Papers

New on arXiv

Mon 19 Jan 2026

331 papers

PRIME: Policy-Reinforced Iterative Multi-agent Execution for Algorithmic Reasoning in Large Language ModelsLLM reasoning → Reasoning for agents, planning and code +1
Analyzing Diffusion and Autoregressive Vision Language Models in Multimodal Embedding SpaceMultimodal LLMs / vision-language models → Image editing and quality assessment +1
Distributional Reinforcement Learning for Condition-Based Maintenance of Multi-Pump EquipmentDeep reinforcement learning methods → RL for energy, networks and operations
Enhancing few-shot time series forecasting with LLM-guided diffusionDiffusion models → Flow matching, time series and science
LSSF: Safety Alignment for Large Language Models through Low-Rank Safety Subspace FusionLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
In Vino Veritas and Vulnerabilities: Examining LLM Safety via Drunk Language InducementLLM safety and security → Jailbreaking
CORVUS: Red-Teaming Hallucination Detectors via Internal Signal Camouflage in Large Language ModelsLLM safety and security → Machine unlearning
PhysicsSolutionAgent: Towards Multimodal Explanations for Numerical Physics Problem SolvingLLM reasoning → Spatial and embodied reasoning
Analyzing VLM-Based Approaches for Anomaly Classification and SegmentationMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Who Endorsed It? Measuring Authority Bias Across Expertise Levels in Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Diffusion Representations for Fine-Grained Image Classification: A Marine Plankton Case StudyDiffusion models → Diffusion foundations and sampling
Reasoning with Pixel-level Precision: QVLM Architecture and SQuID Dataset for Quantitative Geospatial AnalyticsLLM reasoning → Spatial and embodied reasoning +1
Beyond Memorization: Testing LLM Reasoning on Unseen Theory of Computation TasksLLM reasoning → Reasoning benchmarks and evaluation
Confidence over Time: Confidence Calibration with Temporal Logic for Large Language Model ReasoningLLM reasoning → Test-time compute and efficient thinking
A Lightweight Modular Framework for Constructing Autonomous Agents Driven by Large Language Models: Design, Implementation, and Applications in AgentForgeLLM agents and tool use → General LLM agent frameworks and benchmarks
Recurrent Confidence Chain: Temporal-Aware Uncertainty Quantification in Large Language ModelsLLM reasoning → Test-time compute and efficient thinking
Sockpuppetting: Jailbreaking LLMs by Combining Prefilling with OptimizationLLM safety and security → Jailbreaking
The Geometry of Thought: How Scale Restructures Reasoning In Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
CausalSpatial: A Benchmark for Object-Centric Causal Spatial ReasoningLLM reasoning → Spatial and embodied reasoning
CooperBench: Why Coding Agents Cannot be Your Teammates YetLLM agents and tool use → General LLM agent frameworks and benchmarks
Improving the Safety and Trustworthiness of Medical AI via Multi-Agent Evaluation LoopsLLM agents and tool use → Medical and scientific domain agents
Unlearning in LLMs: Methods, Evaluation, and Open ChallengesLLM safety and security → Machine unlearning
CURE-Med: Curriculum-Informed Reinforcement Learning for Multilingual Medical ReasoningLLM reasoning → Clinical and biomedical applications
Aligning Agentic World Models via Knowledgeable Experience LearningWorld models → World models for language agents +1
A Comprehensive Evaluation of LLM Reasoning: From Single-Model to Multi-Agent ParadigmsLLM reasoning → Reasoning for agents, planning and code +1
A Semantic Decoupling-Based Two-Stage Rainy-Day Attack for Revealing Weather Robustness Deficiencies in Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
ConvMambaNet: A Hybrid CNN-Mamba State Space Architecture for Accurate and Real-Time EEG Seizure DetectionState-space sequence models (S4, Mamba) → Physiological signal modeling
Autoregressive Models Rival Diffusion Models at ANY-ORDER GenerationDiffusion models → Discrete and text diffusion
Insider Knowledge: How Much Can RAG Systems Gain from Evaluation Secrets?Retrieval-augmented generation → RAG pipeline design and evaluation
Incorporating Q&A Nuggets into Retrieval-Augmented GenerationRetrieval-augmented generation → RAG robustness and adaptive retrieval
Beyond Single-shot Writing: Deep Research Agents are Unreliable at Multi-turn Report RevisionLLM agents and tool use → Web, search, and research agents
Diffusion-Driven Synthetic Tabular Data Generation for Enhanced DoS/DDoS Attack ClassificationDiffusion models → Flow matching, time series and science
Prompt Injection Mitigation with Agentic AI, Nested Learning, and AI Sustainability via Semantic CachingLLM agents and tool use → Agent threats, guardrails and governance +1
OpenExempt: A Diagnostic Benchmark for Legal Reasoning and a Framework for Creating Custom Benchmarks on DemandLLM reasoning → Reasoning for agents, planning and code
Probe and Skip: Self-Predictive Token Skipping for Efficient Long-Context LLM InferenceLLM reasoning → Test-time compute and efficient thinking
FastAV: Efficient Token Pruning for Audio-Visual Large Language Model InferenceEfficient LLM inference and serving → Visual token pruning in multimodal models
TVWorld: Foundations for Remote-Control TV AgentsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
AMDM-SE: Attention-based Multichannel Diffusion Model for Speech EnhancementDiffusion models → Audio, speech and music
CLIP-Guided Adaptable Self-Supervised Learning for Human-Centric Visual TasksMixture-of-experts → Vision mixture-of-experts
SplatReasoner: Enhancing Embodied Reasoning and Grounding by Novel View SynthesisLLM reasoning → Spatial and embodied reasoning +1
PhaseMark: A Post-hoc, Optimization-Free Watermarking of AI-generated Images in the Latent Frequency DomainDiffusion models → Robustness, privacy and adversarial aspects +1
Agentic Conversational Search with Contextualized Reasoning via Reinforcement LearningLLM reasoning → Knowledge- and tool-augmented reasoning +1
IntAgent: NWDAF-Based Intent LLM Agent Towards Advanced Next Generation NetworksLLM agents and tool use → Agentic AI for wireless networks
CODE: A Contradiction-Based Deliberation Extension Framework for Overthinking Attacks on Retrieval-Augmented GenerationRetrieval-augmented generation → Privacy and security of RAG
Leveraging Lora Fine-Tuning and Knowledge Bases for Construction IdentificationRetrieval-augmented generation → RAG pipeline design and evaluation
Recursive Meta-Distillation: An Axiomatic Framework for Iterative Knowledge RefinementKnowledge distillation → General knowledge distillation methods
METIS: Mentoring Engine for Thoughtful Inquiry & SolutionsLLM reasoning → Reasoning benchmarks and evaluation
MagicGUI-RMS: A Multi-Agent Reward Model System for Self-Evolving GUI Agents via Automated Feedback RefluxLLM agents and tool use → GUI and mobile agents +1
Static Is Not Enough: A Comparative Study of VR and SpaceMouse in Static and Dynamic Teleoperation TasksOffline RL and imitation learning from data → Robot manipulation imitation learning
Complex Aircraft Maneuvering using Reinforcement-Learning-Augmented Sliding Mode-Based ControlDeep reinforcement learning methods → RL for robot navigation and UAV control
Think3D: Thinking with Space for Spatial ReasoningLLM reasoning → Spatial and embodied reasoning +1
PASs-MoE: Mitigating Misaligned Co-drift among Router and Experts via Pathway Activation Subspaces for Continual LearningMixture-of-experts → Mixture of LoRA experts
OFA-MAS: One-for-All Multi-Agent System Topology Design based on Mixture-of-Experts Graph Generative ModelsLLM agents and tool use → General LLM agent frameworks and benchmarks +1
Graph Reasoning Paradigm: Structured and Symbolic Reasoning with Topology-Aware Reinforcement Learning for Large Language ModelsLLM reasoning → Knowledge- and tool-augmented reasoning
Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment GeneralizationVision-language-action models → Human-video and latent-action pretraining
RAGExplorer: A Visual Analytics System for the Comparative Diagnosis of RAG SystemsRetrieval-augmented generation → RAG pipeline design and evaluation
PaperGuide: Making Small Language-Model Paper-Reading Agents More EfficientLLM agents and tool use → Reinforcement learning for LLM agents
The Bitter Lesson of Diffusion Language Models for Agentic Workflows: A Comprehensive Reality CheckDiffusion models → Discrete and text diffusion +1
Sutradhara: An Intelligent Orchestrator-Engine Co-design for Tool-based Agentic InferenceEfficient LLM inference and serving → Efficient reasoning and agent inference
Deterministic Dynamics of Sampling Processes in Score-Based Diffusion Models with Multiplicative Noise ConditioningDiffusion models → Diffusion foundations and sampling
StyMam: A Mamba-Based Generator for Artistic Style TransferState-space sequence models (S4, Mamba) → Video, motion and generative Mamba models
Imitation learning-based spacecraft rendezvous and docking method with Expert DemonstrationOffline RL and imitation learning from data → Imitation learning methods
ImmersiveFlow: Stereo-to-7.1.4 spatial audio generation with flow matchingDiffusion models → Audio, speech and music
Active Inference-Driven World Modeling for Adaptive UAV Swarm Trajectory DesignWorld models → World models for navigation
On the Evidentiary Limits of Membership Inference for Copyright AuditingLLM safety and security → Membership inference attacks
Membership Inference Test: Auditing Training Data in Object Classification ModelsLLM safety and security → Membership inference attacks
ForeDiffusion: Foresight-Conditioned Diffusion Policy via Future View Construction for Robot ManipulationDiffusion models → Diffusion and flow policies for robotics
SciCoQA: Quality Assurance for Scientific Paper--Code AlignmentLLM agents and tool use → Web, search, and research agents
From Prefix Cache to Fusion RAG Cache: Accelerating LLM Inference in Retrieval-Augmented GenerationRetrieval-augmented generation → Generation and context integration +1
Sparse ActionGen: Accelerating Diffusion Policy with Real-time PruningDiffusion models → Diffusion and flow policies for robotics
Hierarchical Sparse Circuit Extraction from Billion-Parameter Language Models through Scalable Attribution Graph DecompositionMechanistic interpretability → Circuits and mechanistic analysis
Proxy Robustness in Vision Language Models is Effortlessly TransferableMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
SCULPT: Constraint-Guided Pruned MCTS that Carves Efficient Paths for Mathematical ReasoningLLM reasoning → Mathematical reasoning +1
Seeing Isn't Always Believing: Analysis of Grad-CAM Faithfulness and Localization Reliability in Lung Cancer CT ClassificationLLM reasoning → Clinical and biomedical applications
MirrorGuard: Toward Secure Computer-Use Agents via Simulation-to-Real Reasoning CorrectionLLM agents and tool use → Agent threats, guardrails and governance
Left-Right Symmetry Breaking in CLIP-style Vision-Language Models Trained on Synthetic Spatial-Relation DataMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Semi-supervised Instruction Tuning for Large Language Models on Text-Attributed GraphsLLM reasoning → Knowledge- and tool-augmented reasoning
FRoM-W1: Towards General Humanoid Whole-Body Control with Language InstructionsDeep reinforcement learning methods → RL for robot locomotion and manipulation
PhyG-MoE: A Physics-Guided Mixture-of-Experts Framework for Energy-Efficient GNSS Interference RecognitionMixture-of-experts → MoE theory and gating functions
VIRO: Robust and Efficient Neuro-Symbolic Reasoning with Verification for Referring Expression ComprehensionLLM reasoning → Multimodal reasoning
Teaching LLMs to Learn Tool Trialing and Execution through Environment InteractionLLM agents and tool use → Tool learning and function calling
Moaw: Unleashing Motion Awareness for Video Diffusion ModelsDiffusion models → Video generation
VISPA: Pluralistic Alignment via Automatic Value Selection and ActivationLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
PAIR-SAFE: A Paired-Agent Approach for Runtime Auditing and Refining AI-Mediated Mental Health SupportLLM agents and tool use → Medical and scientific domain agents
Towards Robust Process Reward Modeling via Noise-aware LearningLLM reasoning → Reinforcement learning for reasoning +1
Vision Language Models for Optimization-Driven Intent Processing in Autonomous NetworksMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
AirHunt: Bridging VLM Semantics and Continuous Planning for Efficient Aerial Object NavigationMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
A Shared Geometry of Difficulty in Multilingual Language ModelsMechanistic interpretability → Knowledge and representations inside LLMs
Teaching Large Reasoning Models Effective ReflectionLLM reasoning → Test-time compute and efficient thinking
S2DiT: Sandwich Diffusion Transformer for Mobile Streaming Video GenerationDiffusion models → Video generation
Decoding Rewards in Competitive Games: Inverse Game Theory with Entropy RegularizationOffline RL and imitation learning from data → Inverse reinforcement learning +1
Towards Spectroscopy: Susceptibility Clusters in Language ModelsMechanistic interpretability → Sparse autoencoder training and evaluation
MetaToolAgent: Towards Generalizable Tool Usage in LLMs through Meta-LearningLLM agents and tool use → Tool learning and function calling
VILTA: A VLM-in-the-Loop Adversary for Enhancing Driving Policy RobustnessEnd-to-end autonomous driving → Adversarial scenarios and safety testing
Augmenting Question Answering with A Hybrid RAG ApproachRetrieval-augmented generation → RAG robustness and adaptive retrieval
Ethical Risks in Deploying Large Language Models: An Evaluation of Medical Ethics JailbreakingLLM safety and security → Safety alignment, refusal and guardrails
Objective Matters: Fine-Tuning Objectives Shape Safety, Robustness, and Persona DriftLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks
Topology-Aware Multiscale Mixture of Experts for Efficient Molecular Property PredictionMixture-of-experts → MoE theory and gating functions
Convolutional Surrogate for 3D Discrete Fracture-Matrix Tensor Upscaling
ReCQR: Incorporating conversational query rewriting to improve Multimodal Image Retrieval

All 331