Papers

New on arXiv

Tue 24 Mar 2026

435 papers

Reciprocal Co-Training (RCT): Coupling Gradient-Based and Non-Differentiable Models via Reinforcement LearningDeep reinforcement learning methods → RL training of language and agent models
Hierarchical Retrieval Augmented Generation for Adversarial Technique Annotation in Cyber Threat Intelligence TextRetrieval-augmented generation → Privacy and security of RAG
StepCache: Step-Level Reuse with Lightweight Verification and Selective Patching for LLM ServingEfficient LLM inference and serving → KV cache optimization
Learning to Select Visual In-Context DemonstrationsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
LogicDiff: Logic-Guided Denoising Improves Zero-Shot Reasoning in Masked Diffusion Language ModelsLLM reasoning → Diffusion reasoning and test-time scaling +1
Quantized Vision-Language Models for Damage Assessment: A Comparative Study of LLaVA-1.5-7B Quantization LevelsMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Edge Reliability Gap in Vision-Language Models: Quantifying Failure Modes of Compressed VLMs Under Visual CorruptionMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Aesthetic Assessment of Chinese Handwritings Based on Vision Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
A training-free framework for high-fidelity appearance transfer via diffusion transformersDiffusion models → Image generation, text-to-image and editing
Bitboard version of Tetris AIDeep reinforcement learning methods → RL for games and combinatorial optimization
ARC-AGI-3: A New Challenge for Frontier Agentic IntelligenceLLM agents and tool use → General LLM agent frameworks and benchmarks
Sparse Autoencoders for Interpretable Medical Image Representation LearningMechanistic interpretability → Interpreting biomedical and science models
Manifold Generalization Provably Proceeds Memorization in Diffusion ModelsDiffusion models → Diffusion foundations and sampling
The Cognitive Firewall:Securing Browser Based AI Agents Against Indirect Prompt Injection Via Hybrid Edge Cloud DefenseLLM agents and tool use → Agent threats, guardrails and governance +1
Lightweight Fairness for LLM-Based Recommendations via Kernelized Projection and Gated AdaptersMixture-of-experts → MoE for recommender systems
Self Paced Gaussian Contextual Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Efficient Benchmarking of AI AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
BXRL: Behavior-Explainable Reinforcement LearningDeep reinforcement learning methods → Interpretable and logic-guided RL
The Diminishing Returns of Early-Exit Decoding in Modern LLMsEfficient LLM inference and serving → Early exit and edge inference
Utilizing Adversarial Training for Robust Voltage Control: An Adaptive Deep Reinforcement Learning MethodDeep reinforcement learning methods → RL for energy, networks and operations
Can LLM Agents Be CFOs? Benchmarking Long-Horizon Resource Allocation in an Uncertain Enterprise EnvironmentLLM agents and tool use → Financial trading and forecasting agents
A Theory of LLM Information SusceptibilityLLM agents and tool use → General LLM agent frameworks and benchmarks
Environment Maps: Structured Environmental Representations for Long-Horizon AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
Reasoning models do not yet follow their reasoning in autonomous driving: The KITScenes LongTail DatasetLLM reasoning → Chain-of-thought reasoning for driving
The Geometric Price of Discrete Logic: Context-driven Manifold Dynamics of Number RepresentationsLLM reasoning → Reasoning benchmarks and evaluation
Safe Reinforcement Learning with Preference-based Constraint InferenceDeep reinforcement learning methods → Safe and constrained RL
MedObvious: Exposing the Medical Moravec's Paradox in VLMs via Clinical TriageMultimodal LLMs / vision-language models → Medical vision-language models
UniGRPO: Unified Policy Optimization for Reasoning-Driven Visual GenerationDiffusion models → Diffusion and flow policies for robotics
DA-Flow: Degradation-Aware Optical Flow Estimation with Diffusion ModelsDiffusion models → Inverse problems and image restoration
WildWorld: A Large-Scale Dataset for Dynamic World Modeling with Actions and Explicit State toward Generative ARPGWorld models → Video-generation world models and simulators +1
VISion On Request: Enhanced VLLM efficiency with sparse, dynamically selected, vision-language interactionsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Foveated Diffusion: Efficient Spatially Adaptive Image and Video GenerationDiffusion models → Video generation
AgentRVOS: Reasoning over Object Tracks for Zero-Shot Referring Video Object SegmentationLLM reasoning → Multimodal reasoning
TETO: Tracking Events with Teacher Observation for Motion Estimation and Frame InterpolationKnowledge distillation → Detection and dense prediction distillation
SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and PlanningLLM reasoning → Reasoning for agents, planning and code +2
VTAM: Video-Tactile-Action Models for Complex Physical Interaction Beyond VLAsWorld models → Embodied and robot world models
ConceptCoder: Improve Code Reasoning via Concept LearningLLM reasoning → Reasoning for agents, planning and code
End-to-End Efficient RL for Linear Bellman Complete MDPs with Deterministic TransitionsDeep reinforcement learning methods → RL theory and sample complexity
DetPO: In-Context Learning with Multi-Modal LLMs for Few-Shot Object DetectionMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection
Code Review Agent BenchmarkLLM agents and tool use → Software engineering coding agents
Similarity-Aware Mixture-of-Experts for Data-Efficient Continual LearningMixture-of-experts → Mixture of LoRA experts
SortedRL: Accelerating RL Training for LLMs through Online Length-Aware SchedulingLLM reasoning → Reinforcement learning for reasoning
Beyond Preset Identities: How Agents Form Stances and Boundaries in Generative SocietiesLLM agents and tool use → Social simulation and multi-agent debate
Unleashing Spatial Reasoning in Multimodal Large Language Models via Textual Representation Guided ReasoningLLM reasoning → Spatial and embodied reasoning +1
Graph Energy Matching: Transport-Aligned Energy-Based Modeling for Graph GenerationDiffusion models → Flow matching, time series and science
ABot-PhysWorld: Interactive World Foundation Model for Robotic Manipulation with Physics AlignmentWorld models → Embodied and robot world models
Off-Policy Value-Based Reinforcement Learning for Large Language ModelsLLM reasoning → Reinforcement learning for reasoning
ViBe: Ultra-High-Resolution Video Synthesis Born from Pure ImagesDiffusion models → Video generation
ARGENT: Adaptive Hierarchical Image-Text RepresentationsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Steering LLMs for Culturally Localized GenerationMechanistic interpretability → Knowledge and representations inside LLMs
Designing Agentic AI-Based Screening for Portfolio InvestmentLLM agents and tool use → Financial trading and forecasting agents
Mamba-driven MRI-to-CT Synthesis for MRI-only Radiotherapy PlanningState-space sequence models (S4, Mamba) → Mamba for medical image analysis
A Multimodal Framework for Human-Multi-Agent InteractionLLM agents and tool use → Multimodal and embodied agents
Not All Tokens Are Created Equal: Query-Efficient Jailbreak Fuzzing for LLMsLLM safety and security → Jailbreaking
SafeSeek: Universal Attribution of Safety Circuits in Language ModelsLLM safety and security → Safety alignment, refusal and guardrails +1
Permutation-Symmetrized Diffusion for Unconditional Molecular GenerationDiffusion models → Flow matching, time series and science
GO-Renderer: Generative Object Rendering with 3D-aware Controllable Video Diffusion ModelsDiffusion models → 3D, point cloud and scene generation
Neural ODE and SDE Models for Adaptation and Planning in Model-Based Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
MemCollab: Cross-Model Memory Collaboration via Contrastive Trajectory DistillationLLM agents and tool use → Agent memory systems
GEM: Guided Expectation-Maximization for Behavior-Normalized Candidate Action Selection in Offline RLOffline RL and imitation learning from data → Offline RL algorithms
PERMA: Benchmarking Personalized Memory Agents via Event-Driven Preference and Realistic Task EnvironmentsLLM agents and tool use → Agent memory systems +1
I Came, I Saw, I Explained: Benchmarking Multimodal LLMs on Figurative Meaning in MemesMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Efficient Hybrid SE(3)-Equivariant Visuomotor Flow Policy via Spherical Harmonics for Robot ManipulationDiffusion models → Diffusion and flow policies for robotics
Gaze-Regularized Vision-Language-Action Models for Robotic ManipulationVision-language-action models → Generalist VLA models and benchmarks
ImplicitRM: Unbiased Reward Modeling from Implicit Preference Data for LLM alignmentLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Reasoning over Semantic IDs Enhances Generative RecommendationLLM reasoning → LLM reasoning for recommendation
Path Planning and Reinforcement Learning-Driven Control of On-Orbit Free-Flying Multi-Arm RobotsDeep reinforcement learning methods → RL for robot navigation and UAV control
Gimbal360: Canonicalizing Planar Diffusion for Spherical Panorama CompletionDiffusion models → 3D, point cloud and scene generation
SAiW: Source-Attributable Invisible Watermarking for Proactive Deepfake DefenseLLM safety and security → Watermarking generated content and models
Adaptively Robust LLM Monitoring via Activation WatermarkingLLM safety and security → Watermarking generated content and models
Describe-Then-Act: Proactive Agent Steering via Distilled Language-Action World ModelsWorld models → Embodied and robot world models +1
SMSP: A Plug-and-Play Strategy of Multi-Scale Perception for MLLMs to Perceive Visual IllusionsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
TRAP: Hijacking VLA CoT-Reasoning via Adversarial PatchesLLM reasoning → Spatial and embodied reasoning +1
AgentFoX: LLM Agent-Guided Fusion with eXplainability for AI-Generated Image DetectionLLM agents and tool use → Multimodal and embodied agents
When Language Models Lose Their Mind: The Consequences of Brain MisalignmentLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
When Models Learn to Ask Why: Adaptive Causal Reasoning for Trustworthy Medical Vision-Language ModelsLLM reasoning → Clinical and biomedical applications +1
MLLM-HWSI: A Multimodal Large Language Model for Hierarchical Whole Slide Image UnderstandingMultimodal LLMs / vision-language models → Medical vision-language models
PCR: A Prefetch-Enhanced Cache Reuse System for Low-Latency RAG ServingRetrieval-augmented generation → Generation and context integration +1
Parametric Knowledge in RAG-SFT for Domain-Specific Document GenerationRetrieval-augmented generation → RAG pipeline design and evaluation
YOLO with Kolmogorov-Arnold networks and vision-language foundation models for interpretable object detection with trustworthy multimodal AI in computer vision perceptionMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection
Traffic Sign Recognition in Autonomous Driving: Dataset, Benchmark, and Field ExperimentEnd-to-end autonomous driving → VLMs for autonomous driving
Cog3DMap: Multi-View Vision-Language Reasoning with 3D Cognitive MapsLLM reasoning → Spatial and embodied reasoning +1
Zero-Shot Personalization of Objects via Textual InversionDiffusion models → Image generation, text-to-image and editing
AgentRAE: Remote Action Execution through Notification-based Visual Backdoors against Screenshots-based Mobile GUI AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Can Large Language Models Reason and Optimize Under Constraints?LLM reasoning → Test-time compute and efficient thinking
VQ-Jarvis: Retrieval-Augmented Video Restoration Agent with Sharp Vision and Fast ThoughtRetrieval-augmented generation → Multimodal RAG
A Critical Review on the Effectiveness and Privacy Threats of Membership Inference AttacksLLM safety and security → Membership inference attacks
Markov-Enforced Discrete Diffusion Model for Digital Semantic Symbol Error CorrectionDiffusion models → Diffusion for compression and wireless
Asymptotic Learning Curves for Diffusion Models with Random Features Score and Manifold DataDiffusion models → Diffusion foundations and sampling
From Morality Installation in LLMs to LLMs in Morality-as-a-SystemLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
PersonalQ: Select, Quantize, and Serve Personalized Diffusion Models for Efficient InferenceDiffusion models → Image generation, text-to-image and editing
Optimizing Small Language Models for NL2SQL via Chain-of-Thought Fine-TuningLLM reasoning → Knowledge- and tool-augmented reasoning
ProGRank: Probe-Gradient Reranking to Defend Dense-Retriever RAG from Corpus PoisoningRetrieval-augmented generation → Privacy and security of RAG
SoK: The Attack Surface of Agentic AI - Tools and AutonomyRetrieval-augmented generation → Privacy and security of RAG +2
Quality Over Clicks: Iterative Reinforcement Learning for Early-Stage E-Commerce Query SuggestionDeep reinforcement learning methods → RL training of language and agent models
EVA: Efficient Reinforcement Learning for End-to-End Video AgentMultimodal LLMs / vision-language models → Video understanding with MLLMs +2
IntentWeave: A Progressive Entry Ladder for Multi-Surface Browser Agents in Cloud PortalsLLM agents and tool use → Web, search, and research agents
ForestPrune: High-ratio Visual Token Compression for Video Multimodal Large Language Models via Spatial-Temporal Forest ModelingMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
EchoKV: Efficient KV Cache Compression via Similarity-Based ReconstructionEfficient LLM inference and serving → KV cache optimization
Separating Diagnosis from Control: Auditable Policy Adaptation in Agent-Based Simulations with LLM-Based DiagnosticsLLM agents and tool use → Medical and scientific domain agents

All 435