Papers

New on arXiv

Tue 4 Mar 2025

386 papers

Closing the Intent-to-Behavior Gap via Fulfillment Priority LogicDeep reinforcement learning methods → RL for games and combinatorial optimization
Multi-agent Auto-Bidding with Latent Graph Diffusion ModelsDiffusion models → Diffusion and flow policies for robotics
Self-Evolved Preference Optimization for Enhancing Mathematical Reasoning in Small Language ModelsLLM reasoning → Mathematical reasoning +2
Learning from Failures in Multi-Attempt Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning
Generating Novel Brain Morphology by Deforming Learned TemplatesDiffusion models → Medical imaging and segmentation
A2Perf: Real-World Autonomous Agents BenchmarkOffline RL and imitation learning from data → Imitation learning methods
LLM Misalignment via Adversarial RLHF PlatformsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks
SAFE: A Sparse Autoencoder-Based Framework for Robust Query Enrichment and Hallucination Mitigation in LLMsMechanistic interpretability → Hallucination detection from model internals
MoSE: Hierarchical Self-Distillation Enhances Early Layer EmbeddingsKnowledge distillation → Knowledge distillation for LLMs
Benchmarking Diffusion Annealing-Based Bayesian Inverse Problem SolversDiffusion models → Inverse problems and image restoration
Effectively Steer LLM To Follow Preference via Building Confident DirectionsLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Teaching AI to Handle Exceptions: Supervised Fine-Tuning with Human-Aligned JudgmentLLM agents and tool use → General LLM agent frameworks and benchmarks
LINGOLY-TOO: Disentangling Reasoning from Knowledge with Templatised Orthographic ObfuscationLLM reasoning → Reasoning benchmarks and evaluation
Koopman-Based Generalization of Deep Reinforcement Learning With Application to Wireless CommunicationsDeep reinforcement learning methods → RL for energy, networks and operations
KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for CodingLLM reasoning → Reasoning for agents, planning and code +1
LiteWebAgent: The Open-Source Suite for VLM-Based Web-Agent ApplicationsLLM agents and tool use → Software engineering coding agents
Diverse Controllable Diffusion Policy with Signal Temporal LogicDiffusion models → Diffusion and flow policies for robotics
Optimizing open-domain question answering with graph-based retrieval augmented generationRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Straight-Line Diffusion Model for Efficient 3D Molecular GenerationDiffusion models → Flow matching, time series and science
Interpretable Few-Shot Retinal Disease Diagnosis with Concept-Guided Prompting of Vision-Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
Text2Scenario: Text-Driven Scenario Generation for Autonomous Driving TestEnd-to-end autonomous driving → Large language models for driving
Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich ManipulationDiffusion models → Diffusion and flow policies for robotics +1
Language Models can Self-Improve at State-Value Estimation for Better SearchSynthetic data and self-improvement for LLMs → Self-evolving search agents
The First Few Tokens Are All You Need: An Efficient and Effective Unsupervised Prefix Fine-Tuning Method for Reasoning ModelsLLM reasoning → Test-time compute and efficient thinking
FairSense-AI: Responsible AI Meets SustainabilityMultimodal LLMs / vision-language models → Social bias and fairness in VLMs
Mask-DPO: Generalizable Fine-grained Factuality Alignment of LLMsLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
AlignDistil: Token-Level Language Model Alignment as Adaptive Policy DistillationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Meta-Learning to Explore via Memory Density FeedbackDeep reinforcement learning methods → Deep RL algorithms and representations
Feynman-Kac Correctors in Diffusion: Annealing, Guidance, and Product of ExpertsDiffusion models → Diffusion foundations and sampling
Q-Filters: Leveraging QK Geometry for Efficient KV Cache CompressionEfficient LLM inference and serving → KV cache optimization
MX-Font++: Mixture of Heterogeneous Aggregation Experts for Few-shot Font GenerationMixture-of-experts → Multimodal MoE models
Teaching Your Models to Understand Code via Focal Preference AlignmentLLM alignment / RLHF / preference optimisation → Process reward models for reasoning
Implicit Bias in LLMs: A SurveyLLM safety and security → Safety alignment, refusal and guardrails
BatchGEMBA: Token-Efficient Machine Translation Evaluation with Batched Prompting and Prompt CompressionEfficient LLM inference and serving → Prompt and context compression
Bridging VLM and KMP: Enabling Fine-grained robotic manipulation via Semantic Keypoints RepresentationMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Variable-Friction In-Hand Manipulation for Arbitrary Objects via Diffusion-Based Imitation LearningOffline RL and imitation learning from data → Robot manipulation imitation learning
MPO: Boosting LLM Agents with Meta Plan OptimizationLLM agents and tool use → General LLM agent frameworks and benchmarks
Multidimensional Consistency Improves Reasoning in Language ModelsLLM reasoning → Mathematical reasoning
Human-aligned Safe Reinforcement Learning for Highway On-Ramp Merging in Dense TrafficDeep reinforcement learning methods → RL for autonomous driving and traffic
XFMamba: Cross-Fusion Mamba for Multi-View Medical Image ClassificationState-space sequence models (S4, Mamba) → Mamba for medical image analysis
Reinforcement Learning-based Threat AssessmentDeep reinforcement learning methods → Adversarial robustness and RL in finance
OkraLong: A Flexible Retrieval-Augmented Framework for Long-Text Query ProcessingRetrieval-augmented generation → Generation and context integration
Seeing is Understanding: Unlocking Causal Attention into Modality-Mutual Attention for Multimodal LLMsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
StageDesigner: Artistic Stage Generation for Scenography via Theater ScriptsDiffusion models → 3D, point cloud and scene generation
SPG: Improving Motion Diffusion by Smooth Perturbation GuidanceDiffusion models → Video generation
LLM-Safety Evaluations Lack RobustnessLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
World Models for Anomaly Detection during Model-Based Reinforcement Learning InferenceWorld models → Latent dynamics models and model-based RL +1
SpecInF: Exploiting Idle GPU Resources in Distributed DL Training via Speculative Inference FillingEfficient LLM inference and serving → Serving systems and request scheduling
RectifiedHR: Enable Efficient High-Resolution Synthesis via Energy RectificationDiffusion models → Diffusion foundations and sampling
Generator-Assistant Stepwise Rollback Framework for Large Language Model AgentLLM agents and tool use → General LLM agent frameworks and benchmarks
Q&C: When Quantization Meets Cache in Efficient Image GenerationDiffusion models → Diffusion foundations and sampling
Union of Experts: Adapting Hierarchical Routing to Equivalently Decomposed TransformerMixture-of-experts → Expert routing strategies
BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion ModelingDiffusion models → Flow matching, time series and science +2
AILS-NTUA at SemEval-2025 Task 4: Parameter-Efficient Unlearning for Large Language Models using Data ChunkingLLM safety and security → Machine unlearning
A comparison of visual representations for real-world reinforcement learning in the context of vacuum grippingDeep reinforcement learning methods → RL for robot locomotion and manipulation
AutoEval: A Practical Framework for Autonomous Evaluation of Mobile AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
PersonaX: A Recommendation Agent Oriented User Modeling Framework for Long Behavior SequenceLLM agents and tool use → Agentic recommender systems
Vision-Language Model IP Protection via Prompt-based LearningMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
ReSo: A Reward-driven Self-organizing LLM-based Multi-Agent System for Reasoning TasksLLM reasoning → Reasoning for agents, planning and code +2
An Efficient and Precise Training Data Construction Framework for Process-supervised Reward Model in Mathematical ReasoningLLM reasoning → Reinforcement learning for reasoning +1
Teaching Metric Distance to Discrete Autoregressive Language ModelsLLM reasoning → Reinforcement learning for reasoning
JPDS-NN: Reinforcement Learning-Based Dynamic Task Allocation for Agricultural Vehicle Routing OptimizationDeep reinforcement learning methods → RL for games and combinatorial optimization
Are Large Vision Language Models Good Game Players?Multimodal LLMs / vision-language models → Embodied and 3D vision-language models
Controllable Motion Generation via Diffusion Modal CouplingDiffusion models → Diffusion and flow policies for robotics
DeLTa: A Decoding Strategy based on Logit Trajectory Prediction Improves Factuality and Reasoning AbilityLLM reasoning → Test-time compute and efficient thinking
Unlocking a New Rust Programming Experience: Fast and Slow Thinking with LLMs to Conquer Undefined BehaviorsLLM reasoning → Reasoning for agents, planning and code
BiasICL: In-Context Learning and Demographic Biases of Vision Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
COMMA: Coordinate-aware Modulated Mamba Network for 3D Dispersed Vessel SegmentationState-space sequence models (S4, Mamba) → Mamba for medical image analysis
PromptCoT: Synthesizing Olympiad-level Problems for Mathematical Reasoning in Large Language ModelsLLM reasoning → Mathematical reasoning
Audio-Reasoner: Improving Reasoning Capability in Large Audio Language ModelsLLM reasoning → Audio-language model reasoning
Go Beyond Your Means: Unlearning with Per-Sample Gradient OrthogonalizationLLM safety and security → Machine unlearning
Target Return Optimizer for Multi-Game Decision TransformerOffline RL and imitation learning from data → Decision transformers and sequence modeling
PD-VLA: Accelerating Vision-Language-Action Model Integrated with Action Chunking via Parallel DecodingVision-language-action models → Efficient VLA inference and adaptation +1
A Token-level Text Image Foundation Model for Document UnderstandingMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Diffusion-Based mmWave Radar Point Cloud Enhancement Driven by Range ImagesDiffusion models → 3D, point cloud and scene generation
DreamerV3 for Traffic Signal Control: Hyperparameter Tuning and PerformanceDeep reinforcement learning methods → RL for autonomous driving and traffic
Active Robot Curriculum Learning from Online Human DemonstrationsOffline RL and imitation learning from data → Robot learning from demonstration
Experience Replay with Random ReshufflingDeep reinforcement learning methods → Deep RL algorithms and representations
AppAgentX: Evolving GUI Agents as Proficient Smartphone UsersLLM agents and tool use → GUI and mobile agents
WMNav: Integrating Vision-Language Models into World Models for Object Goal NavigationWorld models → World models for navigation +1
Haste Makes Waste: Evaluating Planning Abilities of LLMs for Efficient and Feasible Multitasking with Time Constraints Between ActionsLLM reasoning → Reasoning for agents, planning and code +1
Language-Guided Visual Perception Disentanglement for Image Quality Assessment and Conditional Image GenerationMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Words or Vision: Do Vision-Language Models Have Blind Faith in Text?Multimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
ATLaS: Agent Tuning via Learning Critical StepsLLM agents and tool use → General LLM agent frameworks and benchmarks
h-Edit: Effective and Flexible Diffusion-Based Editing via Doob's h-TransformDiffusion models → Image generation, text-to-image and editing
DivPrune: Diversity-based Visual Token Pruning for Large Multimodal ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Adversarial TokenizationLLM safety and security → Jailbreaking
KGCompiler: Deep Learning Compilation Optimization for Knowledge Graph Complex Logical Query AnsweringLLM reasoning → Knowledge- and tool-augmented reasoning
Is Bellman Equation Enough for Learning Control?Deep reinforcement learning methods → RL theory and sample complexity
LLM-TabLogic: Preserving Inter-Column Logical Relationships in Synthetic Tabular Data via Prompt-Guided Latent DiffusionDiffusion models → Flow matching, time series and science
MedHEval: Benchmarking Hallucinations and Mitigation Strategies in Medical Large Vision-Language ModelsMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
Tabby: A Language Model Architecture for Tabular and Structured Data SynthesisSynthetic data and self-improvement for LLMs → Privacy-preserving tabular data synthesis
Four Principles for Physically Interpretable World ModelsWorld models → Latent predictive world models
Memory and Bandwidth are All You Need for Fully Sharded Data Parallel
PointSplit: Towards On-device 3D Object Detection with Heterogeneous Low-power Accelerators
From Voices to Worlds: Developing an AI-Powered Framework for 3D Object Generation in Augmented Reality
Radar Pulse Deinterleaving with Transformer Based Deep Metric Learning
Cross-Subject Depression Level Classification Using EEG Signals with a Sample Confidence Method
Robust Detection of Extremely Thin Lines Using 0.2mm Piano Wire
CORDIC Is All You Need

All 386