling and Retrieval for Evolving KnowledgeRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Steer-MoE: Efficient Audio-Language Alignment with a Mixture-of-Experts Steering ModuleMixture-of-experts → Multimodal MoE models
Attention Illuminates LLM Reasoning: The Preplan-and-Anchor Rhythm Enables Fine-Grained Policy OptimizationLLM reasoning → Test-time compute and efficient thinking
In-Browser LLM-Guided Fuzzing for Real-Time Prompt Injection Testing in Agentic AI BrowsersLLM agents and tool use → Agent threats, guardrails and governance +1
Offline and Online KL-Regularized RLHF under Differential PrivacyLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Confidence as a Reward: Transforming LLMs into Reward ModelsLLM reasoning → Reinforcement learning for reasoning +1
ConsintBench: Evaluating Language Models on Real-World Consumer Intent UnderstandingLLM reasoning → Reasoning benchmarks and evaluation
DistilCLIP-EEG: Enhancing Epileptic Seizure Detection Through Multi-modal Learning and Knowledge DistillationKnowledge distillation → Distillation for medical imaging
ExpressNet-MoE: A Hybrid Deep Neural Network for Emotion RecognitionMixture-of-experts → Multimodal MoE models
Bridge the Gap: Enhancing Quadruped Locomotion with Vertical Ground PerturbationsDeep reinforcement learning methods → RL for robot locomotion and manipulation
Who Speaks for the Trigger? Dynamic Expert Routing in Backdoored Mixture-of-Experts TransformersLLM safety and security → Mixture-of-experts routing safety
Toward Efficient Inference Attacks: Shadow Model Sharing via Mixture-of-ExpertsLLM safety and security → Membership inference attacks
Steerable Conditional Diffusion for Domain Adaptation in PET Image ReconstructionDiffusion models → Medical imaging and segmentation
M2PO: Multi-Perspective Multi-Pair Preference Optimization for Machine TranslationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Reinforcement Learning Meets Masked Generative Models: Mask-GRPO for Text-to-Image GenerationLLM reasoning → Reinforcement learning for reasoning
Assessing LLM Reasoning Through Implicit Causal Chain Discovery in Climate DiscourseLLM reasoning → Causal reasoning in LLMs
Doing Things with Words: Rethinking Theory of Mind Simulation in Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Spatial-DISE: A Unified Benchmark for Evaluating Spatial Reasoning in Vision-Language ModelsLLM reasoning → Spatial and embodied reasoning +1
DepthVLA: Enhancing Vision-Language-Action Models with Depth-Aware Spatial ReasoningLLM reasoning → Spatial and embodied reasoning +1
MADREC: A Multi-Aspect Driven LLM Agent for Explainable and Adaptive RecommendationLLM agents and tool use → Agentic recommender systems
A New Perspective on Transformers in Online Reinforcement Learning for Continuous ControlDeep reinforcement learning methods → Deep RL algorithms and representations
Language as a Label: Zero-Shot Multimodal Classification of Everyday Postures under Data ScarcityMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
D-SMART: Enhancing LLM Dialogue Consistency via Dynamic Structured Memory And Reasoning TreeLLM reasoning → Reasoning for agents, planning and code
Improving Visual Recommendation on E-commerce Platforms Using Vision-Language ModelsMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Adversarial Fine-tuning in Offline-to-Online Reinforcement Learning for Robust Robot ControlOffline RL and imitation learning from data → Offline RL algorithms
Protect: Towards Robust Guardrailing Stack for Trustworthy Enterprise LLM SystemsLLM safety and security → Safety alignment, refusal and guardrails
UniMoE-Audio: Unified Speech and Music Generation with Dynamic-Capacity MoEMixture-of-experts → Multimodal MoE models
Taming the Fragility of KV Cache Eviction in LLM InferenceEfficient LLM inference and serving → KV cache optimization
Embedding-Based Context-Aware RerankerRetrieval-augmented generation → Generation and context integration
Tactile-Conditioned Diffusion Policy for Force-Aware Robotic ManipulationOffline RL and imitation learning from data → Robot manipulation imitation learning
Self-Aug: Query and Entropy Adaptive Decoding for Large Vision-Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency
ChatR1: Reinforcement Learning for Conversational Reasoning and Retrieval Augmented Question AnsweringLLM reasoning → Knowledge- and tool-augmented reasoning +1
Km-scale dynamical downscaling through conformalized latent diffusion modelsDiffusion models → Flow matching, time series and science
In-Distribution Steering: Balancing Control and Coherence in Language Model GenerationMechanistic interpretability → Activation steering methods

All 407