Papers

New on arXiv

Sat 19 Sep 2026

214 papers

Learning When to Refine: Long-Horizon Reinforcement Learning for Budgeted Neural-Operator PDE SolversDeep reinforcement learning methods → RL theory and sample complexity
Environment Steering: Using Data Flow Control to Improve Agent Utility and SafetyLLM agents and tool use → Agent threats, guardrails and governance
From Lexical Baselines to Agentic Retrieval-Augmented Generation: Structured Skill and Responsibility-Level Extraction with the SFIA FrameworkRetrieval-augmented generation → Agentic RAG, agents and memory +1
Be Careful Who You Trust: Coordination Dynamics under Corrupted Communication in LLM Multi-Agent GamesLLM agents and tool use → Social simulation and multi-agent debate
DiffVQE2: An Efficient Low-delay Diffusion Model for Acoustic Echo and Noise ControlDiffusion models → Audio, speech and music
Robot World Models Are Not Invariant to How the Actions Are WrittenWorld models → Embodied and robot world models
Triggers and Diagnostics for LLM-Based Interpretability Failures in Active Inference AgentsLLM agents and tool use → Agent threats, guardrails and governance
Euston: Training Away Mathematical Sycophancy Without Losing the MathematicsLLM reasoning → Mathematical reasoning
CausalWM: Causal Chain-of-Thought Reasoning for Embodied World ModelWorld models → Latent predictive world models +1
SparkDiffusion: Mitigating the High-Sparsity Trap --- A Unified Framework for up to 265× Single-GPU Acceleration of Visual GenerationDiffusion models → Video generation
CraftBench-UE: Deterministic Evaluation for Coding Agents in Unreal EngineLLM agents and tool use → Software engineering coding agents
QwenVLConnector: A Fast, Unified Medical VLM Chatbot for Fine-Grained Clinical Perception and Text GenerationMultimodal LLMs / vision-language models → Medical vision-language models
From Inference Engine to Inference Control Plane: Connecting vLLM, llm-d, and the Evolution of Efficient Distributed LLM ServingEfficient LLM inference and serving → Serving systems and request scheduling
Provably Efficient Reinforcement Learning in Continuous-Time Episodic MDPs with Poisson Decision EpochsDeep reinforcement learning methods → RL theory and sample complexity
MM-ContextFold: Context Folding for Multimodal Agentic RetrievalLLM agents and tool use → Multimodal and embodied agents
Verti-WM: A Physics-Aided Exteroceptive World Model for Off-Road Reinforcement LearningWorld models → World models for navigation +1
DiagGen: Agentic Generation of Deformable Assets with Sim-based Diagnostics for Robotic SimulationLLM agents and tool use → Multimodal and embodied agents
When Does Adversarial Refinement Help? A Negative Result and Open Problem in Adapting R3GAN to Time Series ImputationDiffusion models → Flow matching, time series and science
Measured Joules, Learned Routes: Learning to Route for Energy-Efficient LLM ServingLLM reasoning → Test-time compute and efficient thinking +2
From Concept Alignment to Causal Grounding: An Intervention Test of Chain-of-Thought FaithfulnessLLM reasoning → Prompting and chain-of-thought reasoning +1
LazyAgent: Demand-Driven Materialization and Physical Optimization of Agentic ProgramsLLM agents and tool use → Software engineering coding agents
Bridging Static and Agentic RAG for Taiwanese Historical Question AnsweringRetrieval-augmented generation → RAG robustness and adaptive retrieval +1
Optimizers for Diffusion Models: A Controlled BenchmarkDiffusion models → Discrete and text diffusion
Attributable Post-Rationalization in RAG Citations: A Controlled Reproduction and an RLVR ComparisonRetrieval-augmented generation → Agentic RAG, agents and memory
Spatial-Interactor: Learning Spatial Reasoning through Interaction with the Observable Physical WorldLLM reasoning → Spatial and embodied reasoning +2
Exploiting Residual Reachability for Cross-Model Migration of Graph-Based Indexes in Approximate Nearest Neighbor SearchRetrieval-augmented generation → Vector search and ANN indexes
WaveFront Decoding: Parallelized Self-Speculative Decoding for Looped Language ModelsEfficient LLM inference and serving → Speculative decoding
CrowdCue: Specialist-Cue Conditioning for Vision-Language Crowd CountingMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Dual-Locking Learned AI Models: A PIN-Based Sparse QIM Watermarking and Adaptive Index Permutation ApproachLLM safety and security → Watermarking generated content and models
Beyond Similarity: Coverage-Aware Prompt Selection for Time Series Forecasting with LLMsRetrieval-augmented generation → Time series and agricultural RAG
An Empirical Study and Open Testbed for Federated Fine-Tuning of Vision-Language-Action ModelsVision-language-action models → Reinforcement learning and post-training
When Agentic Trust Crosses Organizational Boundaries: Structural Externalization and a Reference Model for Trust EvidenceLLM agents and tool use → Agent threats, guardrails and governance
AgentRouter: Heterogeneous Model Routing for Cost-Optimal Multi-Step Agentic WorkflowsLLM agents and tool use → Software engineering coding agents
Beyond Single-Model Injection: A Threat Model and Defense Architecture for Prompt Injection in Multi-Agent SystemsLLM agents and tool use → Agent threats, guardrails and governance +1
RewardVerse: Rubric-Guided Policy Optimization for Video Reward ModelingLLM alignment / RLHF / preference optimisation → Reward hacking and misspecification +1
NostrAgent: A Decentralized Identity and Delegation Architecture for Sovereign Agentic SystemsLLM agents and tool use → Agent threats, guardrails and governance
An Evolutionary Agentic Approach for Open-ended Image Quality PerceptionLLM agents and tool use → Multimodal and embodied agents
"Dear LLaVA, Please Drive": A Depth-Aware Vision-Language Agent for Closed-Loop Robotic ControlMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
Computationally efficient safe exploration in reinforcement learningDeep reinforcement learning methods → Safe and constrained RL
H-VLA: Hierarchical Vision-Language-Action Model with Key-Action Reasoning and Motion Planning in a Unified Action SpaceLLM reasoning → Spatial and embodied reasoning +1
Stable and Efficient Real-World Online VLA Post-Training via Asynchronous Replay-Anchored Policy ImprovementVision-language-action models → Reinforcement learning and post-training
Prioritized Rollouts for Efficient World Model-based Vision-Language-Action Policy OptimizationWorld models → Latent dynamics models and model-based RL +2
ISA-Bench: A Benchmark for Computational Reasoning Across Instruction Set ArchitecturesLLM reasoning → Reasoning for agents, planning and code
Towards Full Pipeline FP8 Reinforcement Learning for LLMsLLM agents and tool use → Reinforcement learning for LLM agents
Planning-Aligned Pretraining of BEV Representations with Sparse Action-Conditioned Targets for End-to-End Autonomous DrivingEnd-to-end autonomous driving → End-to-end system architecture
Leveraging Inference-Time Compute for Diffusion Models via Global Scheduling of Denoising TrajectoriesLLM reasoning → Diffusion reasoning and test-time scaling +1
PileBelief: Persistent Physical State for Interaction-Driven World ModelingWorld models → Latent predictive world models
SmoLSTM: A Compact Vision-Language-Action Model with Recurrent Memory that PersistsVision-language-action models → Efficient VLA inference and adaptation
ForceRFT: Refining VLA Actions through Force-Guided Residual Reinforcement LearningVision-language-action models → Reinforcement learning and post-training
Personalized Federated Reinforcement Learning via Model-Agnostic Meta-Learning: Convergence of Exact and Hessian-Free Meta-Policy GradientsDeep reinforcement learning methods → RL for energy, networks and operations
The Price of Safety: Benign-Case Utility and Token Overhead of Memory-Poisoning Defenses in LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance
FIRM-WM: State-factorized factual-interventional recurrent modeling for reward-free visual planningWorld models → Embodied and robot world models
SelfOp: An Optimization Algorithm for Self-Improving Security AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Robust Market Making with Hawkes Order Flow and Price Impact via Adversarial Reinforcement LearningDeep reinforcement learning methods → Adversarial robustness and RL in finance
Look Before You Steer: Geometry Predicts SAE Feature SteerabilityMechanistic interpretability → SAE features and their applications
DriveReferee: Geometric Safety Verdicts Need Not Be Learned for Driving World-Action ModelsWorld models → World models for autonomous driving +1
D-IMPL: A Diffusion-based Solver for Parameterized BBOsDiffusion models → Diffusion foundations and sampling
ProcessLight: Process Supervision for Large Language Model Based Traffic Signal ControlLLM reasoning → Chain-of-thought reasoning for driving
BEACON: Belief-Enabled Adaptive CONtrol for Imitation Learning under UncertaintyOffline RL and imitation learning from data → Robot manipulation imitation learning
MATE: Policy-Aware Security Auditing for Mobile Agents via Synthesis-Driven Trajectory LearningLLM agents and tool use → Agent threats, guardrails and governance
From Code to Requirements: Agentic Reverse Engineering of Business Rules at Enterprise ScaleLLM agents and tool use → Software engineering coding agents
Trustworthy Agentic AI: Failure Modes, Mitigation Strategies, and a Lifecycle Framework for Autonomous LLM SystemsLLM agents and tool use → Agent threats, guardrails and governance +1
LLaDA-PRM: A Bidirectional Step-Level Reasoning EvaluatorLLM reasoning → Reasoning benchmarks and evaluation
COT-TTS: Audio Context-Aware Text-to-Speech with Chain-of-Thought ReasoningLLM reasoning → Audio-language model reasoning
StateMem: Single-State Residual Memory with Adaptive Inference for Vision-Language-Action PoliciesVision-language-action models → Efficient VLA inference and adaptation
From Capability to Assurance in Autonomous Penetration-Testing Harnesses: A Framework and Reference ImplementationLLM agents and tool use → Agent threats, guardrails and governance
Beyond Baseline Severity: Temporal and Disease-Specific Predictors of Depression Outcomes Following Mindfulness Interventions
Comparative review of hybrid forecasting models for short-term prediction of building thermal load
The Price of Greenwashing: Algorithmic Verification and Market Discipline using Conformal Machine Learning
Alignment Forecasting: Predicting Misalignment From Training Data
CLC-YOLO: A Compact Channel-Gated Prototype Network for Real-Time Leakage-Aware Breast Ultrasound Lesion Segmentation
Integrated Deep Learning Framework Designed on Hybrid Optimization Strategies for Automated Health Detection and Analysis in Silkworms
Modernising the Compressed-Domain Video Captioner: A Controlled Study of SigLIP2 and GPT-2 Substitutions
Normalise or condition? Noise-floor front-ends for on-board keyword spotting under UAV rotor ego-noise
RPA: Residual Patch-Token Adapter for Image Retrieval from EEG and MEG
Learning Stiffness Dependent Fluid Structure Dynamics from Coarse Flow Representations
An Accurate and Interpretable Hyper Graph Neural Network for GBM Survival Prediction
FREESIA: Covariance-Aware Posterior Transport for Expressive and Scalable Data Assimilation
The Price of Self-Calibration: Exact Evidence Budgets and Manufactured Blind Sets in Adaptive Monitoring
Exact Quotients of Fresnel-Kummer Surfaces and Certified Biaxial Refraction
SPACE: Semantic Projection and Alignment of CLIP Embeddings for Domain Adaptation
Proximal Residual Value Functions for Consistent Planning and Real-Time Execution
SoK: Formal Methods for Fact-Checking and Information Integrity
Auditing Bayesian Graph Alignment: Diagnostic Comparisons and Reference Failure
ChemCLIR-Bench: Benchmarking Cross-Lingual Information Retrieval in Multilingual Chemical Patents
Stealing profits: Spread-based temporal hierarchy forecasting for day-ahead electricity markets
Causal Inference with Unobserved Confounding: A Mixture Learning Perspective
Bayesian Deck-of-cards-based Ordinal Regression with Sequential Preference Elicitation
SDC-GON: Singular Decomposition and Consistency-Regularized Green's Operator Networks for Solving Partial Differential Equations
Do Not Trust the Benchmark: Limitations of General LLM Rankings and a Case for Task-Specific Evaluation
Enhancing speech representation learning with cross-modal knowledge transfer with HGNN under low resource settings: the case study of Yemba
LLMs as Linguistic Chameleons: Decoupling Semantics and Structure for Privacy-Preserving Communication
Low resource cross-modal alignment using HGNN to enhance speech representation
Neural Residual Modeling for Scientific Data Compression under Guaranteed Error Bounds
K-TRAIL: Simulator-Guided Generative Design of EM/RF Circuits
GrapeSplat: Geometry-Grounded Reconstruction via Amalgamated Pose-Free Encoding for Feed-Forward 3D Gaussian Splatting
Chronologic: Measuring Language Models' Ability to Represent the Past
Conformal Robustness in Prediction-Driven Decision-Making
UltraTex: Unleashing 2K Multi-View Diffusion for 3D Texturing
Signal-Informed Temporal Routing for Vinyl Defect Regime Detection

All 214