Papers

New on arXiv

Mon 9 Jun 2025

422 papers

FingerTip 20K: A Benchmark for Proactive and Personalized Mobile LLM AgentsLLM agents and tool use → GUI and mobile agents
Memory Allocation in Resource-Constrained Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
A Digital Twin Framework for Generation-IV Reactors with Reinforcement Learning-Enabled Health-Aware Supervisory ControlDeep reinforcement learning methods → RL for energy, networks and operations
Knowledge Compression via Question Generation: Enhancing Multihop Document Retrieval without Fine-tuningRetrieval-augmented generation → Generation and context integration
LLMs Caught in the Crossfire: Malware Requests and Jailbreak ChallengesLLM safety and security → Jailbreaking
HiBerNAC: Hierarchical Brain-inspired Robotic Neural Agent Collective for Disentangling Complex ManipulationVision-language-action models → Reasoning and chain-of-thought in VLA
From Passive to Active Reasoning: Can Large Language Models Ask the Right Questions under Incomplete Information?LLM reasoning → Reasoning benchmarks and evaluation
From Debate to Equilibrium: Belief-Driven Multi-Agent LLM Reasoning via Bayesian Nash EquilibriumLLM reasoning → Reasoning for agents, planning and code +1
Serendipitous Recommendation with Multimodal LLMMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Task-conditioned probing of instruction-tuned multimodal LLMs: Region-specific brain alignment patterns under naturalistic stimuliMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
LEANN: A Low-Storage Vector IndexRetrieval-augmented generation → Vector search and ANN indexes
Reinforcement Learning from Human Feedback with High-Confidence Safety ConstraintsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
Temporalizing Confidence: Evaluation of Chain-of-Thought Reasoning with Signal Temporal LogicLLM reasoning → Test-time compute and efficient thinking
Can AI Validate Science? Benchmarking LLMs for Accurate Scientific Claim rightarrow Evidence ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning
GIQ: Benchmarking 3D Geometric Reasoning of Vision Foundation Models with Simulated and Real PolyhedraLLM reasoning → Spatial and embodied reasoning +1
Interpreting Agent Behaviors in Reinforcement-Learning-Based Cyber-Battle Simulation PlatformsDeep reinforcement learning methods → Adversarial robustness and RL in finance
Open World Scene Graph Generation using Vision Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Agentic Surgical AI: Surgeon Style Fingerprinting and Privacy Risk Quantification via Discrete Diffusion in a Vision-Language-Action FrameworkDiffusion models → Diffusion and flow policies for robotics +1
Worst-Case Symbolic Constraints Analysis and Generalisation with Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
BLUR: A Bi-Level Optimization Approach for LLM UnlearningLLM safety and security → Machine unlearning
Ego-centric Learning of Communicative World Models for Autonomous DrivingWorld models → World models for autonomous driving +2
EconWebArena: Benchmarking Autonomous Agents on Economic Tasks in Realistic Web EnvironmentsLLM agents and tool use → Web, search, and research agents
Not All Tokens Matter: Towards Efficient LLM Reasoning via Token Significance in Reinforcement LearningLLM reasoning → Test-time compute and efficient thinking
QA-LIGN: Aligning LLMs through Constitutionally Decomposed QALLM alignment / RLHF / preference optimisation → LLM preference optimization methods
SOP-Bench: Complex Industrial SOPs for Evaluating LLM AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
Towards AI-assisted Neutrino Flavor Theory DesignDeep reinforcement learning methods → RL for energy, networks and operations
Hierarchical Lexical Graph for Enhanced Multi-Hop RetrievalRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
STAMImputer: Spatio-Temporal Attention MoE for Traffic Data ImputationMixture-of-experts → MoE for time series and graphs
ReCogDrive: A Reinforced Cognitive Framework for End-to-End Autonomous DrivingMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection +1
StableMTL: Repurposing Latent Diffusion Models for Multi-Task Learning from Partially Annotated Synthetic DatasetsDiffusion models → Diffusion foundations and sampling
GUI-Reflection: Empowering Multimodal GUI Models with Self-Reflection BehaviorMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Play to Generalize: Learning to Reason Through Game PlayLLM reasoning → Multimodal reasoning +1
Self Forcing: Bridging the Train-Test Gap in Autoregressive Video DiffusionDiffusion models → Video generation
Dreamland: Controllable World Creation with Simulator and Generative ModelsWorld models → Video-generation world models and simulators
Dynamic View Synthesis as an Inverse ProblemDiffusion models → 3D, point cloud and scene generation
MADFormer: Mixed Autoregressive and Diffusion Transformers for Continuous Image GenerationDiffusion models → Image generation, text-to-image and editing
Supporting Construction Worker Well-Being with a Multi-Agent Conversational AI SystemLLM agents and tool use → Medical and scientific domain agents
Rethinking Cross-Modal Interaction in Multimodal Diffusion TransformersDiffusion models → Image generation, text-to-image and editing
Beyond Top Activations: Efficient and Reliable Crowdsourced Evaluation of Automated InterpretabilityMechanistic interpretability → Circuits and mechanistic analysis
Thinking vs. Doing: Agents that Reason by Scaling Test-Time InteractionLLM reasoning → Reasoning for agents, planning and code
HeuriGym: An Agentic Benchmark for LLM-Crafted Heuristics in Combinatorial OptimizationLLM agents and tool use → General LLM agent frameworks and benchmarks
CyberV: Cybernetics for Test-time Scaling in Video UnderstandingLLM reasoning → Multimodal reasoning +1
SpaCE-10: A Comprehensive Benchmark for Multimodal Large Language Models in Compositional Spatial IntelligenceMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
SUDER: Self-Improving Unified Large Multimodal Models for Understanding and Generation with Dual Self-RewardsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +2
BridgeVLA: Input-Output Alignment for Efficient 3D Manipulation Learning with Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
Decoupling the Image Perception and Multimodal Reasoning for Reasoning Segmentation with Digital Twin RepresentationsLLM reasoning → Multimodal reasoning
Mimicking or Reasoning: Rethinking Multi-Modal In-Context Learning in Vision-Language ModelsLLM reasoning → Multimodal reasoning +1
A Generative Physics-Informed Reinforcement Learning-Based Approach for Construction of Representative Drive CycleDeep reinforcement learning methods → RL for autonomous driving and traffic
Solving Inequality Proofs with Large Language ModelsLLM reasoning → Mathematical reasoning
WeThink: Toward General-purpose Vision-Language Reasoning via Reinforcement LearningLLM reasoning → Multimodal reasoning
Diffuse Everything: Multimodal Diffusion Models on Arbitrary State SpacesDiffusion models → Image generation, text-to-image and editing
FunDiff: Diffusion Models over Function Spaces for Physics-Informed Generative ModelingDiffusion models → Flow matching, time series and science
Video Unlearning via Low-Rank Refusal VectorLLM safety and security → Vision-language and text-to-image safety
Diffusion Counterfactual Generation with Semantic AbductionDiffusion models → Diffusion foundations and sampling
Generative Resource Allocation for 6G O-RAN with Diffusion PoliciesDiffusion models → Diffusion for compression and wireless
Versatile Loco-Manipulation through Flexible Interlimb CoordinationDeep reinforcement learning methods → RL for robot locomotion and manipulation
Learning to Focus: Causal Attention Distillation via Gradient-Guided Token PruningEfficient LLM inference and serving → KV cache optimization
SAM2Auto: Auto Annotation Using FLASHMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection
Diffusion models under low-noise regimeDiffusion models → Diffusion foundations and sampling
HAIBU-ReMUD: Reasoning Multimodal Ultrasound Dataset and Model Bridging to General Specific DomainsLLM reasoning → Clinical and biomedical applications +1
Accelerating Diffusion Planners in Offline RL via Reward-Aware Consistency Trajectory DistillationDiffusion models → Diffusion and flow policies for robotics +1
Guideline Forest: Retrieval-Augmented Reasoning with Branching Experience-Induced GuidelinesLLM reasoning → Knowledge- and tool-augmented reasoning +1
WebUIBench: A Comprehensive Benchmark for Evaluating Multimodal Large Language Models in WebUI-to-CodeMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
M2Restore: Mixture-of-Experts-based Mamba-CNN Fusion Framework for All-in-One Image RestorationMixture-of-experts → Vision mixture-of-experts +1
Self-Cascaded Diffusion Models for Arbitrary-Scale Image Super-ResolutionDiffusion models → Inverse problems and image restoration
Image Reconstruction as a Tool for Feature AnalysisMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
LLM Unlearning Should Be Form-IndependentLLM safety and security → Machine unlearning
Re-ranking Reasoning Context with Tree Search Makes Large Vision-Language Models StrongerLLM reasoning → Multimodal reasoning +2
A Neurosymbolic Agent System for Compositional Visual ReasoningLLM reasoning → Multimodal reasoning +1
AbstRaL: Augmenting LLMs' Reasoning by Reinforcing Abstract ThinkingLLM reasoning → Mathematical reasoning
Difference Inversion: Interpolate and Isolate the Difference with Token Consistency for Image Analogy GenerationDiffusion models → Image generation, text-to-image and editing
Graph-Assisted Stitching for Offline Hierarchical Reinforcement LearningOffline RL and imitation learning from data → Offline RL algorithms
RSafe: Incentivizing proactive reasoning to build robust and adaptive LLM safeguardsLLM reasoning → Reasoning for agents, planning and code +2
Consistent Video Editing as Flow-Driven Image-to-Video GenerationDiffusion models → Video generation
Through the Valley: Path to Effective Long CoT Training for Small Language ModelsLLM reasoning → Prompting and chain-of-thought reasoning
Evaluating Robustness in Latent Diffusion Models via Embedding Level AugmentationDiffusion models → Diffusion foundations and sampling
NOVA3D: Normal Aligned Video Diffusion Model for Single Image to 3D GenerationDiffusion models → 3D, point cloud and scene generation
Breaking the Block: Preserving Data Continuity to Train Superior SAEs for Instruct ModelsMechanistic interpretability → SAE features and their applications
QUITE: A Query Rewrite System Beyond Rules with LLM AgentsLLM agents and tool use → Web, search, and research agents
MCPWorld: A Unified Benchmarking Testbed for API, GUI, and Hybrid Computer Use AgentsLLM agents and tool use → GUI and mobile agents
ProARD: progressive adversarial robustness distillation: provide wide range of robust studentsKnowledge distillation → Federated, robust and private distillation
Synthesis by Design: Controlled Data Generation via Structural GuidanceLLM reasoning → Mathematical reasoning
FMaMIL: Frequency-Driven Mamba Multi-Instance Learning for Weakly Supervised Lesion Segmentation in Medical ImagesState-space sequence models (S4, Mamba) → Mamba for medical image analysis
Fast ECoT: Efficient Embodied Chain-of-Thought via Thoughts ReuseLLM reasoning → Spatial and embodied reasoning +1
SWE-Dev: Building Software Engineering Agents with Training and Inference ScalingLLM agents and tool use → Software engineering coding agents
SongBloom: Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion RefinementDiffusion models → Audio, speech and music
Unblocking Fine-Grained Evaluation of Detailed Captions: An Explaining AutoRater and Critic-and-Revise PipelineMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Event-Priori-Based Vision-Language Model for Efficient Visual UnderstandingMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
SceneRAG: Scene-level Retrieval-Augmented Generation for Video UnderstandingRetrieval-augmented generation → Multimodal RAG
TwinBreak: Jailbreaking LLM Security Alignments based on Twin PromptsLLM safety and security → Jailbreaking
Explore the vulnerability of black-box models via diffusion modelsDiffusion models → Robustness, privacy and adversarial aspects
Super Encoding Network: Recursive Association of Multi-Modal Encoders for Video UnderstandingWorld models → Video-generation world models and simulators
Uncertainty-o: One Model-agnostic Framework for Unveiling Uncertainty in Large Multimodal ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
OptiScene: LLM-driven Indoor Scene Layout Generation via Scaled Human-aligned Data Synthesis and Multi-Stage Preference OptimizationSynthetic data and self-improvement for LLMs → Self-improving MLLMs and synthetic data
SAFEFLOW: A Principled Protocol for Trustworthy and Transactional Autonomous Agent SystemsLLM agents and tool use → Agent threats, guardrails and governance
MoE-MLoRA for Multi-Domain CTR Prediction: Efficient Adaptation with Expert SpecializationMixture-of-experts → MoE for recommender systems
GTR-CoT: Graph Traversal as Visual Chain of Thought for Molecular Structure RecognitionLLM reasoning → Clinical and biomedical applications
CheMatAgent: Enhancing LLMs for Chemistry and Materials Science through Tree-Search Based Tool LearningLLM agents and tool use → Medical and scientific domain agents
MoQAE: Mixed-Precision Quantization for Long-Context LLM Inference via Mixture of Quantization-Aware ExpertsMixture-of-experts → MoE large language models
BitVLA: 1-bit Vision-Language-Action Models for Robotics ManipulationVision-language-action models → Efficient VLA inference and adaptation

All 422