Papers

New on arXiv

Fri 17 Oct 2025

373 papers

Enhancing Retrieval-Augmented Generation with Two-Stage Retrieval: FlashRank Reranking and Query ExpansionRetrieval-augmented generation → Generation and context integration
MedBuild AI: An Agent-Based Hybrid Intelligence Framework for Reshaping Agency in Healthcare Infrastructure Planning through Generative Design for Medical ArchitectureLLM agents and tool use → Medical and scientific domain agents
Synera: Synergistic LLM Serving across Device and Cloud at ScaleEfficient LLM inference and serving → Serving systems and request scheduling
Repairing Tool Calls Using Post-tool Execution Reflection and RAGLLM agents and tool use → Tool learning and function calling
NEBULA: Do We Evaluate Vision-Language-Action Agents Correctly?Vision-language-action models → Failure recovery and robustness
Distractor Injection Attacks on Large Reasoning Models: Characterization and DefenseLLM reasoning → Reasoning for agents, planning and code
Towards Low-Resource Alignment to Diverse Perspectives with Sparse FeedbackLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Detecting Adversarial Fine-tuning with Auditing AgentsLLM agents and tool use → Agent threats, guardrails and governance
WEBSERV: A Full-Stack and RL-Ready Web Environment for Training Web Agents at ScaleLLM agents and tool use → Web, search, and research agents +1
VM-BeautyNet: A Synergistic Ensemble of Vision Transformer and Mamba for Facial Beauty PredictionState-space sequence models (S4, Mamba) → Vision Mamba backbones
SentinelNet: Safeguarding Multi-Agent Collaboration Through Credit-Based Dynamic Threat DetectionLLM agents and tool use → Agent threats, guardrails and governance
EgMM-Corpus: A Multimodal Vision-Language Dataset for Egyptian CultureMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Towards Automatic Evaluation and Selection of PHI De-identification Models via Multi-Agent CollaborationLLM agents and tool use → Medical and scientific domain agents
Human-Allied Relational Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Expressive Reward Synthesis with the Runtime Monitoring LanguageDeep reinforcement learning methods → Interpretable and logic-guided RL
The Formalism-Implementation Gap in Reinforcement Learning ResearchDeep reinforcement learning methods → Deep RL algorithms and representations
Alignment is Localized: A Causal Probe into Preference LayersLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
AsyncVoice Agent: Real-Time Explanation for LLM Planning and ReasoningLLM reasoning → Reasoning for agents, planning and code +1
Agentic AI for Ultra-Modern Networks: Multi-Agent Framework for RAN Autonomy and AssuranceLLM agents and tool use → Agentic AI for wireless networks
Expert Merging in Sparse Mixture of Experts with Nash BargainingMixture-of-experts → Expert routing strategies
GuideFlow3D: Optimization-Guided Rectified Flow For Appearance TransferDiffusion models → 3D, point cloud and scene generation
A Minimal-Assumption Analysis of Q-Learning with Time-Varying PoliciesDeep reinforcement learning methods → RL theory and sample complexity
Prompt injections as a tool for preserving identity in GAI image descriptionsLLM safety and security → Prompt injection and agent security
Zero-shot World Models via Search in MemoryWorld models → Latent dynamics models and model-based RL +1
The Hidden Cost of Modeling P(X): Vulnerability to Membership Inference Attacks in Generative Text ClassifiersLLM safety and security → Membership inference attacks
Reliability of Large Language Model Generated Clinical Reasoning in Assisted Reproductive Technology: Blinded Comparative Evaluation StudyLLM reasoning → Clinical and biomedical applications
BIOGEN: Evidence-Grounded Multi-Agent Reasoning Framework for Transcriptomic Interpretation in Antimicrobial ResistanceLLM reasoning → Clinical and biomedical applications +1
TriAgent: Automated Biomarker Discovery with Deep Research Grounding for Triage in Acute Care by LLM-Based Multi-Agent CollaborationLLM agents and tool use → Medical and scientific domain agents
EvolveR: Self-Evolving LLM Agents through an Experience-Driven LifecycleLLM agents and tool use → General LLM agent frameworks and benchmarks +1
BPL: Bias-adaptive Preference Distillation Learning for Recommender SystemKnowledge distillation → Distillation for recommender systems
Can LLMs Correct Themselves? A Benchmark of Self-Correction in LLMsLLM reasoning → Test-time compute and efficient thinking
SIADAFIX: issue description response for adaptive program repairLLM reasoning → Reasoning for agents, planning and code +1
BiomedXPro: Prompt Optimization for Explainable Diagnosis with Biomedical Vision Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
Rethinking the Design of Reinforcement Learning-Based Deep Research AgentsLLM reasoning → Knowledge- and tool-augmented reasoning +2
InfiMed-ORBIT: Aligning LLMs on Open-Ended Complex Tasks via Rubric-Based Incremental TrainingLLM alignment / RLHF / preference optimisation → Medical LLM alignment
BLIP3o-NEXT: Next Frontier of Native Image GenerationDiffusion models → Image generation, text-to-image and editing
Neuro-Symbolic Spatial Reasoning in SegmentationLLM reasoning → Spatial and embodied reasoning
Learning Correlated Reward Models: Statistical Barriers and OpportunitiesLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
ReCon: Region-Controllable Data Augmentation with Rectification and Alignment for Object DetectionDiffusion models → Image generation, text-to-image and editing
Demo: Guide-RAG: Evidence-Driven Corpus Curation for Retrieval-Augmented Generation in Long COVIDRetrieval-augmented generation → Medical and biomedical RAG
Self-evolving expertise in complex non-verifiable subject domains: dialogue as implicit meta-RLSynthetic data and self-improvement for LLMs → Self-evolving LLM agents +1
QSilk: Micrograin Stabilization and Adaptive Quantile Clipping for Detail-Friendly Latent DiffusionDiffusion models → Diffusion foundations and sampling
NDM: A Noise-driven Detection and Mitigation Framework against Implicit Sexual Intentions in Text-to-Image GenerationDiffusion models → Robustness, privacy and adversarial aspects
LLMs Judge Themselves: A Game-Theoretic Framework for Human-Aligned EvaluationLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
AURA: An Agent Autonomy Risk Assessment FrameworkLLM agents and tool use → Agent threats, guardrails and governance
Attention Sinks in Diffusion Language ModelsDiffusion models → Discrete and text diffusion
RLAF: Reinforcement Learning from Automaton FeedbackLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
The 3rd Place Solution of CCIR CUP 2025: A Framework for Retrieval-Augmented Generation in Multi-Turn Legal ConversationRetrieval-augmented generation → Legal and financial RAG
ProSh: Probabilistic Shielding for Model-free Reinforcement LearningDeep reinforcement learning methods → Safe and constrained RL
Cost-Aware Retrieval-Augmentation Reasoning Models with Adaptive Retrieval DepthLLM reasoning → Knowledge- and tool-augmented reasoning +1
Direct Preference Optimization with Unobserved Preference Heterogeneity: The Necessity of Ternary PreferencesLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
GraphMind: Interactive Novelty Assessment System for Accelerating Scientific DiscoveryLLM reasoning → Knowledge- and tool-augmented reasoning
Mixture of Experts Approaches in Dense Retrieval TasksMixture-of-experts → MoE large language models
SQuAI: Scientific Question-Answering with Multi-Agent Retrieval-Augmented GenerationRetrieval-augmented generation → Agentic RAG, agents and memory +1
ProofBridge: Auto-Formalization of Natural Language Proofs in Lean via Joint EmbeddingsRetrieval-augmented generation → Agentic RAG, agents and memory
HEADER: Hierarchical Robot Exploration via Attention-Based Deep Reinforcement Learning with Expert-Guided RewardDeep reinforcement learning methods → RL for robot navigation and UAV control
CarBoN: Calibrated Best-of-N Sampling Improves Test-time ReasoningLLM reasoning → Test-time compute and efficient thinking
Fast and Compact Tsetlin Machine Inference on CPUs Using Instruction-Level OptimizationEfficient LLM inference and serving → Serving systems and request scheduling
CQD-SHAP: Explainable Complex Query Answering via Shapley ValuesLLM reasoning → Knowledge- and tool-augmented reasoning
On-device Semantic Selection Made Low Latency and Memory Efficient with Monolithic ForwardingRetrieval-augmented generation → Generation and context integration
Unleashing Scientific Reasoning for Bio-experimental Protocol Generation via Structured Component-based Reward MechanismLLM reasoning → Clinical and biomedical applications
The Spark Effect: On Engineering Creative Diversity in Multi-Agent AI SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
JudgeSQL: Reasoning over SQL Candidates with Weighted Consensus TournamentLLM reasoning → Knowledge- and tool-augmented reasoning
Think Parallax: Solving Multi-Hop Problems via Multi-View Knowledge-Graph-Based Retrieval-Augmented GenerationLLM reasoning → Knowledge- and tool-augmented reasoning +1
Rethinking Cross-lingual Gaps from a Statistical ViewpointLLM reasoning → Reasoning benchmarks and evaluation
TokenTiming: A Dynamic Alignment Method for Universal Speculative Decoding Model PairsEfficient LLM inference and serving → Speculative decoding
VO-DP: Semantic-Geometric Adaptive Diffusion Policy for Vision-Only Robotic ManipulationDiffusion models → Diffusion and flow policies for robotics +1
LLM Latent Reasoning as Chain of SuperpositionLLM reasoning → Test-time compute and efficient thinking
Revisiting Knowledge Distillation: The Hidden Role of Dataset SizeKnowledge distillation → General knowledge distillation methods
Voting with the Graph: Stable RLAIF via Topological Consistency MaximizationLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Exploring Conditions for Diffusion models in Robotic ControlDiffusion models → Diffusion and flow policies for robotics +1
Perfect Prediction or Plenty of Proposals? What Matters Most in Planning for Autonomous DrivingEnd-to-end autonomous driving → Imitation-learning end-to-end planners
The Road Less Traveled: Enhancing Exploration in LLMs via Sequential SamplingLLM reasoning → Reinforcement learning for reasoning
HarmRLVR: Weaponizing Verifiable Rewards for Harmful LLM AlignmentLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
OffSim: Offline Simulator for Model-based Offline Inverse Reinforcement LearningOffline RL and imitation learning from data → Inverse reinforcement learning
SoK: Systematizing LLM Prompt Security: Taxonomies, Datasets, and Unified Evaluation of Attacks and DefensesLLM safety and security → Jailbreaking
Expediting Reinforcement Learning by Incorporating Knowledge About Temporal Causality in the EnvironmentDeep reinforcement learning methods → Deep RL algorithms and representations
CORE: Reducing UI Exposure in Mobile Agents via Collaboration Between Cloud and Local LLMsLLM agents and tool use → GUI and mobile agents
VDRive: Leveraging Reinforced VLA and Diffusion Policy for End-to-end Autonomous DrivingDiffusion models → Diffusion and flow policies for robotics +1
A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM ReasoningLLM reasoning → Test-time compute and efficient thinking
Select Less, Reason More: Prioritizing Evidence Purity for Video ReasoningLLM reasoning → Multimodal reasoning
Learning to Detect Unknown Jailbreak Attacks in Large Vision-Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Safe, Efficient, and Robust Reinforcement Learning for Ranking and Diffusion ModelsDiffusion models → Diffusion and flow policies for robotics +1
When Seeing Is not Enough: Revealing the Limits of Active Reasoning in MLLMsLLM reasoning → Multimodal reasoning +1
Fine-Tuning MedGemma for Clinical Captioning to Enhance Multimodal RAG over Malaysia CPGsRetrieval-augmented generation → Medical and biomedical RAG
Adaptive Minds: Empowering Agents with LoRA-as-ToolsLLM agents and tool use → General LLM agent frameworks and benchmarks
MARSHAL: Incentivizing Multi-Agent Reasoning via Self-Play with Strategic LLMsLLM reasoning → Reasoning for agents, planning and code +2
Iterative Refinement of Flow Policies in Probability Space for Online Reinforcement LearningDiffusion models → Diffusion and flow policies for robotics +1
Towards Robust Zero-Shot Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Towards Flash Thinking via Decoupled Advantage Policy OptimizationLLM reasoning → Reinforcement learning for reasoning +1
Infinity Parser: Layout Aware Reinforcement Learning for Scanned Document ParsingMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
AutoGraph-R1: End-to-End Reinforcement Learning for Knowledge Graph ConstructionRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG +1
Backdoor or Manipulation? Graph Mixture of Experts Can Defend Against Various Graph Adversarial AttacksMixture-of-experts → Vision mixture-of-experts
VERITAS: Leveraging Vision Priors and Expert Fusion to Improve Multimodal DataMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Accelerating Mobile Language Model via Speculative Decoding and NPU-Coordinated ExecutionRetrieval-augmented generation → RAG pipeline design and evaluation +1
DSSmoothing: Toward Certified Dataset Ownership Verification for Pre-trained Language Models via Dual-Space SmoothingLLM safety and security → Watermarking generated content and models
Latent Diffusion Model without Variational AutoencoderDiffusion models → Diffusion foundations and sampling
MTmixAtt: Integrating Mixture-of-Experts with Multi-Mix Attention for Large-Scale RecommendationMixture-of-experts → MoE for recommender systems
Exemplar-Guided Planing: Enhanced LLM Agent for KGQALLM agents and tool use → Web, search, and research agents
DriveGen3D: Boosting Feed-Forward Driving Scene Generation with Efficient Video DiffusionDiffusion models → Diffusion and flow policies for robotics

All 373