Papers

New on arXiv

Thu 18 Jun 2026

480 papers

The Age of AI Agents Demands A New Scientific Paradigm To Sustain Trustworthy ScienceLLM agents and tool use → Web, search, and research agents
Masked Distillation: Internalizing the Chain-of-Thought in Language ModelsLLM reasoning → Prompting and chain-of-thought reasoning +1
ExecuGraph: A Multi-Agent, Execution-Grounded Framework for Reliable Backend Code Synthesis with Large Language ModelsLLM agents and tool use → Software engineering coding agents
Hidden in Thought: Transferable Chain-of-Thought Artifacts Induce Harmful BehaviorLLM reasoning → Reasoning for agents, planning and code +1
Position: Explainability Research Must Prioritize Foundations over Ad-hoc MethodsMechanistic interpretability → Circuits and mechanistic analysis
Robusto-2: Benchmarking Humans & VLMs for Autonomous Driving in Lima & New York CityEnd-to-end autonomous driving → VLMs for autonomous driving
How Should Agents Read Demonstrations? Hierarchical Structure Beats Flat Action LogsLLM agents and tool use → Agent skill learning and evolution
UNITY: Attention Flow Networks for Adaptive Conditioning in DiffusionDiffusion models → Image generation, text-to-image and editing
Is Our Benchmark Enough? An Analysis of Continual Learning for MLLMsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Learning What Not to Forget: Long-Horizon Agent Memory from a Few Kilobytes of LearningLLM agents and tool use → Agent memory systems
Grouped Query Experts: Mixture-of-Experts on GQA Self-AttentionMixture-of-experts → Expert routing strategies
ELDiff: When Evidential Learning Meets Text-to-Image DiffusionDiffusion models → Image generation, text-to-image and editing
Think Twice Before You Act: Protecting LLM Agents Against Tool Description Poisoning via Isolated PlanningLLM agents and tool use → Agent threats, guardrails and governance +1
PROTON: Prototype-Based Test-Time Online OOD Detection for Medical VLMsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Whose Agent Are You? Multi-Layer Fingerprinting and Attribution of Autonomous Web AgentsLLM agents and tool use → Agent threats, guardrails and governance
Vesta: A Generalist Embodied Reasoning ModelLLM reasoning → Spatial and embodied reasoning
Reinforcement Learning for Risk-Sensitive Investment Management: a Free Energy--Entropy Duality ApproachDeep reinforcement learning methods → Adversarial robustness and RL in finance
When Do Intrinsic Rewards Work for Code Reasoning? A Comprehensive StudyLLM reasoning → Reinforcement learning for reasoning
Adversarial observations in probabilistic State-Space Models for robust Reinforcement LearningDeep reinforcement learning methods → Adversarial robustness and RL in finance
SciLens: Multi-modal Scientific Claim Verification with Agentic Entailment and GroundingLLM agents and tool use → Web, search, and research agents
Geometric Entropy: When Trajectory Diversity Helps and Hurts in Imitation LearningOffline RL and imitation learning from data → Robot manipulation imitation learning
FOCA: Future-Oriented Conditioning for Data-Efficient Vision-Language-Action AdaptationVision-language-action models → Reinforcement learning and post-training +1
Evolutionary Discovery of Developmental Reward Schedules in Deep Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
SignVLA: Real-Time Sign Language-Guided Robotic Manipulation via Attention LSTM and Vision-Language-Action ModelsVision-language-action models → Generalist VLA models and benchmarks
Translating Inference-Time Control to Radiology Vision-Language Models: Activation Steering for Pneumonia Classification on Chest X-raysMultimodal LLMs / vision-language models → Medical vision-language models +1
Process-Reward Tactic Evolution for Long-Horizon Bioinformatics WorkflowsLLM agents and tool use → Medical and scientific domain agents
ReLaTS: a Reinforcement Learning-based method for dynamically determining the coupling Time Step in multi-scale simulations of self-gravitating systemsDeep reinforcement learning methods → RL for robot navigation and UAV control
B[FM]2: Brain Foundation Model via Flow Matching with SplitUNetDiffusion models → Medical imaging and segmentation
Fara-1.5: Scalable Learning Environments for Computer Use AgentsLLM agents and tool use → GUI and mobile agents
World Action Models: A SurveyWorld models → Embodied and robot world models
Beyond 'One Language, One Script': Quantifying Orthographic Bias in Multilingual VLMs with PuMVRMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
One Image is All You Need: Agentic One-Shot Image Generation via Text-Based World Models for Long-Tail Spatial PerceptionWorld models → World models for autonomous driving +2
From Sparse X-rays to 3D CT: Training-Free Reconstruction with Diffusion PriorsDiffusion models → Medical imaging and segmentation
Integrating Large Language Model Agents with Digital Twins for Industrial Autonomous SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
Perturbation-Based Epistemic Uncertainty for Failure Detection in Vision-Language-Action ModelsVision-language-action models → Failure recovery and robustness
MemoryWAM: Efficient World Action Modeling with Persistent MemoryWorld models → Embodied and robot world models
TimeProVe: Propose, then Verify for Efficient Long Video Temporal Reasoning in Activities of Daily LivingLLM reasoning → Multimodal reasoning
How Transparent is DiffusionGemma?LLM reasoning → Diffusion reasoning and test-time scaling
From Efficiency to Leakage -- Privacy Backdoor in Federated Language Model Fine-TuningLLM safety and security → Federated unlearning and poisoning
Current World Models Lack a Persistent State CoreWorld models → Video-generation world models and simulators
Toward Calibrated Mixture-of-Experts Under Distribution ShiftMixture-of-experts → MoE theory and gating functions
SSD: Spatially Speculative Decoding Accelerates Autoregressive Image GenerationEfficient LLM inference and serving → Fast autoregressive visual generation
Execution-State Capsules: Graph-Bound Execution-State Checkpoint and Restore for Low-Latency, Small-Batch, On-Device Physical-AI ServingEfficient LLM inference and serving → Serving systems and request scheduling
How Do Instructions Shape Speech? Cross-Attention Attribution for Style-Captioned Text-to-SpeechDiffusion models → Audio, speech and music
LedgerAgent: Structured State for Policy-Adherent Tool-Calling AgentsLLM agents and tool use → Agent threats, guardrails and governance
StylisticBias: A Few Human Visual Cues Drive Most Social Biases in MLLMsMultimodal LLMs / vision-language models → Social bias and fairness in VLMs
Sovereign Execution Broker: Enforcing Certificate-Bound Authority in Agentic Control PlanesLLM agents and tool use → Agent threats, guardrails and governance
FlowEdit: Associative Memory for Lifelong Pronunciation Adaptation in Flow-Matching TTSDiffusion models → Audio, speech and music
S-Agent: Spatial Tool-Use Elicits Reasoning for Spatial IntelligenceLLM agents and tool use → Multimodal and embodied agents
Probe-and-Refine Tuning of Repository Guidance for Coding AgentsLLM agents and tool use → Software engineering coding agents
What Do Safety-Aligned LLMs Learn From Mixed Compliance Demonstrations?LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
Contagion Networks: Evaluator Preference Propagation in Multi-Agent LLM SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
Your Mouse and Eyes Secretly Leak Your Preference: LLM Alignment using Implicit Feedback from UsersLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
GroundControl: Anticipating Navigation Failures in Vision-Language Agents via Trajectory-Consistent Uncertainty EstimatesLLM agents and tool use → Multimodal and embodied agents
Scalable Training of Spatially Grounded 2D Vision-Language Models for RadiologyMultimodal LLMs / vision-language models → Medical vision-language models
Analyzing Defensive Misdirection Against Model-Guided Automated Attacks on Agentic AI SystemsLLM agents and tool use → Agent threats, guardrails and governance +1
Agentic Symbolic Search: Characterizing PDEs Beyond Hand-crafted Expressions, Meshes, and Neural NetworksLLM agents and tool use → Web, search, and research agents
Repurposing a Speech Classifier for Guided Diffusion-Based Speech GenerationDiffusion models → Audio, speech and music
Spectral Query-Key Product Weight Steering for Training-Free VLM Hallucination MitigationMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
On the Redundancy of Timestep Embeddings in Diffusion ModelsDiffusion models → Diffusion foundations and sampling
Direct Advantage Estimation for Scalable and Sample-efficient Deep Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
NRT-Bench: Benchmarking Multi-Turn Red-Teaming of LLM Operator Agents in Safety-Critical Control RoomsLLM agents and tool use → Agent threats, guardrails and governance +1
The Significance of Style Diversity in Annotation-Free Synthetic Data GenerationSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Agentic AutoResearch forSpace Autonomy: An Auditable, LLM-Driven Research Agent for Aerospace Control ProblemsLLM agents and tool use → Web, search, and research agents
Rethinking Shrinkage Bias in LLM FP4 Pretraining: Geometric Origin, Systemic Impact, and UFP4 RecipeLLM reasoning → Test-time compute and efficient thinking +1
CRAX: Fast Safe Reinforcement Learning BenchmarkingDeep reinforcement learning methods → Safe and constrained RL
AutoPass: Evidence-Guided LLM Agents for Compiler Performance TuningLLM agents and tool use → Software engineering coding agents
A Model-Driven Approach for Developing Families of Reinforcement Learning EnvironmentsDeep reinforcement learning methods → Deep RL algorithms and representations
Through the PRISM: Preference Representation in Intermediate States of Video Diffusion ModelsDiffusion models → Video generation +1
CMDS-AD: Cross-Modal Dual-Stream Decoupling for Few-Shot Anomaly DetectionDiffusion models → Flow matching, time series and science
Co-VLA: Coordination-Aware Structured Action Modeling for Dual-Arm Vision-Language-Action SystemsVision-language-action models → Human-video and latent-action pretraining
U2Mamba: A Two-level Nested U-structure Mamba for Salient Object DetectionState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
ELVA: Exploring Ranking-Driven Universal Multimodal RetrievalMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Lagrange: An Open-Vocabulary, Energy-Based Sparse Framework for Generalized End-to-End DrivingEnd-to-end autonomous driving → Reasoning VLA and VLM driving policies
Transcript-Free Flow-Matching Text-to-Speech via Speech Feature ConditioningDiffusion models → Audio, speech and music
Finetuning Vision-Language-Action Models Requires Fewer Layers Than You ThinkVision-language-action models → Efficient VLA inference and adaptation
SPOT-E: Test-Time Entropy Shaping with Visual Spotlights for Frozen VLMsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Phoenix: Safe GitHub Issue Resolution via Multi-Agent LLMsLLM agents and tool use → Software engineering coding agents
ScholarQuest: A Taxonomy-Guided Benchmark for Agentic Academic Paper Search in Open Literature EnvironmentsLLM agents and tool use → Web, search, and research agents
QMFOL: Benchmarking Large Language Model Reasoning via Quantifiable Monadic First-Order Logic Test Case GenerationLLM reasoning → Reasoning benchmarks and evaluation
Augmenting Game AI with Deep Reinforcement LearningDeep reinforcement learning methods → RL for games and combinatorial optimization
FlowMaps: Modeling Long-Term Multimodal Object Dynamics with Flow MatchingDiffusion models → Diffusion and flow policies for robotics
Off-Policy Evaluation for Missingness-Aware Policies in MDPs with Rewards Missing Not at RandomOffline RL and imitation learning from data → Offline RL algorithms +1
Stable Transformer-Actor-Critic Model Predictive Control: A Contraction Analysis ApproachDeep reinforcement learning methods → RL theory and sample complexity
Evaluating and Enhancing Negation Comprehension in Remote Sensing MLLMsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
MedRLM: Recursive Multimodal Health Intelligence for Long-Context Clinical Reasoning, Sensor-Guided Screening, Evidence-Grounded Decision Support, and Community-to-Tertiary Referral OptimizationLLM reasoning → Clinical and biomedical applications
N-Version Programming with Coding AgentsLLM agents and tool use → Software engineering coding agents
From Texts to Scores: Tracing the Emergence of Essay Quality Representations in Large Language ModelsMechanistic interpretability → SAE features and their applications
Frequency-Aware Flow Matching for Continuous and Consistent Robotic Action GenerationDiffusion models → Diffusion and flow policies for robotics
SAM3 Self-Distillation for Fine-Grained GOOSE 2D Semantic SegmentationKnowledge distillation → Detection and dense prediction distillation
Dual-Agent Framework for Cross-Model Verified Translation of Natural-Language Protocols into Robotic Laboratory PlatformLLM agents and tool use → Web, search, and research agents
Pose6DAug: Physically Plausible Multi-view Object Swapping for Robot Data AugmentationVision-language-action models → 3D and spatial-aware VLA
When Does Streaming Tool Use Help? Characterizing Tool-Intent Stabilization in Streaming Retrieval-Augmented GenerationRetrieval-augmented generation → RAG pipeline design and evaluation
Pixel-Level Residual Diffusion Transformer: Scalable 3D CT Volume GenerationDiffusion models → Medical imaging and segmentation
FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion ModelDiffusion models → Diffusion and flow policies for robotics
Quantile of Means: A Bonus-Free Ensemble Method for Minimax Optimal Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
Sensorimotor World Models: Perception for Action via Inverse DynamicsWorld models → Latent predictive world models
RFM-Editing 2: Text-Guided Audio Editing with Rectified Flow Matching and Coarse-to-Fine Diffusion TransformersDiffusion models → Audio, speech and music
Site-Specific MIMO Channel Generation via Diffusion and Flow Matching: Fidelity, Efficiency, and Downstream UtilityDiffusion models → Diffusion for compression and wireless
MakeupMirror: Improving Facial Attribute Preservation in Diffusion Models for Makeup TransferDiffusion models → Image generation, text-to-image and editing

All 480