Papers

New on arXiv

Wed 26 Feb 2025

357 papers

OS-Kairos: Adaptive Interaction for MLLM-Powered GUI AgentsLLM agents and tool use → GUI and mobile agents
Medical Hallucinations in Foundation Models and Their Impact on HealthcareLLM reasoning → Hallucination detection and mitigation
FAA-CLIP: Federated Adversarial Adaptation of CLIPMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
BEYONDWORDS is All You Need: Agentic Generative AI based Social Media Themes ExtractorLLM agents and tool use → Social simulation and multi-agent debate
Uncertainty Comes for Free: Human-in-the-Loop Policies with Diffusion ModelsDiffusion models → Diffusion and flow policies for robotics
MedVLM-R1: Incentivizing Medical Reasoning Capability of Vision-Language Models (VLMs) via Reinforcement LearningLLM reasoning → Clinical and biomedical applications +1
Agentic Mixture-of-Workflows for Multi-Modal Chemical SearchLLM agents and tool use → Medical and scientific domain agents
3D Nephrographic Image Synthesis in CT Urography with the Diffusion Model and Swin TransformerDiffusion models → Medical imaging and segmentation
Weaker LLMs' Opinions Also Matter: Mixture of Opinions Enhances LLM's Mathematical ReasoningLLM reasoning → Mathematical reasoning
Self-rewarding correction for mathematical reasoningLLM reasoning → Test-time compute and efficient thinking +1
Program Synthesis Dialog Agents for Interactive Decision-MakingLLM agents and tool use → General LLM agent frameworks and benchmarks
Reference-Aligned Retrieval-Augmented Question Answering over Heterogeneous Proprietary DocumentsRetrieval-augmented generation → RAG robustness and adaptive retrieval
NeoBERT: A Next-Generation BERTLLM reasoning → Reasoning benchmarks and evaluation
Atlas: A Framework for ML Lifecycle Provenance & TransparencyLLM safety and security → Machine unlearning
Diffusion-based Planning with Learned Viability FiltersDiffusion models → Diffusion and flow policies for robotics
Stay Focused: Problem Drift in Multi-Agent DebateLLM agents and tool use → Social simulation and multi-agent debate
Distill Not Only Data but Also Rewards: Can Smaller Language Models Surpass Larger Ones?LLM reasoning → Reinforcement learning for reasoning +2
Assessing Autonomous Inspection Regimes: Active Versus Passive Satellite InspectionDeep reinforcement learning methods → RL for robot navigation and UAV control
CNS-Obsidian: A Neurosurgical Vision-Language Model Built From Scientific PublicationsMultimodal LLMs / vision-language models → Medical vision-language models
Winning Big with Small Models: Knowledge Distillation vs. Self-Training for Reducing Hallucination in Product QA AgentsLLM agents and tool use → Web, search, and research agents +2
Efficient Reinforcement Learning by Guiding World Models with Non-Curated DataWorld models → Latent dynamics models and model-based RL +1
No, of Course I Can! Deeper Fine-Tuning Attacks That Bypass Token-Level Safety MechanismsLLM safety and security → Safety alignment, refusal and guardrails
Retrieval Augmented Anomaly Detection (RAAD): Nimble Model Adjustment Without RetrainingRetrieval-augmented generation → Privacy and security of RAG
Static Vs. Agentic Game Master AI for Facilitating Solo Role-Playing ExperiencesLLM agents and tool use → Reinforcement learning for LLM agents
Conversational Planning for Personal PlansLLM reasoning → Reasoning for agents, planning and code
On the Interpolation Effect of Score Smoothing in Diffusion ModelsDiffusion models → Diffusion foundations and sampling
Hedging and Non-Affirmation: Quantifying LLM Alignment on Questions of Human RightsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
FLAP: Fully-controllable Audio-driven Portrait Video Generation through 3D head conditioned diffusion modelDiffusion models → Audio, speech and music
TransVDM: Motion-Constrained Video Diffusion Model for Transparent Video SynthesisDiffusion models → Video generation
Can Language Models Falsify? Evaluating Algorithmic Reasoning with Counterexample CreationLLM reasoning → Mathematical reasoning
Code to Think, Think to Code: A Survey on Code-Enhanced Reasoning and Reasoning-Driven Code Intelligence in LLMsLLM reasoning → Reasoning for agents, planning and code
ImageChain: Advancing Sequential Image-to-Text Reasoning in Multimodal Large Language ModelsLLM reasoning → Multimodal reasoning +1
Can Large Language Models Detect Errors in Long Chain-of-Thought Reasoning?LLM reasoning → Prompting and chain-of-thought reasoning
Recurrent Auto-Encoders for Enhanced Deep Reinforcement Learning in Wilderness Search and Rescue PlanningDeep reinforcement learning methods → RL for robot navigation and UAV control
Hybrid Robot Learning for Automatic Robot Motion Planning in ManufacturingDeep reinforcement learning methods → RL for robot locomotion and manipulation
Agentic Reward Modeling: Integrating Human Preferences with Verifiable Correctness Signals for Reliable Reward SystemsLLM reasoning → Reinforcement learning for reasoning +2
FSPO: Few-Shot Optimization of Synthetic Preferences Personalizes to Real UsersLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
WOFOSTGym: A Crop Simulator for Learning Annual and Perennial Crop Management StrategiesDeep reinforcement learning methods → RL for energy, networks and operations
Rethinking LLM Unlearning Objectives: A Gradient Perspective and Go BeyondLLM reasoning → Reasoning for agents, planning and code +1
Agent-centric Information AccessLLM agents and tool use → Web, search, and research agents
On the Importance of Text Preprocessing for Multimodal Representation Learning and Pathology Report GenerationMultimodal LLMs / vision-language models → Medical vision-language models
Efficient Federated Search for Retrieval-Augmented Generation using Lightweight RoutingRetrieval-augmented generation → RAG pipeline design and evaluation
CritiQ: Mining Data Quality Criteria from Human PreferencesLLM reasoning → Reasoning benchmarks and evaluation
Neural Antidote: Class-Wise Prompt Tuning for Purifying Backdoors in CLIPMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Drop-Upcycling: Training Sparse Mixture of Experts with Partial Re-initializationMixture-of-experts → MoE large language models
Poster: Long PHP webshell files detection based on sliding window attentionLLM safety and security → Prompt injection and agent security
Can RLHF be More Efficient with Imperfect Reward Models? A Policy Coverage PerspectiveLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
ObjectVLA: End-to-End Open-World Object Manipulation Without DemonstrationVision-language-action models → Human-video and latent-action pretraining +1
Two Heads Are Better Than One: Dual-Model Verbal Reflection at Inference-TimeLLM reasoning → Test-time compute and efficient thinking +1
Bi'an: A Bilingual Benchmark and Model for Hallucination Detection in Retrieval-Augmented GenerationRetrieval-augmented generation → Hallucination detection in RAG
FaithUn: Toward Faithful Forgetting in Language Models by Investigating the Interconnectedness of KnowledgeLLM safety and security → Machine unlearning
HDM: Hybrid Diffusion Model for Unified Image Anomaly DetectionDiffusion models → Flow matching, time series and science
BIG-Bench Extra HardLLM reasoning → Reasoning benchmarks and evaluation
Knowledge Distillation for Semantic Segmentation: A Label Space Unification ApproachKnowledge distillation → Detection and dense prediction distillation
MEDDxAgent: A Unified Modular Agent Framework for Explainable Automatic Differential DiagnosisLLM agents and tool use → Medical and scientific domain agents
TestNUC: Enhancing Test-Time Computing Approaches and Scaling through Neighboring Unlabeled Data ConsistencyLLM reasoning → Test-time compute and efficient thinking
When Personalization Meets Reality: A Multi-Faceted Analysis of Personalized Preference LearningLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Combining Large Language Models and Symbolic Reasoning for Multi-Robot Temporal Planning through Explainable Knowledge BasesLLM reasoning → Spatial and embodied reasoning
Voting or Consensus? Decision-Making in Multi-Agent DebateLLM agents and tool use → Social simulation and multi-agent debate
Nexus: A Lightweight and Scalable Multi-Agent Framework for Complex Tasks AutomationLLM reasoning → Reasoning for agents, planning and code +1
EndoMamba: An Efficient Foundation Model for Endoscopic Videos via Hierarchical Pre-trainingState-space sequence models (S4, Mamba) → Mamba for medical image analysis
MathDebugger: Detecting and Diagnosing Errors in Synthetic Mathematical DataLLM reasoning → Mathematical reasoning
A Dual-Purpose Framework for Backdoor Defense and Backdoor Amplification in Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects +1
Beyond Surface-Level Patterns: An Essence-Driven Defense Framework Against Jailbreak Attacks in LLMsLLM safety and security → Jailbreaking
FungalZSL: Zero-Shot Fungal Classification with Image Captioning Using a Synthetic Data ApproachMultimodal LLMs / vision-language models → Agricultural vision-language models
Distilling Reinforcement Learning Algorithms for In-Context Model-Based PlanningDeep reinforcement learning methods → Deep RL algorithms and representations
MEBench: Benchmarking Large Language Models for Cross-Document Multi-Entity Question AnsweringRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
OntologyRAG: Better and Faster Biomedical Code Mapping with Retrieval-Augmented Generation (RAG) Leveraging Ontology Knowledge Graphs and Large Language ModelsRetrieval-augmented generation → Medical and biomedical RAG
Evaluating Membership Inference Attacks in heterogeneous-data setupsLLM safety and security → Membership inference attacks
Cycles and collusion in congestion games under Q-learningDeep reinforcement learning methods → RL for games and combinatorial optimization
PEToolLLM: Towards Personalized Tool Learning in Large Language ModelsLLM agents and tool use → Tool learning and function calling
Fewer May Be Better: Enhancing Offline Reinforcement Learning with Reduced DatasetOffline RL and imitation learning from data → Offline RL algorithms
DualSpec: Text-to-spatial-audio Generation via Dual-Spectrogram Guided Diffusion ModelDiffusion models → Audio, speech and music
Towards Label-Only Membership Inference Attack against Pre-trained Large Language ModelsLLM safety and security → Membership inference attacks
JailBench: A Comprehensive Chinese Security Assessment Benchmark for Large Language ModelsLLM safety and security → Jailbreaking
MegaTTS 3: Sparse Alignment Enhanced Latent Diffusion Transformer for Zero-Shot Speech SynthesisDiffusion models → Audio, speech and music
VEM: Environment-Free Exploration for Training GUI Agent with Value Environment ModelMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +2
Learning to Generate Structured Output with Schema Reinforcement LearningDeep reinforcement learning methods → RL training of language and agent models
Multi-LLM Collaborative Search for Complex Problem SolvingLLM reasoning → Reasoning for agents, planning and code
A Theoretical Perspective: How to Prevent Model Collapse in Self-consuming Training LoopsSynthetic data and self-improvement for LLMs → Model collapse from synthetic data
One-shot Optimized Steering Vectors Mediate Safety-relevant Behaviors in LLMsMechanistic interpretability → Steering and probing for LLM safety
Marking Code Without Breaking It: Code Watermarking for Detecting LLM-Generated CodeLLM safety and security → Watermarking generated content and models
RL-OGM-Parking: Lidar OGM-Based Hybrid Reinforcement Learning Planner for Autonomous ParkingDeep reinforcement learning methods → RL for autonomous driving and traffic
REALM-Bench: A Benchmark for Evaluating Multi-Agent Systems on Real-world, Dynamic Planning and Scheduling TasksLLM agents and tool use → General LLM agent frameworks and benchmarks
Data-Efficient Multi-Agent Spatial Planning with LLMsLLM agents and tool use → General LLM agent frameworks and benchmarks
Judge as A Judge: Improving the Evaluation of Retrieval-Augmented Generation through the Judge-Consistency of Large Language ModelsRetrieval-augmented generation → RAG pipeline design and evaluation
Grad-ECLIP: Gradient-based Visual and Textual Explanations for CLIPMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Holistic Audit Dataset Generation for LLM Unlearning via Knowledge Graph Traversal and Redundancy RemovalLLM safety and security → Machine unlearning
Optimal Stochastic Trace Estimation in Generative ModelingDiffusion models → Diffusion foundations and sampling
Towards Optimal Multi-draft Speculative DecodingEfficient LLM inference and serving → Speculative decoding
Research on Edge Computing and Cloud Collaborative Resource Scheduling Optimization Based on Deep Reinforcement LearningDeep reinforcement learning methods → RL for energy, networks and operations
Reward Shaping to Mitigate Reward Hacking in RLHFLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Ptychographic Image Reconstruction from Limited Data via Score-Based Diffusion Models with Physics-GuidanceDiffusion models → Inverse problems and image restoration
AgentSociety Challenge: Designing LLM Agents for User Modeling and Recommendation on Web PlatformsLLM agents and tool use → Agentic recommender systems
ZEBRA: Leveraging Model-Behavioral Knowledge for Zero-Annotation Preference Dataset ConstructionLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
QueryAdapter: Rapid Adaptation of Vision-Language Models in Response to Natural Language QueriesMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Random Forest-of-Thoughts: Uncertainty-aware Reasoning for Computational Social ScienceLLM reasoning → Prompting and chain-of-thought reasoning
TrajLLM: A Modular LLM-Enhanced Agent-Based Framework for Realistic Human Trajectory SimulationLLM agents and tool use → Multimodal and embodied agents
Multi-Perspective Attention Mechanism for Bias-Aware Sequential Recommendation
Truth in Text: A Meta-Analysis of ML-Based Cyber Information Influence Detection Approaches

All 357