Papers

New on arXiv

Mon 29 Dec 2025

272 papers

An Agentic LLM Framework for Adverse Media Screening in AML ComplianceLLM agents and tool use → Agent threats, guardrails and governance
Temporal Attack Pattern Detection in Multi-Agent AI Workflows: An Open Framework for Training Trace-Based Security ModelsLLM agents and tool use → Agent threats, guardrails and governance
Breaking Audio Large Language Models by Attacking Only the Encoder: A Universal Targeted Latent-Space Audio AttackLLM reasoning → Audio-language model reasoning
CASCADE: Cumulative Agentic Skill Creation through Autonomous Development and EvolutionLLM agents and tool use → General LLM agent frameworks and benchmarks
From Illusion to Insight: Change-Aware File-Level Software Defect Prediction Using Agentic AILLM agents and tool use → Software engineering coding agents
Max-Entropy Reinforcement Learning with Flow Matching and A Case Study on LQRDiffusion models → Diffusion and flow policies for robotics +1
Learning to Feel the Future: DreamTacVLA for Contact-Rich ManipulationVision-language-action models → Tactile, force-aware and humanoid VLA
Yggdrasil: Bridging Dynamic Speculation and Static Runtime for Latency-Optimal Tree-Based LLM DecodingEfficient LLM inference and serving → Speculative decoding
Flow Matching Neural ProcessesDiffusion models → Flow matching, time series and science
Integrating Domain Knowledge for Financial QA: A Multi-Retriever RAG Approach with LLMsRetrieval-augmented generation → Legal and financial RAG
From Correctness to Collaboration: Toward a Human-Centered Framework for Evaluating AI Agent Behavior in Software EngineeringLLM agents and tool use → Software engineering coding agents
Adversarial Lens: Exploiting Attention Layers to Generate Adversarial Examples for EvaluationMechanistic interpretability → Steering and probing for LLM safety
Retrieval Augmented Question Answering: When Should LLMs Admit Ignorance?Retrieval-augmented generation → RAG robustness and adaptive retrieval
Improved Bounds for Private and Robust AlignmentLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
FineFT: Efficient and Risk-Aware Ensemble Reinforcement Learning for Futures TradingDeep reinforcement learning methods → Adversarial robustness and RL in finance
SB-TRPO: Towards Safe Reinforcement Learning with Hard ConstraintsDeep reinforcement learning methods → Safe and constrained RL
Entropy-Aware Token Rejection for Improving Speculative DecodingEfficient LLM inference and serving → Speculative decoding
Stream-DiffVSR: Low-Latency Streamable Video Super-Resolution via Auto-Regressive DiffusionDiffusion models → Inverse problems and image restoration
Diffusion Knows Transparency: Repurposing Video Diffusion for Transparent Object Depth and Normal EstimationDiffusion models → 3D, point cloud and scene generation
Robo-Dopamine: General Process Reward Modeling for High-Precision Robotic ManipulationLLM reasoning → Spatial and embodied reasoning +1
Bellman Calibration for V-Learning in Offline Reinforcement LearningOffline RL and imitation learning from data → Offline RL algorithms
Fine-Tuning LLMs with Fine-Grained Human Feedback on Text SpansLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Multilingual Hidden Prompt Injection Attacks on LLM-Based Academic ReviewingLLM safety and security → Prompt injection and agent security
Web World ModelsWorld models → World models for language agents
Nested Browser-Use Learning for Agentic Information SeekingLLM agents and tool use → Web, search, and research agents
Active Perception Agent for Omnimodal Audio-Video UnderstandingLLM agents and tool use → Multimodal and embodied agents
BOAD: Discovering Hierarchical Software Engineering Agents via Bandit OptimizationLLM reasoning → Reasoning for agents, planning and code +1
Close the Loop: Synthesizing Infinite Tool-Use Data via Multi-Agent Role-PlayingLLM agents and tool use → Tool learning and function calling
Same or Not? Enhancing Visual Perception in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
LiveTalk: Real-Time Multimodal Interactive Video Diffusion via Improved On-Policy DistillationDiffusion models → Video generation
Instruction-Following Evaluation of Large Vision-Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
ThinkGen: Generalized Thinking for Visual GenerationLLM reasoning → Multimodal reasoning +1
RxnBench: A Multimodal Benchmark for Evaluating Large Language Models on Chemical Reaction Understanding from Scientific LiteratureMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
VL-RouterBench: A Benchmark for Vision-Language Model RoutingMultimodal LLMs / vision-language models → Visual token pruning and efficiency
Toward Trustworthy Agentic AI: A Multimodal Framework for Preventing Prompt Injection AttacksLLM agents and tool use → Agent threats, guardrails and governance +1
PurifyGen: A Risk-Discrimination and Semantic-Purification Model for Safe Text-to-Image GenerationDiffusion models → Robustness, privacy and adversarial aspects
PathFound: An Agentic Multimodal Model Activating Evidence-seeking Pathological DiagnosisMultimodal LLMs / vision-language models → Medical vision-language models +1
Act2Goal: From World Model To General Goal-conditioned PolicyWorld models → Embodied and robot world models +1
Iterative Inference-time Scaling with Adaptive Frequency Steering for Image Super-ResolutionDiffusion models → Inverse problems and image restoration
Single LLM Debate, MoLaCE: Mixture of Latent Concept Experts Against Confirmation BiasLLM agents and tool use → Social simulation and multi-agent debate
Alpha-R1: Alpha Screening with LLM Reasoning via Reinforcement LearningLLM reasoning → Reasoning for agents, planning and code
Beyond Correctness: Exposing LLM-generated Logical Flaws in Reasoning via Multi-step Automated Theorem ProvingLLM reasoning → Reasoning benchmarks and evaluation
The Gaining Paths to Investment Success: Information-Driven LLM Graph Reasoning for Venture Capital PredictionLLM reasoning → Reasoning for agents, planning and code
Agentic AI for Autonomous Defense in Software Supply Chain Security: Beyond Provenance to Vulnerability MitigationLLM agents and tool use → Agent threats, guardrails and governance
HY-Motion 1.0: Scaling Flow Matching Models for Text-To-Motion GenerationDiffusion models → Video generation
Eliminating Inductive Bias in Reward Models with Information-Theoretic GuidanceLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Replay Failures as Successes: Sample-Efficient Reinforcement Learning for Instruction FollowingLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
CoFi-Dec: Hallucination-Resistant Decoding via Coarse-to-Fine Generative Feedback in Large Vision-Language ModelsMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
Dynamic Subspace Composition: Efficient Adaptation via Contractive Basis ExpansionMixture-of-experts → MoE theory and gating functions
Coupling Experts and Routers in Mixture-of-Experts via an Auxiliary LossMixture-of-experts → Expert routing strategies
ClinDEF: A Dynamic Evaluation Framework for Large Language Models in Clinical ReasoningLLM reasoning → Clinical and biomedical applications
Direct Diffusion Score Preference Optimization via Stepwise Contrastive Policy-Pair SupervisionDiffusion models → Diffusion and flow policies for robotics +1
AKG kernel Agent: A Multi-Agent Framework for Cross-Platform Kernel SynthesisLLM agents and tool use → Software engineering coding agents
DriveLaW:Unifying Planning and Video Generation in a Latent Driving WorldWorld models → World models for autonomous driving +1
MindWatcher: Toward Smarter Multimodal Tool-Integrated ReasoningLLM agents and tool use → Multimodal and embodied agents
SoulX-FlashTalk: Real-Time Infinite Streaming of Audio-Driven Avatars via Self-Correcting Bidirectional DistillationDiffusion models → Audio, speech and music
Diffusion priors enhanced velocity model building from time-lag images using a neural operatorDiffusion models → Flow matching, time series and science
AGRO-SQL: Agentic Group-Relative Optimization with High-Fidelity Data SynthesisLLM reasoning → Reinforcement learning for reasoning +1
SpatialMosaic: A Multiview VLM Dataset for Partial VisibilityLLM reasoning → Spatial and embodied reasoning +1
A Stepwise-Enhanced Reasoning Framework for Large Language Models Based on External Subgraph GenerationLLM reasoning → Knowledge- and tool-augmented reasoning
ISOPO: Proximal policy gradients without pi-oldDeep reinforcement learning methods → RL theory and sample complexity
AI Meets Brain: Memory Systems from Cognitive Neuroscience to Autonomous AgentsLLM agents and tool use → Agent memory systems
CubeBench: Diagnosing Interactive, Long-Horizon Spatial Reasoning Under Partial ObservationsLLM reasoning → Spatial and embodied reasoning
RobustMask: Certified Robustness against Adversarial Neural Ranking Attack via Randomized MaskingRetrieval-augmented generation → Privacy and security of RAG
MedGemma vs GPT-4: Open-Source and Proprietary Zero-shot Medical Disease Classification from ImagesMultimodal LLMs / vision-language models → Medical vision-language models
AI4Reading: Chinese Audiobook Interpretation System Based on Multi-Agent CollaborationLLM agents and tool use → Multimodal and embodied agents
Agentic Physical AI toward a Domain-Specific Foundation Model for Energy Systems: A Case Study on Nuclear Reactor ControlLLM reasoning → Spatial and embodied reasoning +1
Flow2GAN: Hybrid Flow Matching and GAN with Multi-Resolution Network for Few-step High-Fidelity Audio GenerationDiffusion models → Audio, speech and music
YOLO-Master: MOE-Accelerated with Specialized Transformers for Enhanced Real-time DetectionMixture-of-experts → Vision mixture-of-experts
On the Inverse Flow Matching Problem in the One-Dimensional and Gaussian CasesDiffusion models → Flow matching, time series and science
Interpretable Safety Alignment via SAE-Constructed Low-Rank Subspace AdaptationLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +2
Plug-and-Play Fidelity Optimization for Diffusion Transformer Acceleration via Cumulative Error MinimizationDiffusion models → Diffusion foundations and sampling
ASemConsist: Adaptive Semantic Feature Control for Training-Free Identity-Consistent GenerationDiffusion models → Image generation, text-to-image and editing
RAH-VLA: Resolution-Adaptive Hierarchical Vision-Language Alignment for Multimodal Remote Sensing UnderstandingMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
KernelEvolve: Scaling Agentic Kernel Coding for Heterogeneous AI Accelerators at MetaLLM agents and tool use → Software engineering coding agents
SURE Guided Posterior Sampling: Trajectory Correction for Diffusion-Based Inverse ProblemsDiffusion models → Inverse problems and image restoration
A Human-Oriented Cooperative Driving Approach: Integrating Driving Intention, State, and ConflictEnd-to-end autonomous driving → Cooperative and V2X driving
MM-UAVBench: How Well Do Multimodal Large Language Models See, Think, and Plan in Low-Altitude UAV Scenarios?LLM reasoning → Spatial and embodied reasoning +1
Scoring, Reasoning, and Selecting the Best! Ensembling Large Language Models via a Peer-Review ProcessLLM reasoning → Reasoning benchmarks and evaluation
Task-oriented Learnable Diffusion Timesteps for Universal Few-shot Learning of Dense TasksDiffusion models → Diffusion foundations and sampling
From Model Choice to Model Belief: Establishing a New Measure for LLM-Based ResearchSynthetic data and self-improvement for LLMs → Self-rewarding alignment and self-training
GaussianDWM: 3D Gaussian Driving World Model for Unified Scene Understanding and Multi-Modal GenerationWorld models → World models for autonomous driving +1
EquaCode: A Multi-Strategy Jailbreak Approach for Large Language Models via Equation Solving and Code CompletionLLM safety and security → Jailbreaking
Certifying the Right to Be Forgotten: Primal-Dual Optimization for Sample and Label Unlearning in Vertical Federated LearningLLM safety and security → Federated unlearning and poisoning
REVEALER: Reinforcement-Guided Visual Reasoning for Element-Level Text-Image Alignment EvaluationLLM reasoning → Multimodal reasoning +1
SPIRAL: Symbolic LLM Planning via Grounded and Reflective SearchLLM reasoning → Reasoning for agents, planning and code
Cosmos-H-Surgical: Learning Surgical Robot Policies from Videos via World ModelingWorld models → Medical and surgical world models +1
Multi-Agent Framework for Threat Mitigation and Resilience in AI-Based SystemsLLM agents and tool use → Agent threats, guardrails and governance +1
It's a TRAP! Task-Redirecting Agent Persuasion Benchmark for Web AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
InSPO: Unlocking Intrinsic Self-Reflection for LLM Preference OptimizationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Althea: The Fact-Checking--Metalearning Tradeoff in AI-Assisted Verification
BIRD: A Museum Open Dataset Combining Behavior Patterns and Identity Types to Better Model Visitors' Experience
Leveraging Adaptive Group Negotiation for Heterogeneous Multi-Robot Collaboration with Large Language Models
An Empirical Analysis of Fine-Tuning Large Language Models on Bioinformatics Literature: PRSGPT and BioStarsGPT
Discrete Semantic States and Hamiltonian Dynamics in LLM Embedding Spaces
Benchmarking Cross-Lingual Semantic Alignment in Multilingual Embeddings
Geometric Patterns of Meaning: A PHATE Manifold Analysis of Multi-lingual Embeddings
OptFormer: Optical Flow-Guided Attention and Phase Space Reconstruction for SST Forecasting
One if by Land, Two if by Sea, Three if by Four Seas, and More to Come -- Values of Perception, Prediction, Communication, and Common Sense in Decision Making
Jamming Detection in Cell-Free MIMO with Dynamic Graphs

All 272