Papers

New on arXiv

Mon 30 Mar 2026

405 papers

Math Takes Two: A test for emergent mathematical reasoning in communicationLLM reasoning → Mathematical reasoning
Diffusion-Based Generative Priors for Efficient Beam Alignment in Directional NetworksDiffusion models → Diffusion for compression and wireless
FlowHijack: A Dynamics-Aware Backdoor Attack on Flow-Matching Vision-Language-Action ModelsVision-language-action models → Safety and adversarial attacks +2
FluidFlow: a flow-matching generative model for fluid dynamics surrogates on unstructured meshesDiffusion models → Flow matching, time series and science
CSAttention: Centroid-Scoring Attention for Accelerating LLM InferenceEfficient LLM inference and serving → KV cache optimization
Learning to Retrieve from Agent TrajectoriesLLM agents and tool use → Web, search, and research agents
From PDF to RAG-Ready: Evaluating Document Conversion Frameworks for Domain-Specific Question AnsweringRetrieval-augmented generation → RAG robustness and adaptive retrieval
Downscaling weather forecasts from Low- to High-Resolution with Diffusion ModelsDiffusion models → Flow matching, time series and science
Beyond Static Vision: Scene Dynamic Field Unlocks Intuitive Physics Understanding in Multi-modal Large Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Beyond Message Passing: A Semantic View of Agent Communication ProtocolsLLM agents and tool use → General LLM agent frameworks and benchmarks
Realistic Market Impact Modeling for Reinforcement Learning Trading EnvironmentsDeep reinforcement learning methods → Adversarial robustness and RL in finance
PAR2-RAG: Planned Active Retrieval and Reasoning for Multi-Hop Question AnsweringLLM reasoning → Knowledge- and tool-augmented reasoning
CivicShield: A Cross-Domain Defense-in-Depth Framework for Securing Government-Facing AI Chatbots Against Multi-Turn Adversarial AttacksLLM safety and security → Prompt injection and agent security
Trojan-Speak: Bypassing Constitutional Classifiers with No Jailbreak Tax via Adversarial FinetuningLLM safety and security → Jailbreaking
MMFace-DiT: A Dual-Stream Diffusion Transformer for High-Fidelity Multimodal Face GenerationDiffusion models → Image generation, text-to-image and editing
The Model Says Walk: How Surface Heuristics Override Implicit Constraints in LLM ReasoningLLM reasoning → Test-time compute and efficient thinking
Improving Efficiency of GPU Kernel Optimization Agents using a Domain-Specific Language and Speed-of-Light GuidanceLLM agents and tool use → Software engineering coding agents
Understand and Accelerate Memory Processing Pipeline for Large Language Model InferenceRetrieval-augmented generation → Generation and context integration
Design Principles for the Construction of a Benchmark Evaluating Security Operation Capabilities of Multi-agent AI SystemsLLM agents and tool use → Agent threats, guardrails and governance
An Explicit Surrogate for Gaussian Mixture Flow Matching with Wasserstein Gap BoundsDiffusion models → Flow matching, time series and science
Drop the Hierarchy and Roles: How Self-Organizing LLM Agents Outperform Designed StructuresLLM agents and tool use → General LLM agent frameworks and benchmarks
Mimosa Framework: Toward Evolving Multi-Agent Systems for Scientific ResearchLLM agents and tool use → Web, search, and research agents
Stepper: Stepwise Immersive Scene Generation with Multiview PanoramasWorld models → Video-generation world models and simulators
A Pontryagin Method of Model-based Reinforcement Learning via Hamiltonian Actor-CriticDeep reinforcement learning methods → RL theory and sample complexity
AutoWorld: Learning Multi-Agent Traffic Simulation with Self-Supervised World ModelsWorld models → World models for autonomous driving
Multi-Agent LLMs for Adaptive Acquisition in Bayesian OptimizationLLM agents and tool use → Reinforcement learning for LLM agents
Enhancing Policy Learning with World-Action ModelWorld models → Embodied and robot world models +1
ReproMIA: A Comprehensive Analysis of Model Reprogramming for Proactive Membership Inference AttacksDiffusion models → Robustness, privacy and adversarial aspects +1
Optimistic Online LQR via Intrinsic RewardsDeep reinforcement learning methods → RL for games and combinatorial optimization
CrossTrace: A Cross-Domain Dataset of Grounded Scientific Reasoning Traces for Hypothesis GenerationLLM reasoning → Knowledge- and tool-augmented reasoning
A Semantic Observer Layer for Autonomous Vehicles: Pre-Deployment Feasibility Study of VLMs for Low-Latency Anomaly DetectionEnd-to-end autonomous driving → VLMs for autonomous driving
OccSim: Multi-kilometer Simulation with Long-horizon Occupancy World ModelsWorld models → World models for autonomous driving
Gen-Searcher: Reinforcing Agentic Search for Image GenerationLLM agents and tool use → Multimodal and embodied agents
PoseDreamer: Scalable and Photorealistic Human Data Generation Pipeline with Diffusion ModelsDiffusion models → 3D, point cloud and scene generation
On-the-fly Repulsion in the Contextual Space for Rich Diversity in Diffusion TransformersDiffusion models → Image generation, text-to-image and editing
Stop Probing, Start Coding: Why Linear Probes and Sparse Autoencoders Fail at Compositional GeneralisationMechanistic interpretability → Sparse autoencoder training and evaluation
Rethinking Language Model Scaling under Transferable Hypersphere OptimizationLLM reasoning → Test-time compute and efficient thinking +1
FocusVLA: Focused Visual Utilization for Vision-Language-Action ModelsVision-language-action models → VLA architectures and generalization
SOLE-R1: Video-Language Reasoning as the Sole Reward for On-Robot Reinforcement LearningLLM reasoning → Spatial and embodied reasoning +1
Stepwise Credit Assignment for GRPO on Flow-Matching ModelsDiffusion models → Diffusion and flow policies for robotics
Dynamic Dual-Granularity Skill Bank for Agentic RLLLM agents and tool use → Agent skill learning and evolution
DreamLite: A Lightweight On-Device Unified Model for Image Generation and EditingDiffusion models → Image generation, text-to-image and editing
AdaptToken: Entropy-based Adaptive Token Selection for MLLM Long Video UnderstandingMultimodal LLMs / vision-language models → Video understanding with MLLMs
Functional Natural Policy GradientsDeep reinforcement learning methods → RL theory and sample complexity
AMIGO: Agentic Multi-Image Grounding Oracle BenchmarkMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning +1
Dynamic Lookahead Distance via Reinforcement Learning-Based Pure Pursuit for Autonomous RacingDeep reinforcement learning methods → RL for autonomous driving and traffic
Seeing with You: Perception-Reasoning Coevolution for Multimodal ReasoningLLM reasoning → Multimodal reasoning +1
ResAdapt: Adaptive Resolution for Efficient Multimodal ReasoningLLM reasoning → Multimodal reasoning +1
Optimistic Actor-Critic with Parametric Policies for Linear Markov Decision ProcessesDeep reinforcement learning methods → RL theory and sample complexity
MonitorBench: A Comprehensive Benchmark for Chain-of-Thought Monitorability in Large Language ModelsLLM reasoning → Prompting and chain-of-thought reasoning
Navigating the Mirage: A Dual-Path Agentic Framework for Robust Misleading Chart Question AnsweringMultimodal LLMs / vision-language models → Chart understanding and anomaly detection +1
Learning Partial Action Replacement in Offline MARLOffline RL and imitation learning from data → Offline multi-agent RL
Unrestrained Simplex Denoising for Discrete Data. A Non-Markovian Approach Applied to Graph GenerationDiffusion models → Flow matching, time series and science
CirrusBench: Evaluating LLM-based Agents Beyond Correctness in Real-World Cloud Service EnvironmentsLLM agents and tool use → General LLM agent frameworks and benchmarks
XSPA: Crafting Imperceptible X-Shaped Sparse Adversarial Perturbations for Transferable Attacks on VLMsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
StreamingVLA: Streaming Vision-Language-Action Model with Action Flow Matching and Adaptive Early ObservationVision-language-action models → Efficient VLA inference and adaptation +1
T-Norm Operators for EU AI Act Compliance Classification: An Empirical Comparison of Lukasiewicz, Product, and G"odel Semantics in a Neuro-Symbolic Reasoning SystemLLM reasoning → Reasoning benchmarks and evaluation
Domain-Invariant Prompt Learning for Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Hydra: Unifying Document Retrieval and Generation in a Single Vision-Language ModelMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
"What Did It Actually Do?": Understanding Risk Awareness and Traceability for Computer-Use AgentsLLM agents and tool use → Agent threats, guardrails and governance
Seen2Scene: Completing Realistic 3D Scenes with Visibility-Guided FlowDiffusion models → 3D, point cloud and scene generation
GEditBench v2: A Human-Aligned Benchmark for General Image EditingLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
ManipArena: Comprehensive Real-world Evaluation of Reasoning-Oriented Generalist Robot ManipulationWorld models → Embodied and robot world models +1
GraphWalker: Agentic Knowledge Graph Question Answering via Synthetic Trajectory CurriculumLLM agents and tool use → Web, search, and research agents
Intelligent Radio Resource Slicing for 6G In-Body SubnetworksDeep reinforcement learning methods → RL for energy, networks and operations
RAD-LAD: Rule and Language Grounded Autonomous Driving in Real-TimeEnd-to-end autonomous driving → Reasoning VLA and VLM driving policies
MRI-to-CT synthesis using drifting modelsDiffusion models → Medical imaging and segmentation
Video Generation Models as World Models: Efficient Paradigms, Architectures and AlgorithmsWorld models → Video-generation world models and simulators
Courtroom-Style Multi-Agent Debate with Progressive RAG and Role-Switching for Controversial Claim VerificationLLM agents and tool use → Social simulation and multi-agent debate
Tac2Real: Reliable and GPU Visuotactile Simulation for Online Reinforcement Learning and Zero-Shot Real-World DeploymentDeep reinforcement learning methods → RL for robot locomotion and manipulation
Rdm: Re-conceptualizing Distribution Matching as a Reward for Diffusion DistillationDiffusion models → Diffusion foundations and sampling +1
Entropic Claim Resolution: Uncertainty-Driven Evidence Selection for RAGRetrieval-augmented generation → RAG robustness and adaptive retrieval
LG-HCC: Local Geometry-Aware Hierarchical Context Compression for 3D Gaussian SplattingEfficient LLM inference and serving → Token pruning and merging for ViTs
IsoQuant: Hardware-Aligned SO(4) Isoclinic Rotations for LLM KV Cache CompressionEfficient LLM inference and serving → KV cache quantization
Synergy: A Next-Generation General-Purpose Agent for Open Agentic WebLLM agents and tool use → General LLM agent frameworks and benchmarks
Active Stereo-Camera Outperforms Multi-Sensor Setup in ACT Imitation Learning for Humanoid ManipulationOffline RL and imitation learning from data → Robot manipulation imitation learning
MiroEval: Benchmarking Multimodal Deep Research Agents in Process and OutcomeLLM agents and tool use → Web, search, and research agents
EdgeDiT: Hardware-Aware Diffusion Transformers for Efficient On-Device Image GenerationDiffusion models → Diffusion foundations and sampling
Within the MDT Room: Situated in Multidisciplinary Team-Grounded Agent Debate for Clinical DiagnosisLLM agents and tool use → Medical and scientific domain agents
Prompts Without Evidence: How Neuroimaging Mentions Shift Clinical Vision-Language Model PredictionsMultimodal LLMs / vision-language models → Medical vision-language models
Critic-Free Deep Reinforcement Learning for Maritime Coverage Path Planning on Irregular Hexagonal GridsDeep reinforcement learning methods → RL for robot navigation and UAV control
Membership Inference Attacks against Large Audio Language ModelsLLM safety and security → Membership inference attacks
Marco DeepResearch: Unlocking Efficient Deep Research Agents via Verification-Centric DesignLLM agents and tool use → Web, search, and research agents
Deep Research of Deep Research: From Transformer to Agent, From AI to AI for ScienceLLM reasoning → Knowledge- and tool-augmented reasoning +1
CoE: Collaborative Entropy for Uncertainty Quantification in Agentic Multi-LLM SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
VistaGEN: Consistent Driving Video Generation with Fine-Grained Control Using Multiview Visual-Language ReasoningEnd-to-end autonomous driving → Generative simulation for driving
A Multi-Agent Rhizomatic Pipeline for Non-Linear Literature AnalysisLLM agents and tool use → Web, search, and research agents
Integrating Multimodal Large Language Model Knowledge into Amodal CompletionMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
LIBERO-Para: A Diagnostic Benchmark and Metrics for Paraphrase Robustness in VLA ModelsVision-language-action models → Failure recovery and robustness
Competitor-aware Race Management for Electric Endurance RacingDeep reinforcement learning methods → RL for autonomous driving and traffic
Corruption-robust Offline Multi-agent Reinforcement Learning From Human FeedbackLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
DiffAttn: Diffusion-Based Drivers' Visual Attention Prediction with LLM-Enhanced Semantic ReasoningLLM reasoning → Chain-of-thought reasoning for driving
Reasoning as Energy Minimization over Structured Latent TrajectoriesLLM reasoning → Test-time compute and efficient thinking
Cost-Matching Model Predictive Control for Efficient Reinforcement Learning in Humanoid LocomotionDeep reinforcement learning methods → RL for robot locomotion and manipulation
Explaining CLIP Zero-shot Predictions Through ConceptsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
ERPO: Token-Level Entropy-Regulated Policy Optimization for Large Reasoning ModelsLLM reasoning → Reinforcement learning for reasoning
A Deep Reinforcement Learning Framework for Closed-loop Guidance of Fish Schools via Virtual AgentsDeep reinforcement learning methods → RL for robot navigation and UAV control
EpiPersona: Persona Projection and Episode Coupling for Pluralistic Preference ModelingLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
PReD: An LLM-based Foundation Multimodal Model for Electromagnetic Perception, Recognition, and DecisionLLM reasoning → Multimodal reasoning +1
Evaluating Privilege Usage of Agents with Real-World ToolsLLM agents and tool use → Agent threats, guardrails and governance

All 405