Papers

New on arXiv

Fri 18 Sep 2026

388 papers

GPEC: Efficient Pre-LLM Gaussian Process Embedding Correction for Cardiac Video Caption GenerationMultimodal LLMs / vision-language models → Medical vision-language models
HeadGuard: Selective Head Protection for Low-Bit VLM KV-Cache QuantizationMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Video Captioning in Low-Light Conditions through Efficient Uncertainty-Aware Caption CorrectionMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Monotone-Constrained Diffusion Models for Long-Horizon Production ForecastingDiffusion models → Flow matching, time series and science
ConsistWorld: Evidence Routing for Consistent Multi-Agent World ModelsWorld models → Video-generation world models and simulators
VT-Bridge: Bridging Pretrained Foundation VLAs to VTLAs via Lightweight Residual AdaptationVision-language-action models → Tactile, force-aware and humanoid VLA
From Certain Doom to Survival: Agent-Driven Self-Governance in LLM Agent SocietiesLLM agents and tool use → Social simulation and multi-agent debate
AutoGym: Blueprint-First Generation of Verifiable Agent GymsLLM agents and tool use → Reinforcement learning for LLM agents
Seeing is not Enough: Vision-Language Models Perceive Evidence but Fail to ActMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
SEABED: SouthEast Asian Benchmark for Evaluating Audio ReasoningLLM reasoning → Audio-language model reasoning
Augmenting PID Control with Deep Reinforcement Learning: A Hybrid Approach to the Industrial BenchmarkDeep reinforcement learning methods → RL for energy, networks and operations
Zero-Trust Authorization and Discovery for Enterprise MCPLLM agents and tool use → Agent threats, guardrails and governance
Do Student LLMs Inherit OOD Robustness? Invariance-Weighted Distillation for Reliable Knowledge TransferKnowledge distillation → General knowledge distillation methods
Learning Control Policies from Heterogeneous Multi-Horizon Time Series in Battery Energy Management SystemsOffline RL and imitation learning from data → Decision transformers and sequence modeling
Defusing Explosive Prompts: Understanding and Preventing Trigger-Based Prompt Injections in LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Semantics Delivery Network: Rethinking Web Retrieval Infrastructure for LLM AgentsLLM reasoning → Knowledge- and tool-augmented reasoning +2
A Governance-Aware Large Language Model Orchestrated Agentic Digital Twin for Transmission System Operator Control Room Decision SupportLLM agents and tool use → Agentic AI for wireless networks
Efficient Mixture-of-Experts with Speculative Decoding via Expert CoactivationMixture-of-experts → Efficient MoE training and inference systems +1
VLPSA: Vision-Language-Poisson-Safe Actions for Full-Body Safety of Learned PoliciesVision-language-action models → Safety and adversarial attacks
SeeQ: Training Generalist Value Functions for Long-Horizon Robotic ManipulationVision-language-action models → Reinforcement learning and post-training
MintAct: A Unified Visual Agent for Digital EnvironmentsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
CodeMidas: Scaling Agentic Coding RL Environments from Code ItselfLLM agents and tool use → Software engineering coding agents
Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClawLLM agents and tool use → Social simulation and multi-agent debate
Benchmarking World Models for Continual Learning on Compositional TasksWorld models → Embodied and robot world models
Available Guardrails: Certifying Selective Prediction across ML SystemsLLM safety and security → Safety alignment, refusal and guardrails
An Interpretable Memory Decision Controller for LLM Agents Based on Three-Signal Complementarity: Decoupling Confidence and ConsistencyLLM agents and tool use → Agent memory systems
λ-Controlled GRPO: Turning Flow-Matching Ratio Instability into a Budgeted ResourceLLM reasoning → Diffusion reasoning and test-time scaling +1
PRIME: Perception Feedback with Situational Memory Embeddings in VLA ModelsVision-language-action models → VLA for autonomous driving
DiaVLo: Diagnosing Behaviours of Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
RecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use AgentsLLM agents and tool use → GUI and mobile agents
Bayesian Belief Layer for Controllable Opinion Dynamics in LLM AgentsLLM agents and tool use → Social simulation and multi-agent debate
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't RevealLLM safety and security → Machine unlearning
Time series generation with spectrally aligned latent flow matchingDiffusion models → Flow matching, time series and science
SkelWAM: A Skeleton-Guided World-Action Model for Zero-Shot Cross-Embodiment ManipulationWorld models → Embodied and robot world models
CARF: Contrastive Attraction-Repulsion of Failure-Guided Flow MatchingDiffusion models → Diffusion and flow policies for robotics
Schedule optimization for tau-leaping in masked discrete diffusionDiffusion models → Diffusion foundations and sampling
GALA: Geometry-Aware Latent Action Modeling for Vision-Language-Action Model Pretraining across EmbodimentsVision-language-action models → Human-video and latent-action pretraining
Learning to Move Cities: Deep Meta-Models and Reinforcement Policies for Calibration and Control in Urban NetworksDeep reinforcement learning methods → RL for autonomous driving and traffic
Beyond Kinematics: Benchmarking Simulation Fidelity for Muscle-Driven Imitation LearningOffline RL and imitation learning from data → Real-world robot offline RL +1
CommitFlow: Semantic Commitment Verification and Local Correction for Long-Horizon Robot Manipulation VLA ExecutionVision-language-action models → Failure recovery and robustness
Intervention Granularity Matters: Coherent Treatment Bundles in Counterfactual Simulation with Clinical World ModelsWorld models → Medical and surgical world models
ExpBoN: Exponential-Noise Best-of-n for Efficient Test-Time LLM AlignmentLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Benchmarking the Explanatory Quality of Open-Weight Vision-Language Models in Face RecognitionMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Watermarkable Multi-Draft Speculative Sampling via Poisson ProcessesEfficient LLM inference and serving → Speculative decoding
Do Personality-Tuned LLMs Make Better Social Agents?LLM agents and tool use → Social simulation and multi-agent debate
EnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the EnterpriseLLM agents and tool use → Software engineering coding agents
RheoSampling: Resolving the One-Hot Dilemma in Stochastic Dynamic-Tree Speculative DecodingEfficient LLM inference and serving → Speculative decoding
A Sim-to-Real Integration Pipeline for Training and Deployment of Chunk-Based VLA Manipulation PoliciesVision-language-action models → Human-video and latent-action pretraining
Classification-oriented adaptive sensing via posterior samplingDiffusion models → Inverse problems and image restoration
An Agentic Just-in-Time Adaptive Intervention System for Personalized Sleep Support: Proof-of-Concept Study with N of 1 DataLLM agents and tool use → Medical and scientific domain agents
CASCADE Against Jailbreaks: Combination Across Stages with Controlled Attack-Defense EvaluationLLM safety and security → Jailbreaking
AcousticDiffusion: Semantically Conditioned Audio-Guided Diffusion Policy for Search-and-Rescue AssistanceDiffusion models → Diffusion and flow policies for robotics
Compact but Moving: Intervention-Relevant Geometry in Recurrent World ModelsWorld models → Latent predictive world models
PointLAM: Local Attentive Mamba for Efficient Point-based 3D Object DetectionState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
Scaling Vision-Language Reward Learning for Robot Manipulation in Parallel SimulationMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Beyond Benchmark Scores: Auditing Medical Vision-Language Models for Chest X-Ray Tuberculosis ScreeningMultimodal LLMs / vision-language models → Medical vision-language models
GraphSkillEvo: Evolutionary Optimization of Graph-Structured Agent SkillsLLM agents and tool use → Agent skill learning and evolution
World Modeling in TransformersWorld models → World models for language agents
Sandwich-Residuals: Parameter-Efficient Test-time Adaptation of World ModelsWorld models → Latent predictive world models
GEM-MPC: Balancing Exploration and Exploitation through Expert-Guided PlanningDeep reinforcement learning methods → Deep RL algorithms and representations
ZYT-World: A Real-Time Controllable World Model for Closed-Loop Autonomous-Driving SimulationWorld models → World models for autonomous driving +1
SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive LLM InferenceEfficient LLM inference and serving → Speculative decoding
CIPL: A Channel-Aware Framework for Recoverable Privacy Leakage in LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance
GUARD: Natural Forgetting in Large Reasoning Models via Guided Answer-Reasoning DistillationLLM reasoning → Test-time compute and efficient thinking +2
DRT: Dense Reasoning Trace for Efficient and Grounded Multimodal ReasoningLLM reasoning → Multimodal reasoning +1
Accelerating Dense LLMs via L0-regularized Mixture-of-ExpertsMixture-of-experts → MoE large language models
Rethinking Human-Aligned Evaluation: An Analysis of Semantic Metrics Beyond WERLLM reasoning → Audio-language model reasoning
When Steering Fails in Latent Reasoning: A Latent-to-Language Transition GapLLM reasoning → Prompting and chain-of-thought reasoning +1
Outcome-Conditioned End-Effector Geometry Across Vision-Language-Action PoliciesVision-language-action models → Failure recovery and robustness
SynthDemo-RL: Breaking the Zero-Reward Barrier in VLA Adaptation with LLM-Guided Synthetic DemonstrationsVision-language-action models → Reinforcement learning and post-training
Calibrating Teacher--Student Discrepancy for On-Policy DistillationLLM reasoning → Reinforcement learning for reasoning +1
CounterPlay: Counterfactual Post-Training for Self-Play Driving PoliciesEnd-to-end autonomous driving → Imitation-learning end-to-end planners
Potential-Field Action Representation for Reinforcement Learning in Contact-Rich ManipulationDeep reinforcement learning methods → RL for robot locomotion and manipulation
Micro-Collaborative Poisoning: A Distributed Attack on RAG SystemsRetrieval-augmented generation → Privacy and security of RAG
GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State AssertionsLLM agents and tool use → Software engineering coding agents
On Repulsive and Attractive Teachers: Separating Correctness from Behavior in Self-DistillationKnowledge distillation → On-policy and self-distillation for LLMs
MIRAGE: Multi-Perspective Creative Language Model Reasoning with Reinforcement Learning GuidanceLLM reasoning → Reasoning for agents, planning and code
MACE: Memory-Agent Co-Evolution with Adaptive Memory Graphs for Multi-Agent SystemsLLM agents and tool use → Agent memory systems
OpenMAS-GCom. A Diagnostic Benchmark for Graph-enhanced Multi-Agent SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
IncentRL: The Trade-Off Between Preference Guidance and Task PerformanceDeep reinforcement learning methods → RL training of language and agent models
ServeGuard: Verifiable, Bounded-Residual Confinement of Operator-Invisible Channels Without Revealing the Certified Read FactorLLM safety and security → Jailbreaking
Skel-WAM: A Hand-Skeleton-Conditioned World Action Model for Human-to-Robot Manipulation TransferWorld models → Embodied and robot world models
The Communication Bottleneck: A Round-Trip Study of Tree-Structured Expression Serialization in Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Adaptive World Memory 3D Foundation Model for Scalable 3D Mapping, Localization, and RenderingLLM reasoning → Spatial and embodied reasoning
PolyBridgeBench: Benchmarking Multimodal LLMs for Physics-Grounded Bridge DesignMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
LogicTrack: Auditing Reasoning Trajectories of Large Language Models with Formal Logic SolversLLM reasoning → Test-time compute and efficient thinking
Driving on Registers, Reasoning on Risk: Risk-Aware Occupancy for Register-Based End-to-End Autonomous DrivingEnd-to-end autonomous driving → End-to-end system architecture
HE-Guardrail: A Homomorphic Guardrail Against Jailbreak Attacks for Encrypted Large Language Model InferenceLLM safety and security → Jailbreaking
Weave: Fine-Grained Dynamic SM Scheduling in an MoE Megakernel for Compute-Communication OverlapMixture-of-experts → Efficient MoE training and inference systems
Adaptive Rollout Truncation Based on Epistemic Uncertainty for Efficient Offline World Model TrainingWorld models → Latent dynamics models and model-based RL
MT-WAM: Reorienting the One-Pass Predictive Representation Toward Action GenerationDiffusion models → Video generation
Risk-Aware Occupancy for Safety-Oriented End-to-End Autonomous DrivingEnd-to-end autonomous driving → End-to-end system architecture
SkillIR: Evolving Scene-Aware Skills for Agentic Image RestorationLLM agents and tool use → Multimodal and embodied agents
CompAdapt: Adaptable Composite Motion Modeling for Physics-Consistent Text-to-Video GenerationDiffusion models → Video generation
ME-Dex 1.0: Bringing Heterogeneous Tactile Sensing into World Action ModelingWorld models → Embodied and robot world models
Benign Nonconvex Landscape for Policy Optimization: Infinite-Horizon Discounted MDPs with General State and Action SpacesDeep reinforcement learning methods → RL theory and sample complexity
GVPO++: Group Variance Policy Optimization for LLM Post-Training and On-Policy DistillationLLM reasoning → Reinforcement learning for reasoning +1
Quantization-Aware Kalman Estimation for Diffusion SamplingDiffusion models → Diffusion foundations and sampling
Stabilizing Trajectory Outputs in End-to-End Autonomous Driving via SC-IMM Based Teacher SignalsEnd-to-end autonomous driving → Imitation-learning end-to-end planners
WS-NeRF: A Mamba-Driven World-State-Aware Adaptive Deblurring Neural Radiance FieldState-space sequence models (S4, Mamba) → Mamba for image restoration

All 388