Papers

New on arXiv

Sun 1 Mar 2026

246 papers

Mixture of Sequence: Theme-Aware Mixture-of-Experts for Long-Sequence RecommendationMixture-of-experts → MoE for recommender systems
Enhancing Safety of Large Language Models via Embedding Space SeparationLLM safety and security → Safety alignment, refusal and guardrails
ICPRL: Acquiring Physical Intuition from Interactive ControlWorld models → Latent predictive world models
Zipage: Maintain High Request Concurrency for LLM Reasoning through Compressed PagedAttentionLLM reasoning → Test-time compute and efficient thinking +1
GameVerse: Can Vision-Language Models Learn from Video-based Reflection?Multimodal LLMs / vision-language models → Embodied and 3D vision-language models
ProFocus: Proactive Perception and Focused Reasoning in Vision-and-Language NavigationLLM reasoning → Spatial and embodied reasoning
Contextual Invertible World Models: A Neuro-Symbolic Agentic Framework for Colorectal Cancer Drug ResponseWorld models → Medical and surgical world models +1
Characterizing VLA Models: Identifying the Action Generation Bottleneck for Edge AI ArchitecturesVision-language-action models → Efficient VLA inference and adaptation
SWE-Adept: An LLM-Based Agentic Framework for Deep Codebase Analysis and Structured Issue ResolutionLLM agents and tool use → Software engineering coding agents
Truth as a Trajectory: What Internal Representations Reveal About Large Language Model ReasoningLLM reasoning → Reasoning benchmarks and evaluation +2
PAC Guarantees for Reinforcement Learning: Sample Complexity, Coverage, and StructureDeep reinforcement learning methods → RL theory and sample complexity
AG-VAS: Anchor-Guided Zero-Shot Visual Anomaly Segmentation with Large Multimodal ModelsMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Hybrid TD3: Overestimation Bias Analysis and Stable Policy Optimization for Hybrid Action SpaceDeep reinforcement learning methods → RL theory and sample complexity
When Does RL Help Medical VLMs? Disentangling Vision, SFT, and RL GainsLLM reasoning → Clinical and biomedical applications +1
I Can't Believe It's Not Robust: Catastrophic Collapse of Safety Classifiers under Embedding DriftLLM reasoning → Reasoning for agents, planning and code
Theoretical Perspectives on Data Quality and Synergistic Effects in Pre- and Post-Training Reasoning ModelsLLM reasoning → Reinforcement learning for reasoning
Integrating LTL Constraints into PPO for Safe Reinforcement LearningDeep reinforcement learning methods → Safe and constrained RL
JailNewsBench: Multi-Lingual and Regional Benchmark for Fake News Generation under Jailbreak AttacksLLM safety and security → Jailbreaking
Individual Turing Test: A Case Study of LLM-based Simulation Using Longitudinal Personal DataRetrieval-augmented generation → RAG pipeline design and evaluation
Efficient Extractive Summarization with MAMBA-Transformer Hybrids for Low-Resource ScenariosState-space sequence models (S4, Mamba) → Core SSM architectures for language
Attention Smoothing Is All You Need For UnlearningLLM safety and security → Machine unlearning
The Informational Cost of Agency: A Bounded Measure of Interaction Efficiency for Deployed Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
MOSAIC: A Universal Agent-Level Interface for Cross-Paradigm Agent Mixing and Human-AI CollaborationLLM agents and tool use → Reinforcement learning for LLM agents
Cross-Modal Guidance for Fast Diffusion-Based Computed TomographyDiffusion models → Medical imaging and segmentation
Defensive Refusal Bias: How Safety Alignment Fails Cyber DefendersLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
TARSE: Test-Time Adaptation via Retrieval of Skills and Experience for Reasoning AgentsLLM reasoning → Clinical and biomedical applications +1
AgilePruner: An Empirical Study of Attention and Diversity for Adaptive Visual Token Pruning in Large Vision-Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Towards Policy-Adaptive Image Guardrail: Benchmark and MethodLLM safety and security → Safety alignment, refusal and guardrails
Can Thinking Models Think to Detect Hateful Memes?LLM reasoning → Reasoning benchmarks and evaluation
Monocular 3D Object Position Estimation with VLMs for Human-Robot InteractionMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Learn Hard Problems During RL with Reference Guided Fine-tuningLLM reasoning → Reinforcement learning for reasoning
Epistemic Gain, Aleatoric Cost: Uncertainty Decomposition in Multi-Agent Debate for Math ReasoningLLM reasoning → Reasoning for agents, planning and code +1
Reasoning Boosts Opinion Alignment in LLMsLLM reasoning → Reasoning benchmarks and evaluation
Can AI Agents Agree?LLM agents and tool use → Social simulation and multi-agent debate
Agents Learn Their Runtime: Interpreter Persistence as Training-Time SemanticsLLM agents and tool use → Software engineering coding agents
Subliminal Signals in Preference LabelsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
VisNec: Measuring and Leveraging Visual Necessity for Multimodal Instruction TuningLLM reasoning → Multimodal reasoning
Reasoning or Rationalization? The Role of Justifications in Masked Diffusion Models for Fact VerificationLLM reasoning → Diffusion reasoning and test-time scaling
HAVEN: High-Bandwidth Flash Augmented Vector Engine for Large-Scale Approximate Nearest-Neighbor Search AccelerationRetrieval-augmented generation → RAG pipeline design and evaluation
VP-Hype: A Hybrid Mamba-Transformer Framework with Visual-Textual Prompting for Hyperspectral Image ClassificationState-space sequence models (S4, Mamba) → Hyperspectral image analysis with Mamba
FREE-Edit: Using Editing-aware Injection in Rectified Flow Models for Zero-shot Image-Driven Video EditingDiffusion models → Video generation
BeautyGRPO: Aesthetic Alignment for Face Retouching via Dynamic Path Guidance and Fine-Grained Preference ModelingLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Demystifying Group Relative Policy Optimization: Its Policy Gradient is a U-StatisticLLM reasoning → Reinforcement learning for reasoning +1
Semantic XPath: Structured Agentic Memory Access for Conversational AILLM agents and tool use → Agent memory systems
DeepResearch-9K: A Challenging Benchmark Dataset of Deep-Research AgentLLM agents and tool use → Web, search, and research agents
AutoSkill: Experience-Driven Lifelong Learning via Skill Self-EvolutionLLM agents and tool use → Agent memory systems
TC-SSA: Token Compression via Semantic Slot Aggregation for Gigapixel Pathology ReasoningLLM reasoning → Clinical and biomedical applications +2
MedCollab: IBIS-Guided Multi-Agent Collaboration with Hierarchical Disease Relation Chains for Clinical DiagnosisLLM agents and tool use → Medical and scientific domain agents
Predictive Reasoning with Augmented Anomaly Contrastive Learning for Compositional Visual RelationsLLM reasoning → Multimodal reasoning
ClinCoT: Clinical-Aware Visual Chain-of-Thought for Medical Vision Language ModelsLLM reasoning → Clinical and biomedical applications +1
HVR-Met: A Hypothesis-Verification-Replanning Agentic System for Extreme Weather DiagnosisLLM agents and tool use → Financial trading and forecasting agents
From Dialogue to Execution: Mixture-of-Agents Assisted Interactive Planning for Behavior Tree-Based Long-Horizon Robot ExecutionLLM agents and tool use → General LLM agent frameworks and benchmarks
DeAR: Fine-Grained VLM Adaptation by Decomposing Attention Head RolesMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Compact Task-Aligned Imitation Learning for Laboratory AutomationOffline RL and imitation learning from data → Robot manipulation imitation learning
DIVA-GRPO: Enhancing Multimodal Reasoning through Difficulty-Adaptive Variant AdvantageLLM reasoning → Reinforcement learning for reasoning +1
Egocentric Co-Pilot: Web-Native Smart-Glasses Agents for Assistive Egocentric AILLM agents and tool use → Multimodal and embodied agents
Data-Efficient Brushstroke Generation with Diffusion Models for Oil PaintingDiffusion models → Image generation, text-to-image and editing
Unified Vision-Language Modeling via Concept Space AlignmentMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
CARD: Towards Conditional Design of Multi-agent Topological StructuresLLM agents and tool use → General LLM agent frameworks and benchmarks
Can Vision Language Models Assess Graphic Design Aesthetics? A Benchmark, Evaluation, and Dataset PerspectiveMultimodal LLMs / vision-language models → Image editing and quality assessment
Beyond Global Similarity: Towards Fine-Grained, Multi-Condition Multimodal RetrievalMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Flow Matching-enabled Test-Time Refinement for Unsupervised Cardiac MR RegistrationDiffusion models → Medical imaging and segmentation
How RL Unlocks the Aha Moment in Geometric Interleaved ReasoningLLM reasoning → Spatial and embodied reasoning
LLaDA-o: An Effective and Length-Adaptive Omni Diffusion ModelDiffusion models → Image generation, text-to-image and editing
Unleashing VLA Potentials in Autonomous Driving via Explicit Learning from FailuresVision-language-action models → VLA for autonomous driving +1
GroupGPT: A Token-efficient and Privacy-preserving Agentic Framework for Multi-User Chat AssistantLLM agents and tool use → Agent threats, guardrails and governance
TriMoE: Augmenting GPU with AMX-Enabled CPU and DIMM-NDP for High-Throughput MoE Inference via OffloadingMixture-of-experts → Efficient MoE training and inference systems
MM-DeepResearch: A Simple and Effective Multimodal Agentic Search BaselineLLM agents and tool use → Web, search, and research agents
Silo-Bench: A Scalable Environment for Evaluating Distributed Coordination in Multi-Agent LLM SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
From Intuition to Investigation: A Tool-Augmented Reasoning MLLM Framework for Generalizable Face Anti-SpoofingLLM reasoning → Multimodal reasoning
One-Token Verification for Reasoning Correctness EstimationLLM reasoning → Test-time compute and efficient thinking
An Open-Source Modular Benchmark for Diffusion-Based Motion Planning in Closed-Loop Autonomous DrivingEnd-to-end autonomous driving → Imitation-learning end-to-end planners
BadRSSD: Backdoor Attacks on Regularized Self-Supervised Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects +1
FastCode: Fast and Cost-Efficient Code Understanding and ReasoningLLM reasoning → Reasoning for agents, planning and code
GeodesicNVS: Probability Density Geodesic Flow Matching for Novel View SynthesisDiffusion models → Flow matching, time series and science
AG-REPA: Causal Layer Selection for Representation Alignment in Audio Flow MatchingDiffusion models → Audio, speech and music
CollabEval: Enhancing LLM-as-a-Judge via Multi-Agent CollaborationLLM agents and tool use → General LLM agent frameworks and benchmarks
Compensation-free Machine Unlearning in Text-to-Image Diffusion Models by Eliminating the Mutual InformationDiffusion models → Robustness, privacy and adversarial aspects +1
Tracking Capabilities for Safer AgentsLLM safety and security → Prompt injection and agent security
Event-Anchored Frame Selection for Effective Long-Video UnderstandingMultimodal LLMs / vision-language models → Video understanding with MLLMs
EraseAnything++: Enabling Concept Erasure in Rectified Flow Transformers Leveraging Multi-Object OptimizationDiffusion models → Robustness, privacy and adversarial aspects
HiMAC: Hierarchical Macro-Micro Learning for Long-Horizon LLM AgentsLLM agents and tool use → Reinforcement learning for LLM agents
Forgetting is Competition: Rethinking Unlearning as Representation Interference in Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects +1
Intent-Context Synergy Reinforcement Learning for Autonomous UAV Decision-Making in Air CombatDeep reinforcement learning methods → RL for robot navigation and UAV control
AWE: Adaptive Agents for Dynamic Web Penetration TestingLLM agents and tool use → Agent threats, guardrails and governance
HierKick: Hierarchical Reinforcement Learning for Vision-Guided Soccer Robot ControlDeep reinforcement learning methods → RL for robot locomotion and manipulation
DAM-VLA: A Dynamic Action Model-Based Vision-Language-Action Framework for Robot ManipulationVision-language-action models → 3D and spatial-aware VLA
The Aftermath of DrawEduMath: Vision Language Models Underperform with Struggling Students and Misdiagnose ErrorsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
DriveCode: Domain Specific Numerical Encoding for LLM-Based Autonomous DrivingEnd-to-end autonomous driving → Large language models for driving
Improving Text-to-Image Generation with Intrinsic Self-Confidence RewardsLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
pySpatial: Generating 3D Visual Programs for Zero-Shot Spatial ReasoningLLM reasoning → Spatial and embodied reasoning +1
Principled Fast and Meta Knowledge Learners for Continual Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Clawdrain: Exploiting Tool-Calling Chains for Stealthy Token Exhaustion in OpenClaw AgentsLLM agents and tool use → Agent threats, guardrails and governance
CHIMERA: Compact Synthetic Data for Generalizable LLM ReasoningLLM reasoning → Reasoning benchmarks and evaluation +1
VEMamba: Efficient Isotropic Reconstruction of Volume Electron Microscopy with Axial-Lateral Consistent MambaState-space sequence models (S4, Mamba) → Mamba for medical image analysis
Active Flow MatchingDiffusion models → Flow matching, time series and science
BioProAgent: Neuro-Symbolic Grounding for Constrained Scientific PlanningLLM reasoning → Spatial and embodied reasoning
MC-Search: Evaluating and Enhancing Multimodal Agentic Search with Structured Long Reasoning ChainsLLM reasoning → Knowledge- and tool-augmented reasoning +3
PPC-MT: Parallel Point Cloud Completion with Mamba-Transformer Hybrid ArchitectureState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
Tiny-Critic RAG: Empowering Agentic Fallback with Parameter-Efficient Small Language ModelsRetrieval-augmented generation → Agentic RAG, agents and memory +1

All 246