Papers

New on arXiv

Wed 17 Sep 2025

323 papers

LLM on a Budget: Active Knowledge Distillation for Efficient Classification of Large Text CorporaKnowledge distillation → Knowledge distillation for LLMs
Recognizing internal states in AI: evidence from patterned preferences in large language modelsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Baseer: A Vision-Language Model for Arabic Document-to-Markdown OCRMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
PiERN: Token-Level Routing for Integrating High-Precision Computation and ReasoningLLM reasoning → Test-time compute and efficient thinking
SIRAG: Towards Stable and Interpretable RAG with A Process-Supervised Multi-Agent FrameworkRetrieval-augmented generation → Agentic RAG, agents and memory +1
Thinking in a Crowd: How Auxiliary Information Shapes LLM ReasoningLLM reasoning → Reasoning benchmarks and evaluation
ZERA: Zero-init Instruction Evolving Refinement Agent -- From Zero Instructions to Structured Prompts via Principle-based OptimizationLLM agents and tool use → General LLM agent frameworks and benchmarks
Synthetic bootstrapped pretrainingSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Multi-Modal Interpretability for Enhanced Localization in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
ViSpec: Accelerating Vision-Language Models with Vision-Aware Speculative DecodingMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Video2Roleplay: A Multimodal Dataset and Framework for Video-Guided Role-playing AgentsLLM agents and tool use → Social simulation and multi-agent debate
An LLM-based multi-agent framework for agile effort estimationLLM agents and tool use → Software engineering coding agents
Process-Supervised Reinforcement Learning for Interactive Multimodal Tool-Use AgentsLLM agents and tool use → Reinforcement learning for LLM agents
Ticket-Bench: A Kickoff for Multilingual and Regionalized Agent EvaluationLLM agents and tool use → General LLM agent frameworks and benchmarks
Causal-Counterfactual RAG: The Integration of Causal-Counterfactual Reasoning into RAGLLM reasoning → Causal reasoning in LLMs +1
Value Alignment of Social Media Ranking AlgorithmsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Local-Canonicalization Equivariant Graph Neural Networks for Sample-Efficient and Generalizable Swarm Robot ControlDeep reinforcement learning methods → RL for robot navigation and UAV control
Detecting Pipeline Failures through Fine-Grained Analysis of Web AgentsLLM agents and tool use → Web, search, and research agents
CRAFT: Coaching Reinforcement Learning Autonomously using Foundation Models for Multi-Robot Coordination TasksDeep reinforcement learning methods → RL training of language and agent models
DreamControl: Human-Inspired Whole-Body Humanoid Control for Scene Interaction via Guided DiffusionDiffusion models → Diffusion and flow policies for robotics
LeVR: A Modular VR Teleoperation Framework for Imitation Learning in Dexterous ManipulationOffline RL and imitation learning from data → Robot manipulation imitation learning
Near-Real-Time Resource Slicing for QoS Optimization in 5G O-RAN using Deep Reinforcement LearningDeep reinforcement learning methods → RL for energy, networks and operations
FlowDrive: Energy Flow Field for End-to-End Autonomous DrivingEnd-to-end autonomous driving → End-to-end system architecture
D4PM: A Dual-branch Driven Denoising Diffusion Probabilistic Model with Joint Posterior Diffusion Sampling for EEG Artifacts RemovalDiffusion models → Medical imaging and segmentation
A Simple and Efficient Jailbreak Method Exploiting LLMs' HelpfulnessLLM safety and security → Jailbreaking
Aegis: Automated Error Generation and Attribution for Multi-Agent SystemsLLM agents and tool use → Software engineering coding agents
Cin'easte: A Fine-grained Contextual Movie Question Answering BenchmarkMultimodal LLMs / vision-language models → Video understanding with MLLMs
Defending Diffusion Models Against Membership Inference Attacks via Higher-Order Langevin DynamicsDiffusion models → Robustness, privacy and adversarial aspects +1
Synthesizing Behaviorally-Grounded Reasoning Chains: A Data-Generation Framework for Personal Finance LLMsLLM reasoning → Reasoning for agents, planning and code
TGPO: Tree-Guided Preference Optimization for Robust Web Agent Reinforcement LearningLLM agents and tool use → Web, search, and research agents +2
AssoCiAm: A Benchmark for Evaluating Association Thinking while Circumventing AmbiguityLLM reasoning → Reasoning benchmarks and evaluation +1
Where Do Tokens Go? Understanding Pruning Behaviors in STEP at High ResolutionsEfficient LLM inference and serving → Token pruning and merging for ViTs
Quantum Reinforcement Learning-Guided Diffusion Model for Image Synthesis via Hybrid Quantum-Classical Generative Model ArchitecturesDiffusion models → Diffusion foundations and sampling +1
MIMIC-D: Multi-modal Imitation for MultI-agent Coordination with Decentralized Diffusion PoliciesDiffusion models → Diffusion and flow policies for robotics
CLAW: A Vision-Language-Action Framework for Weight-Aware Robotic GraspingVision-language-action models → Tactile, force-aware and humanoid VLA
MARS2 2025 Challenge on Multimodal Reasoning: Datasets, Methods, Results, Discussion, and OutlookLLM reasoning → Multimodal reasoning
SeqVLA: Sequential Task Execution for Long-Horizon Manipulation with Completion-Aware Vision-Language-Action ModelVision-language-action models → Reasoning and chain-of-thought in VLA
GeoAware-VLA: Implicit Geometry Aware Vision-Language-Action ModelVision-language-action models → 3D and spatial-aware VLA
CSMoE: An Efficient Remote Sensing Foundation Model with Soft Mixture-of-ExpertsMixture-of-experts → Vision mixture-of-experts
Reasoning Efficiently Through Adaptive Chain-of-Thought Compression: A Self-Optimizing FrameworkLLM reasoning → Test-time compute and efficient thinking
FlightDiffusion: Revolutionising Autonomous Drone Training with Diffusion Models Generating FPV VideoDiffusion models → Diffusion and flow policies for robotics
Online Bayesian Risk-Averse Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
VSE-MOT: Multi-Object Tracking in Low-Quality Video Scenes Guided by Visual Semantic EnhancementMultimodal LLMs / vision-language models → Video understanding with MLLMs
Prompt2Auto: From Motion Prompt to Automated Control via Geometry-Invariant One-Shot Gaussian Process LearningOffline RL and imitation learning from data → Robot learning from demonstration
Enhancing Multi-Agent Debate System Performance via Confidence ExpressionLLM agents and tool use → Social simulation and multi-agent debate
SAIL-VL2 Technical ReportMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning +1
CrowdAgent: Multi-Agent Managed Multi-Source Annotation SystemLLM agents and tool use → General LLM agent frameworks and benchmarks
Early Stopping Chain-of-thoughts in Large Language ModelsLLM reasoning → Prompting and chain-of-thought reasoning
RFM-Editing: Rectified Flow Matching for Text-guided Audio EditingDiffusion models → Audio, speech and music
Slim-SC: Thought Pruning for Efficient Scaling with Self-ConsistencyLLM reasoning → Test-time compute and efficient thinking
LLM Agents for Interactive Workflow Provenance: Reference Architecture and Evaluation MethodologyLLM agents and tool use → Web, search, and research agents
SEG-Parking: Towards Safe, Efficient, and Generalizable Autonomous Parking via End-to-End Offline Reinforcement LearningOffline RL and imitation learning from data → Imitation and offline RL for driving
SHaRe-RL: Structured, Interactive Reinforcement Learning for Contact-Rich Industrial Assembly TasksDeep reinforcement learning methods → RL for robot locomotion and manipulation
Reinforcement Learning for Autonomous Point-to-Point UAV NavigationDeep reinforcement learning methods → RL for robot navigation and UAV control
An Empirical Study on Failures in Automated Issue SolvingLLM agents and tool use → Software engineering coding agents
Noise-Level Diffusion Guidance: Well Begun is Half DoneDiffusion models → Diffusion foundations and sampling
Large Language Model-Empowered Decision Transformer for UAV-Enabled Data CollectionOffline RL and imitation learning from data → Offline multi-agent RL
Linguistic Nepotism: Trading-off Quality for Language Preference in Multilingual RAGRetrieval-augmented generation → Generation and context integration
DSpAST: Disentangled Representations for Spatial Audio Reasoning with Large Language ModelsLLM reasoning → Audio-language model reasoning
MAP: End-to-End Autonomous Driving with Map-Assisted PlanningEnd-to-end autonomous driving → End-to-end system architecture
Towards Robust Defense against Customization via Protective Perturbation Resistant to Diffusion-based PurificationDiffusion models → Robustness, privacy and adversarial aspects
Towards Rationale-Answer Alignment of LVLMs via Self-Rationale CalibrationMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
PhysicalAgent: Towards General Cognitive Robotics with Foundation World ModelsWorld models → Embodied and robot world models +2
Synthetic Data Generation for Screen Time and App UsageSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Combining Evidence and Reasoning for Biomedical Fact-CheckingLLM reasoning → Clinical and biomedical applications
Do LLMs Align Human Values Regarding Social Biases? Judging and Explaining Social Biases with LLMsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Masked Diffusion Models as Energy MinimizationDiffusion models → Diffusion foundations and sampling
SpecDiff: Accelerating Diffusion Model Inference with Self-SpeculationDiffusion models → Diffusion foundations and sampling
Diving into Mitigating Hallucinations from a Vision Perspective for Large Vision-Language ModelsMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
Semi-MoE: Mixture-of-Experts meets Semi-Supervised Histopathology SegmentationMixture-of-experts → Medical and biomedical MoE
BWCache: Accelerating Video Diffusion Transformers through Block-Wise CachingDiffusion models → Video generation
CETUS: Causal Event-Driven Temporal Modeling With Unified Variable-Rate SchedulingState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
Who is Introducing the Failure? Automatically Attributing Failures of Multi-Agent Systems via Spectrum AnalysisLLM agents and tool use → Software engineering coding agents
Dual-Actor Fine-Tuning of VLA Models: A Talk-and-Tweak Human-in-the-Loop ApproachVision-language-action models → Reinforcement learning and post-training
Who Taught the Lie? Responsibility Attribution for Poisoned Knowledge in Retrieval-Augmented GenerationRetrieval-augmented generation → Privacy and security of RAG
AdaThinkDrive: Adaptive Thinking via Reinforcement Learning for Autonomous DrivingLLM reasoning → Chain-of-thought reasoning for driving +2
Generative Image Coding with Diffusion PriorDiffusion models → Diffusion for compression and wireless
THOR: Tool-Integrated Hierarchical Optimization via RL for Mathematical ReasoningLLM reasoning → Mathematical reasoning
Iterative Prompt Refinement for Safer Text-to-Image GenerationMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
A Study on Thinking Patterns of Large Reasoning Models in Code GenerationLLM reasoning → Reasoning for agents, planning and code
Controllable-Continuous Color Editing in Diffusion Model via Color MappingDiffusion models → Image generation, text-to-image and editing
Scrub It Out! Erasing Sensitive Memorization in Code Language Models via Machine UnlearningLLM safety and security → Machine unlearning
Dynamic Adaptive Legged Locomotion Policy via Decoupling Reaction Force Control and Gait ControlDeep reinforcement learning methods → RL for robot locomotion and manipulation
Reinforcement Learning for Robotic Insertion of Flexible Cables in Industrial SettingsDeep reinforcement learning methods → RL for robot locomotion and manipulation
DSPC: Dual-Stage Progressive Compression Framework for Efficient Long-Context ReasoningLLM reasoning → Test-time compute and efficient thinking
Inject, Fork, Compare: Defining an Interaction Vocabulary for Multi-Agent Simulation PlatformsLLM agents and tool use → Social simulation and multi-agent debate
StyleProtect: Safeguarding Artistic Identity in Fine-tuned Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
InfraMind: A Novel Exploration-based GUI Agentic Framework for Mission-critical Industrial ManagementLLM agents and tool use → GUI and mobile agents
Improving Context Fidelity via Native Retrieval-Augmented ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning +1
Re-purposing SAM into Efficient Visual Projectors for MLLM-Based Referring Image SegmentationMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
A High-Quality and Low-Complexity Streamable Neural Speech Codec with Knowledge DistillationKnowledge distillation → Speech and audio distillation
DREAM: Domain-aware Reasoning for Efficient Autonomous Underwater MonitoringLLM reasoning → Spatial and embodied reasoning
Is GPT-4o mini Blinded by its Own Safety Filters? Exposing the Multimodal-to-Unimodal Bottleneck in Hate Speech DetectionMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Softmax as a Lagrangian-Legendrian Seam
Words to Waves: Emotion-Adaptive Music Recommendation System
AquaVLM: Improving Underwater Situation Awareness with Mobile Vision Language Models
Risk-Calibrated Bayesian Streaming Intrusion Detection with SRE-Aligned Decisions
FTSCommDetector: Discovering Behavioral Communities through Temporal Synchronization
VLHSA: Vision-Language Hierarchical Semantic Alignment for Jigsaw Puzzle Solving with Eroded Gaps
Spiking Neural Networks for Mental Workload Classification with a Multimodal Approach

All 323