Papers

New on arXiv

Sun 14 Jun 2026

263 papers

Taming Visual Neglect: A Variational Information Bottleneck Framework for Adaptive Attention in Multimodal In-Context LearningMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
When Vocabulary Comprehension Fails Clinical Reasoning: Evaluating Therapy Bots' Safety Risks for Generation AlphaLLM reasoning → Clinical and biomedical applications
Decentralized Granular Access Control for Agentic AI Systems in Critical InfrastructureLLM agents and tool use → Agent threats, guardrails and governance
Tokengeist: Multi-Turn Attribution Tracing in Agentic ConversationsLLM agents and tool use → General LLM agent frameworks and benchmarks
Ablation, Statistical Inference, and Validation for KV-Cache CompressionEfficient LLM inference and serving → KV cache quantization
Beyond the Reranker: Do RAG Retrieval Enhancements Help Once a Strong Reranker Is Present?Retrieval-augmented generation → Generation and context integration
From Question Answering to Task Completion: A Survey on Agent System and Harness DesignLLM agents and tool use → General LLM agent frameworks and benchmarks
DynAMO:Dynamic Asset Management Orchestration via Topological Multi-Agent SchedulingLLM agents and tool use → Software engineering coding agents
Software Delegation Contracts: Measuring Reviewability in AI Coding-Agent WorkLLM agents and tool use → Software engineering coding agents
Mind-Studio: Executable World Models with Lookahead Evaluation for Partially Observable GamesWorld models → Video-generation world models and simulators
Stepwise Token Selection for Efficient Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
CloudDiffusion: Diffusion-Based Scene Completion in the Point Cloud DomainWorld models → World modeling from LiDAR generation +1
From Argument Components to Graphs: A Multi-Agent Debate with Confidence Gating for Argument RelationsLLM reasoning → Reasoning for agents, planning and code +1
Circuit Tracing in Autoregressive Protein Language ModelsMechanistic interpretability → Interpreting biomedical and science models
Open-SWE-Traces: Advancing Dual-Mode Multilingual Distillation for Software Engineering AgentsLLM agents and tool use → Software engineering coding agents
Theorem-Grounded Execution Ontologies for Interpretable Machine ReasoningLLM reasoning → Reasoning benchmarks and evaluation
GRACE-DS: a Guarded Reward-guided Agent Correction Environment in Data ScienceLLM agents and tool use → Web, search, and research agents
Agentic Framework for Deep Learning workload migration via In-Context LearningLLM agents and tool use → Software engineering coding agents
Mind the Gap: Diagnosing Constraint Discovery Failures in Text-in-Image EditingLLM reasoning → Multimodal reasoning
Formalize Once, Edit the Rest: Efficient Lean-Based Answer Selection for Math ReasoningLLM reasoning → Mathematical reasoning
SAG: SQL-Retrieval Augmented Generation with Query-Time Dynamic HyperedgesLLM reasoning → Knowledge- and tool-augmented reasoning +1
Green SARC: Predictive Cost and Carbon Governance for Agentic AI SystemsLLM agents and tool use → Financial trading and forecasting agents
DeepRoot: A KG-Coordinated Multi-Agent System for Therapeutic Reasoning over Historical Medical TextsLLM reasoning → Clinical and biomedical applications +1
ControlMap: Controllable High-Definition Map Generation for Traffic Scenario SimulationDiffusion models → Diffusion and flow policies for robotics
OmniOPSD: Rationale-Privileged On-Policy Self-Distillation for Affective ComputingMultimodal LLMs / vision-language models → Multimodal emotion understanding +1
Energy-Efficient Arm Reaching for a Humanoid Robot via Deep Reinforcement Learning with Identified Power ModelsDeep reinforcement learning methods → RL for robot locomotion and manipulation
Reinforcement Learning for LLM-based Event ForecastingLLM reasoning → Reinforcement learning for reasoning
On-Policy Distillation with Curriculum Turn-level Guidance for Multi-turn AgentsKnowledge distillation → On-policy and self-distillation for LLMs
Interactor: Agentic RL oriented Iterative Creation for Ad Description Generation in Sponsored SearchLLM agents and tool use → Web, search, and research agents
Calibrated Triage, Not Autonomy: Confidence Estimation for Medical Vision-Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
GeoTLM: Geometry-aware Tactile-Language Models for Contact Motion Orientation Reasoning of Dynamic ObjectsLLM reasoning → Spatial and embodied reasoning
Control-Plane Placement Shapes Forgetting: An Architectural Study of Agent Memory Across Thirteen System ConfigurationsLLM agents and tool use → Agent memory systems
SkillVetBench: LLM-as-Judge for Multi-Dimensional Security Risk Evaluation in Open-Source LLM Agent SkillsLLM agents and tool use → Agent threats, guardrails and governance
Topological Flow MatchingDiffusion models → Flow matching, time series and science
UrbanWell: Benchmarking Multimodal Large Language Models for Spatio-Temporal Urban Wellbeing AnalyticsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Neuron Level Analysis of Large Language Model in Legal Domain ReasoningLLM reasoning → Reasoning for agents, planning and code
Deep Residual Injection for Full-Spectrum Forensic Signal Perception in Multimodal Large Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
Free Energy Heuristics: Fast-And-Frugal Cognition as Active Inference Under Uncertain PrecisionLLM reasoning → Test-time compute and efficient thinking
LLM-as-Code: Agentic Programming for Agent HarnessLLM agents and tool use → Software engineering coding agents
SciOrch: Learning to Orchestrate Expert LLMs for Solving Frontier Multimodal Scientific Reasoning TasksLLM reasoning → Knowledge- and tool-augmented reasoning
Metis: A Generalizable and Efficient World-Action Model for Autonomous Driving and Urban NavigationWorld models → World models for autonomous driving +1
CogCanvas: A Benchmark for Evaluating Multi-Subject Reference-Based Image GenerationDiffusion models → Image generation, text-to-image and editing
STRIDE: Strategic Trajectory Reasoning via Discriminative Estimation for Verifiable Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning
RetailBench: Benchmarking long horizon reasoning and coherent decision making of LLM agents in realistic retail environmentsLLM reasoning → Reasoning for agents, planning and code +1
Object Tokens as a Bridge Between Segmentation and Visual Question Answering in Robotic SurgeryMultimodal LLMs / vision-language models → Medical vision-language models
FlashNav: Training Deployable Robot Navigation Policies in SecondsDeep reinforcement learning methods → RL for robot navigation and UAV control
Multi-Stage Stochastic Optimization and Reinforcement Learning Approaches for Dynamic Inspection of Infrastructure SystemsDeep reinforcement learning methods → RL for energy, networks and operations
Intelligent Multimodal Retrieval and Reasoning for Geospatial Knowledge Discovery on the I-GUIDE PlatformLLM reasoning → Spatial and embodied reasoning +1
Wasserstein Convergence of ODE-Based Samplers in Decentralized Diffusion Model via Velocity Field DecompositionDiffusion models → Diffusion foundations and sampling
Configuration Smells in AGENTS.md Files: Common Mistakes in Configuring Coding AgentsLLM agents and tool use → Software engineering coding agents
TrustedARI: Towards Trust-Native Agentic Routing Infrastructure for Agentic AILLM agents and tool use → Agent threats, guardrails and governance
The Truth Stays in the Family: Enhancing Contextual Grounding via Inherited Truthful Heads in Model LineagesMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
SACE: Concept Erasure at the Semantic Singularity in Visual Autoregressive ModelsLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment +1
Mean-Field Parallel Decoding for Discrete Diffusion Language ModelsDiffusion models → Discrete and text diffusion +1
DifFRACT: Diffusion Feature Reconstruction and Attribution for Circuit TracingDiffusion models → Diffusion foundations and sampling +1
Proximal Policy Optimization for Amortized Discrete SamplingDeep reinforcement learning methods → RL with diffusion and flow models
GAS-Leak-LLM: Genetic Algorithm-Based Suffix Optimization for Black-Box LLM JailbreakingLLM safety and security → Jailbreaking
Mitigating Visual Hallucinations in Multimodal Systems through Retrieval-Augmented Reliability-Aware InferenceMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs +1
Faithful Action-unit Causal Reasoning for Counterfactually Faithful Emotion ExplanationsLLM reasoning → Audio-language model reasoning
ttda704 at SemEval-2026 Task 6: Structured Chain-of-Thought Prompting for Political Evasion DetectionLLM reasoning → Prompting and chain-of-thought reasoning
LaWAM: Latent World Action Models for Efficient Dynamics-Aware Robot PoliciesWorld models → Embodied and robot world models +2
Snyk VulnBench JS 1.0: Can LLMs Find the Same Bugs Twice?LLM agents and tool use → Agent threats, guardrails and governance
From Correlation to Causation in Lane Change Prediction for Automated Driving: A Causal Explanation FrameworkEnd-to-end autonomous driving → End-to-end system architecture
RoboPIN: Grounded Embodied Reasoning via Pinned Chain-of-ThoughtLLM reasoning → Spatial and embodied reasoning
OmniTraffic: A Controllable Generation Pipeline and Benchmark for Spatio-Temporal Traffic ReasoningLLM reasoning → Chain-of-thought reasoning for driving
A Self Consistency Based Reranking for Narrative Question AnsweringLLM reasoning → Test-time compute and efficient thinking
Vernier: Probing Representational Misalignment Behind Lexical Gaps in Causal ReasoningLLM reasoning → Causal reasoning in LLMs
How to Score Experts for One-Shot MoE Expert Pruning: A Unified Formulation and Selection PrincipleMixture-of-experts → MoE large language models
Beyond English: Uncovering the Multilingual Gap in Vision-Language-Action ModelsVision-language-action models → Generalist VLA models and benchmarks
MAF: Multimodal Adaptive Few-shot Prompting for Sentiment Analysis with MLLMsMultimodal LLMs / vision-language models → Multimodal emotion understanding
Multi-agent Framework for Time-Sensitive Complementary Collaboration in MinecraftLLM agents and tool use → Reinforcement learning for LLM agents
ReQAT: Achieving Full-Precision Reasoning Accuracy with 4-bit Floating-Point Quantization-Aware TrainingLLM reasoning → Test-time compute and efficient thinking +1
Self-Questioning Vision-Language Models: Reinforcement Learning for Compositional Visual ReasoningLLM reasoning → Multimodal reasoning +1
Multi-Agent Framework for Audit Risk Assessment with Explicit Uncertainty and Evidence Conflict ModelingLLM agents and tool use → Financial trading and forecasting agents
MambAdapter: Lightweight Mamba-Based Adapters for Parameter-Efficient Transfer Learning in Speech and AudioState-space sequence models (S4, Mamba) → Speech and audio SSMs
Formalizing and Mitigating Structural Distortion in LLM Attention for Graph ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning
Retrieve, Don't Retrain: Extending Vision Language Action Models to New Tasks at Test TimeVision-language-action models → Reinforcement learning and post-training
Conflict-Aware Federated Fine-Tuning of Large Language Models with Mixture-of-ExpertsMixture-of-experts → Federated mixture-of-experts
NeRD: Neuro-Symbolic Rule Distillation for Efficient Ontology-Grounded Chain-of-Thought in Medical Image DiagnosisLLM reasoning → Clinical and biomedical applications
MoECa: Aligning Feature Reuse with Expert Decomposition in Diffusion TransformersDiffusion models → Diffusion foundations and sampling +1
Variational Test-time Optimization for Diffusion SynchronizationDiffusion models → Diffusion foundations and sampling
FragFuse: Bypassing Access Control of Large Language Model Agents via Memory-Based Query Fragmentation and FusionLLM agents and tool use → Agent threats, guardrails and governance
On the Adversarial Robustness of Multimodal LLM JudgesMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
Integrating Reasoning and Generalization in Text-to-SQL via Self-Enhanced Fine-TuningLLM reasoning → Knowledge- and tool-augmented reasoning
Pixels to Proofs: Probabilistically-Safe Latent World Model Control via Parallel Conformal Robust MPCWorld models → Latent dynamics models and model-based RL
Agentic Retrieval and Reinforcement Learned Equation Chains: A Controlled Generation Framework for Complex and Novel Physics Word ProblemsLLM agents and tool use → Educational agents and math problem solving
Unlocking Diffusion Hierarchies: Adaptive Timestep Selection for Zero-Shot SegmentationDiffusion models → Medical imaging and segmentation
Is Code Better Than Language for Algorithmic ReasoningLLM reasoning → Reasoning for agents, planning and code +1
Perfect Demo Makes Poor Teacher: Learning Robust Alignment from Critical Motion SegmentsOffline RL and imitation learning from data → Robot manipulation imitation learning
Localizing Credit at the Divergence: Path-Conditioned Self-Distillation for LLM ReasoningLLM reasoning → Reinforcement learning for reasoning +1
SAPS: Shared Autonomy for Policy Steering by Blending Teleoperation with a Pretrained VLAVision-language-action models → Reinforcement learning and post-training
Service-Induced Congestion in Memory-Constrained LLM ServingEfficient LLM inference and serving → Serving systems and request scheduling
Robots as Tokens: Unified Diffusion Transformer for Coordinated Multi-Robot Trajectory GenerationDiffusion models → Diffusion and flow policies for robotics
One Goal, Many Commands: Characterizing Denylist Fragility in AI AgentsLLM agents and tool use → Agent threats, guardrails and governance
EIBench: A Simulator-Based Benchmark and Turn-Credit RL for Emotion ManagementLLM reasoning → Audio-language model reasoning
Greedy Coordinate Diffusion: Effective and Semantically Coherent Adversarial Attacks via Diffusion GuidanceLLM safety and security → Jailbreaking
SHARD: Safe and Helpful Alignment via Self-Reframing DistillationLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks
Equivariant Cellular Sheaves for Molecular Electronic Structure: Bridging Sheaf Cohomology and E(3)-Equivariant Hamiltonian Learning
Building and Evaluating a Synthetic Bengali Speech Resource for Telecom Customer Care
Evaluating LLMs as Interpretable Controllers for Dynamical Systems

All 263