Papers

New on arXiv

Mon 6 Oct 2025

400 papers

DREAMer-VXS: A Latent World Model for Sample-Efficient AGV Exploration in Stochastic, Unobserved EnvironmentsWorld models → World models for navigation +1
A Multidisciplinary Design and Optimization (MDO) Agent Driven by Large Language ModelsLLM agents and tool use → Multimodal and embodied agents
SER-Diff: Synthetic Error Replay Diffusion for Incremental Brain Tumor SegmentationDiffusion models → Medical imaging and segmentation +1
Surgeons Are Indian Males and Speech Therapists Are White Females: Auditing Biases in Vision-Language Models for Healthcare ProfessionalsMultimodal LLMs / vision-language models → Medical vision-language models
Dynamic Object Masks as Goal Representations for Visual Goal-Conditioned Reinforcement LearningDeep reinforcement learning methods → RL for robot navigation and UAV control
Reproducibility Study of "XRec: Large Language Models for Explainable Recommendation"Mixture-of-experts → MoE large language models
Bridging Reasoning to Learning: Unmasking Illusions using Complexity Out of Distribution GeneralizationLLM reasoning → Test-time compute and efficient thinking
RareGraph-Synth: Knowledge-Guided Diffusion Models for Generating Privacy-Preserving Synthetic Patient Trajectories in Ultra-Rare DiseasesDiffusion models → Medical imaging and segmentation
VAL-Bench: Belief Consistency as a measure for Value Alignment in Language ModelsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Prior-Aligned Meta-RL: Thompson Sampling with Learned Priors and Guarantees in Finite-Horizon MDPsDeep reinforcement learning methods → Deep RL algorithms and representations
AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question AnsweringLLM agents and tool use → Web, search, and research agents
Adversarial Reinforcement Learning for Large Language Model Agent SafetyLLM agents and tool use → Agent threats, guardrails and governance +1
UnitTenX: Generating Tests for Legacy Packages with AI Agents Powered by Formal VerificationLLM agents and tool use → Software engineering coding agents
Draft, Verify, and Improve: Toward Training-Aware Speculative DecodingEfficient LLM inference and serving → Speculative decoding
A Lightweight Large Language Model-Based Multi-Agent System for 2D Frame Structural AnalysisLLM agents and tool use → Multimodal and embodied agents
See the past: Time-Reversed Scene Reconstruction from Thermal Traces Using Visual Language ModelsMultimodal LLMs / vision-language models → Video understanding with MLLMs
AutoDAN-Reasoning: Enhancing Strategies Exploration based Jailbreak Attacks with Test-Time ScalingLLM reasoning → Reasoning for agents, planning and code +1
KVLinC : KV Cache Quantization with Hadamard Rotation and Linear CorrectionEfficient LLM inference and serving → KV cache quantization
LightCache: Memory-Efficient, Training-Free Acceleration for Video GenerationDiffusion models → Video generation
Mitigating Diffusion Model Hallucinations with Dynamic GuidanceDiffusion models → Medical imaging and segmentation
Margin Adaptive DPO: Leveraging Reward Model for Granular Control in Preference OptimizationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
WeatherArchive-Bench: Benchmarking Retrieval-Augmented Reasoning for Historical Weather ArchivesRetrieval-augmented generation → Agentic RAG, agents and memory
Biomedical reasoning in action: Multi-agent System for Auditable Biomedical Evidence SynthesisLLM reasoning → Clinical and biomedical applications +1
DeepV: A Model-Agnostic Retrieval-Augmented Framework for Verilog Code Generation with a High-Quality Knowledge BaseRetrieval-augmented generation → RAG for code
RAG Makes Guardrails Unsafe? Investigating Robustness of Guardrails under RAG-style ContextsLLM safety and security → Safety alignment, refusal and guardrails
When Should Users Check? Modeling Confirmation Frequency in Multi-Step Agentic AI TasksLLM agents and tool use → Software engineering coding agents
Adjusting the Output of Decision Transformer with Action GradientOffline RL and imitation learning from data → Decision transformers and sequence modeling +1
Beyond Monolithic Rewards: A Hybrid and Multi-Aspect Reward Optimization for MLLM AlignmentMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
Chrysalis: A Unified System for Comparing Active Teaching and Passive Learning with AI Agents in EducationLLM agents and tool use → Educational agents and math problem solving
Let it Calm: Exploratory Annealed Decoding for Verifiable Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning +1
Stratum: System-Hardware Co-Design with Tiered Monolithic 3D-Stackable DRAM for Efficient MoE ServingMixture-of-experts → Efficient MoE training and inference systems
Indirect Prompt Injections: Are Firewalls All You Need, or Stronger Benchmarks?LLM safety and security → Prompt injection and agent security
A Data-Driven Prism: Multi-View Source Separation with Diffusion Model PriorsDiffusion models → Inverse problems and image restoration
From surveillance to signalling: escalation channels as environmental controls for agentic AILLM agents and tool use → Agent threats, guardrails and governance
Plug-and-Play Dramaturge: A Divide-and-Conquer Approach for Iterative Narrative Script Refinement via Collaborative LLM AgentsLLM agents and tool use → Multimodal and embodied agents
From Noisy Traces to Stable Gradients: Bias-Variance Optimized Preference Optimization for Aligning Large Reasoning ModelsLLM reasoning → Test-time compute and efficient thinking +1
Factuality Matters: When Image Generation and Editing Meet Structured VisualsLLM reasoning → Multimodal reasoning
Finish First, Perfect Later: Test-Time Token-Level Cross-Validation for Diffusion Large Language ModelsEfficient LLM inference and serving → Diffusion LLM and MoE inference
SAEdit: Token-level control for continuous image editing via Sparse AutoEncoderDiffusion models → Image generation, text-to-image and editing +1
Slm-mux: Orchestrating small language models for reasoningLLM reasoning → Reasoning benchmarks and evaluation
SwiReasoning: Switch-Thinking in Latent and Explicit for Pareto-Superior Reasoning LLMsLLM reasoning → Test-time compute and efficient thinking
Automaton Constrained Q-LearningDeep reinforcement learning methods → Interpretable and logic-guided RL
Staircase Streaming for Low-Latency Multi-Agent InferenceLLM agents and tool use → General LLM agent frameworks and benchmarks
StaMo: Unsupervised Learning of Generalizable Robot Motion from Compact State RepresentationWorld models → Embodied and robot world models
Proactive defense against LLM JailbreakLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
Look-ahead Reasoning with a Learned Model in Imperfect Information GamesLLM reasoning → Reasoning for agents, planning and code
Test-Time Scaling in Diffusion LLMs via Hidden Semi-Autoregressive ExpertsLLM reasoning → Diffusion reasoning and test-time scaling
Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal ModelsLLM reasoning → Multimodal reasoning +1
Causal Abstractions, Categorically UnifiedMechanistic interpretability → Circuits and mechanistic analysis
Imperceptible Jailbreaking against Large Language ModelsLLM safety and security → Jailbreaking
Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignmentLLM alignment / RLHF / preference optimisation → Reward hacking and misspecification
Think Then Embed: Generative Context Improves Multimodal EmbeddingLLM reasoning → Multimodal reasoning
Curiosity-Driven Development of Action and Language in Robots Through Self-ExplorationDeep reinforcement learning methods → RL training of language and agent models
Resource-Efficient Fine-Tuning of LLaMA-3.2-3B for Medical Chain-of-Thought ReasoningLLM reasoning → Clinical and biomedical applications
Reinforce-Ada: An Adaptive Sampling Framework under Non-linear RL ObjectivesLLM reasoning → Reinforcement learning for reasoning
Efficient Navigation in Unknown Indoor Environments with Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
LLM-Hanabi: Evaluating Multi-Agent Gameplays with Theory-of-Mind and Rationale Inference in Imperfect Information Collaboration GameLLM agents and tool use → Reinforcement learning for LLM agents
Aligning Perception, Reasoning, Modeling and Interaction: A Survey on Physical AIWorld models → Latent predictive world models
ActiveMark: on watermarking of visual foundation models via massive activationsLLM safety and security → Watermarking generated content and models
Safe and Compliant Cross-Market Trade Execution via Constrained RL and Zero-Knowledge AuditsDeep reinforcement learning methods → Adversarial robustness and RL in finance
On Structured State-Space DualityState-space sequence models (S4, Mamba) → Core SSM architectures for language
MARS: Co-evolving Dual-System Deep Research via Multi-Agent Reinforcement LearningLLM reasoning → Reasoning for agents, planning and code
REN: Anatomically-Informed Mixture-of-Experts for Interstitial Lung Disease DiagnosisMixture-of-experts → Medical and biomedical MoE
Retrieval-Augmented Code Generation: A Survey with Focus on Repository-Level ApproachesRetrieval-augmented generation → RAG for code
Benchmarking M-LTSF: Frequency and Noise-Based Evaluation of Multivariate Long Time Series Forecasting ModelsState-space sequence models (S4, Mamba) → Time series forecasting with SSMs
HyperVLA: Efficient Inference in Vision-Language-Action Models via HypernetworksVision-language-action models → Generalist VLA models and benchmarks
SocialHarmBench: Revealing LLM Vulnerabilities to Socially Harmful RequestsLLM safety and security → Safety alignment, refusal and guardrails
Where Did It All Go Wrong? A Hierarchical Look into Multi-Agent Error AttributionLLM agents and tool use → Software engineering coding agents
RL Is a Hammer and LLMs Are Nails: A Simple Reinforcement Learning Recipe for Strong Prompt InjectionLLM agents and tool use → Agent threats, guardrails and governance +1
Flow-Matching Based Refiner for Molecular Conformer GenerationDiffusion models → Flow matching, time series and science
Less is More: Recursive Reasoning with Tiny NetworksLLM reasoning → Test-time compute and efficient thinking
Model Predictive Control-Guided Reinforcement Learning for Implicit BalancingDeep reinforcement learning methods → RL for energy, networks and operations
Alignment Tipping Process: How Self-Evolution Pushes LLM Agents Off the RailsLLM agents and tool use → General LLM agent frameworks and benchmarks
ERDE: Entropy-Regularized Distillation for Early-exitEfficient LLM inference and serving → Early exit and edge inference +1
FreshBrew: A Benchmark for Evaluating AI Agents on Java Code MigrationLLM agents and tool use → Software engineering coding agents
LEGOMem: Modular Procedural Memory for Multi-agent LLM Systems for Workflow AutomationLLM agents and tool use → Agent memory systems
Detecting Distillation Data from Reasoning ModelsLLM reasoning → Prompting and chain-of-thought reasoning +1
From Actions to Kinesics: Extracting Human Psychological States through Bodily MovementsDeep reinforcement learning methods → RL for robot locomotion and manipulation
Flow Matching for Conditional MRI-CT and CBCT-CT Image SynthesisDiffusion models → Medical imaging and segmentation
Visual Representations inside the Language ModelMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Natural Language Edge Labelling: Decoupling Intent from Execution in Structured LM ReasoningLLM reasoning → Test-time compute and efficient thinking
Hybrid Architectures for Language Models: Systematic Analysis and Design InsightsState-space sequence models (S4, Mamba) → Core SSM architectures for language
GUISpector: An MLLM Agent Framework for Automated Verification of Natural Language Requirements in GUI PrototypesLLM agents and tool use → Software engineering coding agents
Trade in Minutes! Rationality-Driven Agentic System for Quantitative Financial TradingLLM agents and tool use → Financial trading and forecasting agents
Learning on the Job: Test-Time Curricula for Targeted Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Distribution Preference Optimization: A Fine-grained Perspective for LLM UnlearningLLM alignment / RLHF / preference optimisation → LLM preference optimization methods +1
When Do Credal Sets Stabilize? Fixed-Point Theorems for Credal Set UpdatesDeep reinforcement learning methods → RL theory and sample complexity
ParallelBench: Understanding the Trade-offs of Parallel Decoding in Diffusion LLMsEfficient LLM inference and serving → Diffusion LLM and MoE inference
ModernBERT + ColBERT: Enhancing biomedical RAG through an advanced re-ranking retrieverRetrieval-augmented generation → Medical and biomedical RAG
Speak, Edit, Repeat: High-Fidelity Voice Editing and Zero-Shot TTS with Cross-Attentive MambaState-space sequence models (S4, Mamba) → Speech and audio SSMs
ReactDiff: Fundamental Multiple Appropriate Facial Reaction Diffusion ModelDiffusion models → Audio, speech and music
ID-Consistent, Precise Expression Generation with Blendshape-Guided DiffusionDiffusion models → Image generation, text-to-image and editing
AtomWorld: A Benchmark for Evaluating Spatial Reasoning in Large Language Models on Crystalline MaterialsLLM reasoning → Clinical and biomedical applications
Beyond Outcome Reward: Decoupling Search and Answering Improves LLM AgentsLLM agents and tool use → Web, search, and research agents
Multilingual Routing in Mixture-of-ExpertsMixture-of-experts → Expert routing strategies
Multi-Agent Tool-Integrated Policy OptimizationLLM agents and tool use → Reinforcement learning for LLM agents
ConceptSplit: Decoupled Multi-Concept Personalization of Diffusion Models via Token-wise Adaptation and Attention DisentanglementDiffusion models → Image generation, text-to-image and editing
EduPersona: Benchmarking Subjective Ability Boundaries of Virtual Student AgentsLLM agents and tool use → Educational agents and math problem solving
Predictive Feature Caching for Training-free Acceleration of Molecular Geometry GenerationDiffusion models → Flow matching, time series and science
QuantAgents: Towards Multi-agent Financial System via Simulated TradingLLM agents and tool use → Financial trading and forecasting agents

All 400