Papers

New on arXiv

Fri 3 Oct 2025

358 papers

Brain-Language Model Alignment: Insights into the Platonic Hypothesis and Intermediate-Layer AdvantageLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Synthetic EEG Generation using Diffusion Models for Motor Imagery TasksDiffusion models → Medical imaging and segmentation
Beyond CNNs: Efficient Fine-Tuning of Multi-Modal LLMs for Object Detection on Low-Data RegimesMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection
Physics Knowledge in Frontier Models: A Diagnostic Study of Failure ModesMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
CoT Referring: Improving Referring Expression Tasks with Grounded ReasoningLLM reasoning → Multimodal reasoning +1
Knowledge Graph-Guided Multi-Agent Distillation for Reliable Industrial Question Answering with DatasetsLLM agents and tool use → Web, search, and research agents
Malice in Agentland: Down the Rabbit Hole of Backdoors in the AI Supply ChainLLM agents and tool use → Agent threats, guardrails and governance +1
Lang-PINN: From Language to Physics-Informed Neural Networks via a Multi-Agent FrameworkLLM agents and tool use → Web, search, and research agents
Adversarial Reinforcement Learning for Offensive and Defensive Agents in a Simulated Zero-Sum Network EnvironmentDeep reinforcement learning methods → Adversarial robustness and RL in finance
VeriGuard: Enhancing LLM Agent Safety via Verified Code GenerationLLM agents and tool use → Agent threats, guardrails and governance
DiT-VTON: Diffusion Transformer Framework for Unified Multi-Category Virtual Try-On and Virtual Try-All with Integrated Image EditingDiffusion models → 3D, point cloud and scene generation
Efficient Test-Time Scaling for Small Vision-Language ModelsLLM reasoning → Multimodal reasoning +1
Longitudinal Flow Matching for Trajectory ModelingDiffusion models → Flow matching, time series and science
Machine Unlearning Meets Adversarial Robustness via Constrained Interventions on LLMsLLM safety and security → Jailbreaking
Streaming Drag-Oriented Interactive Video Manipulation: Drag Anything, Anytime!Diffusion models → Video generation
SketchPlan: Diffusion Based Drone Planning From Human SketchesDiffusion models → Diffusion and flow policies for robotics
GraphMed-LT: Patient-Specific Graph Memory with Latent Clinical Thought Refinement for Multi-Turn Medical ConversationsLLM reasoning → Clinical and biomedical applications
Certifiable Safe RLHF: Semantic Grounding and Fixed Penalty Constraint Optimization for Safer LLM AlignmentLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks
TS-Reasoner: Aligning Time Series Foundation Models with LLM ReasoningLLM reasoning → Time series reasoning
D2 Actor Critic: Diffusion Actor Meets Distributional CriticDiffusion models → Diffusion and flow policies for robotics +1
AgentHub: A Registry for Discoverable, Verifiable, and Reproducible AI AgentsLLM agents and tool use → Agent threats, guardrails and governance
Trajectory Data Suffices for Statistically Efficient Policy Evaluation in Fixed-Horizon Offline RL with Linear qπ-Realizability and ConcentrabilityOffline RL and imitation learning from data → Offline RL algorithms
Learning Efficient Guardrails for ComplianceLLM agents and tool use → Agent threats, guardrails and governance +1
LLM Agents for Automated Dependency UpgradesLLM agents and tool use → Software engineering coding agents
Bridging LLM Planning Agents and Formal Methods: A Case Study in Plan VerificationLLM agents and tool use → General LLM agent frameworks and benchmarks
ALMAS: an Autonomous LLM-based Multi-Agent Software Engineering FrameworkLLM agents and tool use → Software engineering coding agents
Warm-Starting Optimization-Based Motion Planning for Robotic Manipulators via Point Cloud-Conditioned Flow MatchingDiffusion models → Diffusion and flow policies for robotics
The Argument is the Explanation: Structured Argumentation for Trust in AgentsLLM agents and tool use → Social simulation and multi-agent debate
Spatial-ViLT: Enhancing Visual Spatial Reasoning through Multi-Task LearningLLM reasoning → Spatial and embodied reasoning +1
Paris: A Decentralized Trained Open-Weight Diffusion ModelDiffusion models → Diffusion foundations and sampling
LegalWiz: A Multi-Agent Generation Framework for Contradiction Detection in Legal DocumentsRetrieval-augmented generation → Legal and financial RAG +1
NEXUS: Network Exploration for eXploiting Unsafe Sequences in Multi-Turn LLM JailbreaksLLM safety and security → Jailbreaking
LLMs Lean on Priors, Not Programming Language SemanticsLLM reasoning → Reasoning benchmarks and evaluation
LegalSim: Multi-Agent Simulation of Legal Systems for Discovering Procedural ExploitsLLM agents and tool use → Social simulation and multi-agent debate
Studying the Korean Word-Chain Game with RLVR: Mitigating Reward Conflicts via Curriculum LearningLLM reasoning → Reinforcement learning for reasoning
Visual Language Model as a Judge for Object Detection in Industrial DiagramsMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Conditional Pseudo-Supervised Contrast for Data-Free Knowledge DistillationKnowledge distillation → General knowledge distillation methods
Disentangling Recall and Reasoning in Transformer Models through Layer-wise Attention and Activation AnalysisLLM reasoning → Reasoning benchmarks and evaluation +1
Diffusion-Based, Data-Assimilation-Enabled Super-Resolution of Hub-height WindsDiffusion models → Flow matching, time series and science
Physics-informed Neural-operator Predictive Control for Drag Reduction in Turbulent FlowsDeep reinforcement learning methods → RL with diffusion and flow models
LEAML: Label-Efficient Adaptation to Out-of-Distribution Visual Tasks for Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
Reward Models are Metrics in a Trench CoatLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Attention-Aligned Reasoning for Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Low-probability Tokens Sustain Exploration in Reinforcement Learning with Verifiable RewardLLM reasoning → Reinforcement learning for reasoning
Abstain and Validate: A Dual-LLM Policy for Reducing Noise in Agentic Program RepairLLM agents and tool use → Software engineering coding agents
Cache-to-Cache: Direct Semantic Communication Between Large Language ModelsEfficient LLM inference and serving → KV cache optimization
To Distill or Decide? Understanding the Algorithmic Trade-off in Partially Observable Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Coevolutionary Continuous Discrete Diffusion: Make Your Diffusion Language Model a Latent ReasonerLLM reasoning → Diffusion reasoning and test-time scaling +1
FocusAgent: Simple Yet Effective Ways of Trimming the Large Context of Web AgentsLLM agents and tool use → Agent threats, guardrails and governance
Memory Forcing: Spatio-Temporal Memory for Consistent Scene Generation on MinecraftWorld models → Video-generation world models and simulators +1
CoDA: Agentic Systems for Collaborative Data VisualizationLLM agents and tool use → Multimodal and embodied agents
Superposition disentanglement of neural representations reveals hidden alignmentMechanistic interpretability → Sparse autoencoder training and evaluation
PRISM-Physics: Causal DAG-Based Process Evaluation for Physics ReasoningLLM reasoning → Spatial and embodied reasoning
Simulation to Rules: A Dual-VLM Framework for Formal Visual PlanningMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Q-Learning with Shift-Aware Upper Confidence Bound in Non-Stationary Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
Mixture of Many Zero-Compute Experts: A High-Rate Quantization Theory PerspectiveMixture-of-experts → MoE theory and gating functions
Taming Imperfect Process Verifiers: A Sampling Perspective on BacktrackingLLM reasoning → Test-time compute and efficient thinking
MM-Nav: Multi-View VLA Model for Robust Visual Navigation via Multi-Expert LearningVision-language-action models → 3D and spatial-aware VLA
Enhancing XAI Narratives through Multi-Narrative Refinement and Knowledge DistillationKnowledge distillation → Knowledge distillation for LLMs
A Study of Rule Omission in Raven's Progressive MatricesLLM reasoning → Multimodal reasoning
HAVIR: HierArchical Vision to Image Reconstruction using CLIP-Guided Versatile DiffusionDiffusion models → Medical imaging and segmentation
Listening or Reading? Evaluating Speech Awareness in Chain-of-Thought Speech-to-Text TranslationLLM reasoning → Audio-language model reasoning
Revisiting Direct Speech-to-Text Translation with Speech LLMs: Better Scaling than CoT Prompting?LLM reasoning → Audio-language model reasoning
Latent Diffusion Unlearning: Protecting Against Unauthorized Personalization Through Trajectory Shifted PerturbationsDiffusion models → Robustness, privacy and adversarial aspects +1
A Unified Deep Reinforcement Learning Approach for Close Enough Traveling Salesman ProblemDeep reinforcement learning methods → RL for games and combinatorial optimization
Comparative Analysis of Parameterized Action Actor-Critic Reinforcement Learning Algorithms for Web Search Match Plan GenerationDeep reinforcement learning methods → RL theory and sample complexity
Learning Robust Diffusion Models from Imprecise SupervisionDiffusion models → Diffusion foundations and sampling
Distributional Inverse Reinforcement LearningOffline RL and imitation learning from data → Inverse reinforcement learning +1
3D-CovDiffusion: 3D-Aware Diffusion Policy for Coverage Path PlanningDiffusion models → Diffusion and flow policies for robotics
NonTextual Target AttackLLM safety and security → Jailbreaking
TIT-Score: Evaluating Long-Prompt Based Text-to-Image Alignment via Text-to-Image-to-Text ConsistencyMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
Grounding Large Language Models in Clinical Evidence: A Retrieval-Augmented Generation System for Querying UK NICE Clinical GuidelinesRetrieval-augmented generation → Medical and biomedical RAG
External Data Extraction Attacks against Retrieval-Augmented Large Language ModelsRetrieval-augmented generation → Privacy and security of RAG +1
Leave No TRACE: Black-box Detection of Copyrighted Dataset Usage in Large Language Models via WatermarkingLLM safety and security → Watermarking generated content and models
ContextFlow: Context-Aware Flow Matching For Trajectory Inference From Spatial Omics DataDiffusion models → Medical imaging and segmentation
Ergodic Risk Measures: Towards a Risk-Aware Foundation for Continual Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Self-Reflective Generation at Test TimeLLM reasoning → Test-time compute and efficient thinking
Mechanistic Interpretability of Code Correctness in LLMs via Sparse AutoencodersMechanistic interpretability → Knowledge and representations inside LLMs
Improving Adversarial Robustness of Zero-Shot CLIP with Confidence-Aware WeightingMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
Don't Just Chase "Highlighted Tokens" in MLLMs: Revisiting Visual Holistic Context RetentionMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
FinReflectKG -- MultiHop: Financial QA Benchmark for Reasoning with Knowledge Graph EvidenceLLM reasoning → Reasoning for agents, planning and code
DMark: Order-Agnostic Watermarking for Diffusion Large Language ModelsLLM safety and security → Watermarking generated content and models
Global Convergence of Policy Gradient for Entropy Regularized Linear-Quadratic Control with Multiplicative NoiseDeep reinforcement learning methods → RL theory and sample complexity
RoiRL: Efficient, Self-Supervised Reasoning with Offline Iterative Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning
Consolidating Reinforcement Learning for Multimodal Discrete Diffusion ModelsLLM reasoning → Reinforcement learning for reasoning +2
Action Deviation-Aware Inference for Low-Latency Wireless RobotsEfficient LLM inference and serving → Visual token pruning in multimodal models
Reward Model Routing in AlignmentLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Flamed-TTS: Flow Matching Attention-Free Models for Efficient Generating and Dynamic Pacing Zero-shot Text-to-SpeechDiffusion models → Audio, speech and music
Beyond the Final Answer: Evaluating the Reasoning Trajectories of Tool-Augmented AgentsLLM reasoning → Reasoning for agents, planning and code +1
Attack via Overfitting: 10-shot Benign Fine-tuning to Jailbreak LLMsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
StepChain GraphRAG: Reasoning Over Knowledge Graphs for Multi-Hop Question AnsweringLLM reasoning → Knowledge- and tool-augmented reasoning +1
Multi-scale Autoregressive Models are Laplacian, Discrete, and Latent Diffusion Models in DisguiseDiffusion models → Image generation, text-to-image and editing
NCV: A Node-Wise Consistency Verification Approach for Low-Cost Structured Error Localization in LLM ReasoningLLM reasoning → Test-time compute and efficient thinking
Work Zones challenge VLM Trajectory Planning: Toward Mitigation and Robust Autonomous DrivingLLM reasoning → Chain-of-thought reasoning for driving +2
MaskCD: Mitigating LVLM Hallucinations by Image Head Masked Contrastive DecodingMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
Reasoning Riddles: How Explainability Reveals Cognitive Limits in Vision-Language ModelsLLM reasoning → Multimodal reasoning +1
TokenFlow: Responsive LLM Text Streaming Serving under Request Burst via Preemptive SchedulingEfficient LLM inference and serving → Serving systems and request scheduling
The Path of Self-Evolving Large Language Models: Achieving Data-Efficient Learning via Intrinsic FeedbackLLM reasoning → Reasoning for agents, planning and code +1
Bayesian Test-time Adaptation for Object Recognition and Detection with Vision-language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Retrv-R1: A Reasoning-Driven MLLM Framework for Universal and Efficient Multimodal RetrievalLLM reasoning → Multimodal reasoning

All 358