Papers

New on arXiv

Wed 8 Apr 2026

484 papers

GraphDx: A Cost-Aware Knowledge-Enhanced Multi-Agent Framework for Sequential DiagnosisLLM agents and tool use → Medical and scientific domain agents
Where Did It Go Wrong? Process-Level Evaluation of Web Agents with Semantic State TrackingLLM agents and tool use → Web, search, and research agents
From Context Shift to Stylistic Collapse: Why Training Objectives Matter More Than ScaleLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
LCO: LLM-based Constraint Optimization for Safer Agentic LLMs in Real-world TasksLLM agents and tool use → General LLM agent frameworks and benchmarks +1
ICG: Improving Cover Image Generation via MLLM-based Prompting and Personalized Preference AlignmentMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +2
NanoSpec: Accelerating Speculative Decoding using Minimalist In-Context VocabulariesEfficient LLM inference and serving → Speculative decoding
AI Slop or AI-enhancement? Student perceptions of AI-generated media for an English for Academic Purposes courseRetrieval-augmented generation → Agentic RAG, agents and memory
MoZoo:Unleashing Video Diffusion power in animal fur and muscle simulationDiffusion models → Video generation
When Safety Geometry Collapses: Fine-Tuning Vulnerabilities in Agentic Guard ModelsLLM agents and tool use → Agent threats, guardrails and governance +2
Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning
Prism: An Evolutionary Memory Substrate for Multi-Agent Open-Ended DiscoveryLLM agents and tool use → Agent memory systems
From Inheritance to Saturation: Disentangling the Evolution of Visual Redundancy for Architecture-Aware MLLM Inference AccelerationMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Modelling Gas-Phase Reaction Kinetics with Guided Particle Diffusion SamplingDiffusion models → Flow matching, time series and science
CROP: Token-Efficient Reasoning in Large Language Models via Regularized Prompt OptimizationLLM reasoning → Prompting and chain-of-thought reasoning
ConfigSpec: Profiling-Based Configuration Selection for Distributed Edge--Cloud Speculative LLM ServingEfficient LLM inference and serving → Early exit and edge inference
Agentic Compilation: Mitigating the LLM Rerun Crisis for Minimized-Inference-Cost Web AutomationLLM agents and tool use → Software engineering coding agents
LAST: Leveraging Tools as Hints to Enhance Spatial Reasoning for Multimodal Large Language ModelsLLM reasoning → Spatial and embodied reasoning +1
Semantic Intent Fragmentation: A Single-Shot Compositional Attack on Multi-Agent AI PipelinesLLM agents and tool use → Agent threats, guardrails and governance
From Business Events to Auditable Decisions: Ontology-Governed Graph Simulation for Enterprise AILLM agents and tool use → Agent threats, guardrails and governance
Optimal Decay Spectra for Linear RecurrencesState-space sequence models (S4, Mamba) → Core SSM architectures for language
PRIME: Training Free Proactive Reasoning via Iterative Memory Evolution for User-Centric AgentLLM agents and tool use → General LLM agent frameworks and benchmarks
VSAS-Bench: Real-Time Evaluation of Visual Streaming Assistant ModelsMultimodal LLMs / vision-language models → Video understanding with MLLMs
Program Analysis Guided LLM Agent for Proof-of-Concept GenerationLLM agents and tool use → Agent threats, guardrails and governance
DIVERSED: Relaxed Speculative Decoding via Dynamic Ensemble VerificationEfficient LLM inference and serving → Speculative decoding
Towards Real-Time Human-AI Musical Co-Performance: Accompaniment Generation with Latent Diffusion Models and MAX/MSPDiffusion models → Audio, speech and music
Bootstrapping Sign Language Annotations with Sign Language ModelsSynthetic data and self-improvement for LLMs → Self-rewarding alignment and self-training
ROZA Graphs: Self-Improving Near-Deterministic RAG through Evidence-Centric FeedbackLLM reasoning → Knowledge- and tool-augmented reasoning +2
Too long; didn't solveLLM reasoning → Mathematical reasoning
DCD: Domain-Oriented Design for Controlled Retrieval-Augmented GenerationRetrieval-augmented generation → RAG pipeline design and evaluation
COSMIC: Emotionally Intelligent Agents to Support Mental and Emotional Well-being in Extreme Isolation: Lessons from Analog Astronaut Training MissionsLLM agents and tool use → Medical and scientific domain agents
Reasoning-Based Refinement of Unsupervised Text Clusters with LLMsLLM reasoning → Knowledge- and tool-augmented reasoning
MCP-DPT: A Defense-Placement Taxonomy and Coverage Analysis for Model Context Protocol SecurityLLM agents and tool use → Agent threats, guardrails and governance
RL-ASL: A Dynamic Listening Optimization for TSCH Networks Using Reinforcement LearningDeep reinforcement learning methods → RL for energy, networks and operations
Decompose, Look, and Reason: Reinforced Latent Reasoning for VLMsLLM reasoning → Multimodal reasoning +1
ReflectRM: Boosting Generative Reward Models via Self-Reflection within a Unified Judgment FrameworkLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Beyond Human-Readable: Rethinking Software Engineering Conventions for the Agentic Development EraLLM reasoning → Reasoning for agents, planning and code +1
Triage: Routing Software Engineering Tasks to Cost-Effective LLM Tiers via Code Quality SignalsLLM agents and tool use → Software engineering coding agents
Enabling Intrinsic Reasoning over Dense Geospatial Embeddings with DFR-GemmaLLM reasoning → Spatial and embodied reasoning
CLEAR: Context Augmentation from Contrastive Learning of Experience via Agentic ReflectionLLM agents and tool use → Agent memory systems
Private Seeds, Public LLMs: Realistic and Privacy-Preserving Synthetic Data GenerationSynthetic data and self-improvement for LLMs → Privacy-preserving tabular data synthesis
ConsistRM: Improving Generative Reward Models via Consistency-Aware Self-TrainingLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
M-ArtAgent: Evidence-Based Multimodal Agent for Implicit Art Influence DiscoveryLLM agents and tool use → Multimodal and embodied agents
HY-Embodied-0.5: Embodied Foundation Models for Real-World AgentsLLM agents and tool use → Multimodal and embodied agents
GameWorld: Towards Standardized and Verifiable Evaluation of Multimodal Game AgentsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
Regret-Aware Policy Optimization: Environment-Level Memory for Replay Suppression under Delayed HarmDeep reinforcement learning methods → Deep RL algorithms and representations
GIRL: Generative Imagination Reinforcement Learning via Information-Theoretic Hallucination ControlDeep reinforcement learning methods → RL for robot locomotion and manipulation
Multimodal Large Language Models for Multi-Subject In-Context Image GenerationMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
SPAMoE: Spectrum-Aware Hybrid Operator Framework for Full-Waveform InversionMixture-of-experts → MoE theory and gating functions
ReAlign: Optimizing the Visual Document Retriever with Reasoning-Guided Fine-Grained AlignmentLLM reasoning → Multimodal reasoning
SubSearch: Intermediate Rewards for Unsupervised Guided Reasoning in Complex RetrievalLLM reasoning → Knowledge- and tool-augmented reasoning +1
FORGE: Fine-grained Multimodal Evaluation for Manufacturing ScenariosMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Reinforcement Learning with Reward Machines for Sleep Control in Mobile NetworksDeep reinforcement learning methods → RL for energy, networks and operations
Score Shocks: The Burgers Equation Structure of Diffusion Generative ModelsDiffusion models → Diffusion foundations and sampling
RefineRAG: Word-Level Poisoning Attacks via Retriever-Guided Text RefinementRetrieval-augmented generation → Privacy and security of RAG
SHIELD: A Segmented Hierarchical Memory Architecture for Energy-Efficient LLM Inference on Edge NPUsEfficient LLM inference and serving → Serving systems and request scheduling
A Physical Agentic Loop for Language-Guided Grasping with Execution-State MonitoringLLM agents and tool use → Multimodal and embodied agents
Event-Centric World Modeling with Memory-Augmented Retrieval for Embodied Decision-MakingWorld models → World models for navigation
Label Leakage Attacks in Machine Unlearning: A Parameter and Inversion-Based ApproachLLM safety and security → Machine unlearning
Evaluation as Evolution: Transforming Adversarial Diffusion into Closed-Loop Curricula for Autonomous VehiclesEnd-to-end autonomous driving → Adversarial scenarios and safety testing
Personalized RewardBench: Evaluating Reward Models with Human Aligned PersonalizationLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
ReCodeAgent: A Multi-agent Workflow for Language-Agnostic Translation and Validation of Large-Scale RepositoriesLLM agents and tool use → Software engineering coding agents
Appear2Meaning: A Cross-Cultural Benchmark for Structured Cultural Metadata Inference from ImagesMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Region-Graph Optimal Transport Routing for Mixture-of-Experts Whole-Slide Image ClassificationMixture-of-experts → Medical and biomedical MoE
Android Coach: Improve Online Agentic Training Efficiency with Single State Multiple ActionsLLM agents and tool use → GUI and mobile agents
A Systematic Study of Retrieval Pipeline Design for Retrieval-Augmented Medical Question AnsweringRetrieval-augmented generation → Medical and biomedical RAG
GenLCA: 3D Diffusion for Full-Body Avatars from In-the-Wild VideosDiffusion models → 3D, point cloud and scene generation
Joint Optimization of Reasoning and Dual-Memory for Self-Learning Diagnostic AgentLLM reasoning → Clinical and biomedical applications +1
BATON: A Multimodal Benchmark for Bidirectional Automation Transition Observation in Naturalistic DrivingEnd-to-end autonomous driving → Reasoning VLA and VLM driving policies
How Much Heavy Lifting Can an Agent Harness Do?: Measuring the LLM's Residual Role in a Planning AgentLLM agents and tool use → Software engineering coding agents
PhyEdit: Towards Real-World Object Manipulation via Physically-Grounded Image EditingWorld models → Embodied and robot world models
Robust Quadruped Locomotion via Evolutionary Reinforcement LearningDeep reinforcement learning methods → RL for robot locomotion and manipulation
TraceSafe: A Systematic Assessment of LLM Guardrails on Multi-Step Tool-Calling TrajectoriesLLM agents and tool use → Agent threats, guardrails and governance +1
HIVE: Query, Hypothesize, Verify An LLM Framework for Multimodal Reasoning-Intensive RetrievalLLM reasoning → Multimodal reasoning
VersaVogue: Visual Expert Orchestration and Preference Alignment for Unified Fashion SynthesisDiffusion models → 3D, point cloud and scene generation +1
INSPATIO-WORLD: A Real-Time 4D World Simulator via Spatiotemporal Autoregressive ModelingWorld models → Video-generation world models and simulators
Agent-Driven Corpus Linguistics: A Framework for Autonomous Linguistic DiscoveryLLM agents and tool use → General LLM agent frameworks and benchmarks
Smart Commander: A Hierarchical Reinforcement Learning Framework for Fleet-Level PHM Decision OptimizationDeep reinforcement learning methods → RL for robot navigation and UAV control
Critical Inker: Scaffolding Critical Thinking in AI-Assisted Writing Through Socratic QuestioningLLM reasoning → Reasoning benchmarks and evaluation
Reason in Chains, Learn in Trees: Self-Rectification and Grafting for Multi-turn Agent Policy OptimizationLLM reasoning → Reinforcement learning for reasoning +1
Multi-Turn Reasoning LLMs for Task Offloading in Mobile Edge ComputingLLM reasoning → Reasoning for agents, planning and code
Dynamic Context Evolution for Scalable Synthetic Data GenerationSynthetic data and self-improvement for LLMs → Model collapse from synthetic data
Learning to Search: A Decision-Based Agent for Knowledge-Based Visual Question AnsweringMultimodal LLMs / vision-language models → Document understanding and multimodal RAG +1
Autopoiesis: A Self-Evolving System Paradigm for LLM Serving Under Runtime DynamicsEfficient LLM inference and serving → Serving systems and request scheduling
Persona Matters: Effects of Activation Steering on Short Answer Generation and ScoringMixture-of-experts → MoE large language models +1
STRIDE-ED: A Strategy-Grounded Stepwise Reasoning Framework for Empathetic Dialogue SystemsLLM reasoning → Audio-language model reasoning
Flow Motion Policy: Manipulator Motion Planning with Flow Matching ModelsDiffusion models → Diffusion and flow policies for robotics
MARVEL: Multimodal Adaptive Reasoning-intensiVe Expand-rerank and retrievaLLLM reasoning → Multimodal reasoning
Epistemic Robust Offline Reinforcement LearningOffline RL and imitation learning from data → Offline RL algorithms
EVGeoQA: Benchmarking LLMs on Dynamic, Multi-Objective Geo-Spatial ExplorationLLM reasoning → Spatial and embodied reasoning
Production-Ready Automated ECU Calibration using Residual Reinforcement LearningDeep reinforcement learning methods → RL for autonomous driving and traffic
AV-SQL: Decomposing Complex Text-to-SQL Queries with Agentic ViewsLLM agents and tool use → Web, search, and research agents
ReDAct: Uncertainty-Aware Deferral for LLM AgentsLLM agents and tool use → Medical and scientific domain agents
Unified Deployment-Aware Evaluation of Open Reasoning Language ModelsLLM reasoning → Prompting and chain-of-thought reasoning
KITE: Keyframe-Indexed Tokenized Evidence for VLM-Based Robot Failure AnalysisMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
MoE Routing Testbed: Studying Expert Specialization and Routing Behavior at Small ScaleMixture-of-experts → Expert routing strategies
Strategic Persuasion with Trait-Conditioned Multi-Agent Systems for Iterative Legal ArgumentationLLM agents and tool use → Social simulation and multi-agent debate
Not all tokens contribute equally to diffusion learningDiffusion models → Diffusion foundations and sampling
MARS: Enabling Autoregressive Models Multi-Token GenerationEfficient LLM inference and serving → Speculative decoding
Predictive Representations for Skill Transfer in Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
DTCRS: Dynamic Tree Construction for Recursive SummarizationRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG

All 484