Papers

New on arXiv

Tue 21 Apr 2026

433 papers

ISEE: Interactive Semantic Enrichment for Database FieldsLLM agents and tool use → Web, search, and research agents
Detecting AI-Generated Content on Social Media with Multi-modal Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
NightFeats @ MMU-RAGent NeurIPS 2025: A Context-Optimized Multi-Agent RAG System for the Text-to-Text TrackRetrieval-augmented generation → Agentic RAG, agents and memory +1
The Structural Attention Tax: How Retrieval Format Hijacks In-Context Learning Independent of ContentRetrieval-augmented generation → RAG robustness and adaptive retrieval
Bridging Traditional Explainability Methods and Multimodal Multilingual Models: An XAI-Based AnalysisMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Durable Evaluation Framework: Adversarial Arbitration for Sycophancy Reduction in Large Language ModelsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks
mllm-shap: A Shapley Value Explainability Platform for Text-Audio Multimodal Large Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
CAPruner: Conceptual-Adjacent Scene Graph Pruner for Enhancing 3D Spatial Reasoning of Large Language ModelsLLM reasoning → Spatial and embodied reasoning
Agentic Literacy Debt: A Structural Problem the AI Literacy Field Has Not Yet NamedLLM agents and tool use → General LLM agent frameworks and benchmarks
Toward Reliable Design of LLM-Enabled Agentic Workflows: Optimizing Latency-Reliability-Cost TradeoffsLLM agents and tool use → Software engineering coding agents
TEAM-SimHRA: A Team-Based Simulation Framework for Human Reliability Analysis Using Multi-Agent Large Language ModelsLLM agents and tool use → Social simulation and multi-agent debate
How Much Thinking is Enough? Quantifying and Understanding Redundancy in LLM ReasoningLLM reasoning → Test-time compute and efficient thinking
Query-Aware Flow Diffusion for Graph-Based RAG with Retrieval GuaranteesLLM reasoning → Knowledge- and tool-augmented reasoning +1
More Thinking, More Bias: Length-Driven Position Bias in Reasoning ModelsLLM reasoning → Prompting and chain-of-thought reasoning
AgenticRecTune: Multi-Agent with Self-Evolving Skillhub for Recommendation System OptimizationLLM agents and tool use → Agentic recommender systems +1
Structure Guided Retrieval-Augmented Generation for Factual QueriesRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
AeSlides: Incentivizing Aesthetic Layout in LLM-Based Slide Generation via Verifiable RewardsLLM agents and tool use → Multimodal and embodied agents
Reinforcing privacy reasoning in LLMs via normative simulacra from fictionLLM reasoning → Reasoning for agents, planning and code +1
Frequency-Forcing: From Scaling-as-Time to Soft Frequency GuidanceDiffusion models → Flow matching, time series and science
Bootstrapping Post-training Signals for Open-ended Tasks via Rubric-based Self-play on Pre-training TextSynthetic data and self-improvement for LLMs → Self-play RL for reasoning
Information Aggregation with AI AgentsLLM agents and tool use → Financial trading and forecasting agents
PASTA: A Patch-Agnostic Twofold-Stealthy Backdoor Attack on Vision TransformersLLM safety and security → Backdoor and data poisoning attacks
TriEx: A Game-based Tri-View Framework for Explaining Internal Reasoning in Multi-Agent LLMsLLM reasoning → Reasoning for agents, planning and code +1
Normalizing Flows with Iterative DenoisingDiffusion models → Flow matching, time series and science
Separable Pathways for Causal Reasoning: How Architectural Scaffolding Enables Hypothesis-Space Restructuring in LLM AgentsLLM reasoning → Causal reasoning in LLMs +1
MoBayes: A Modular Bayesian Framework for Separating Reasoning from Language in Conversational Clinical Decision SupportLLM reasoning → Clinical and biomedical applications
Continuous Semantic Caching for Low-Cost LLM ServingEfficient LLM inference and serving → KV cache optimization
EmbodiedMidtrain: Bridging the Gap between Vision-Language Models and Vision-Language-Action Models via Mid-trainingMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
From Recall to Forgetting: Benchmarking Long-Term Memory for Personalized AgentsLLM agents and tool use → Agent memory systems
Semantic Prompting: Agentic Incremental Narrative Refinement through Spatial Semantic InteractionLLM agents and tool use → Multimodal and embodied agents
DistortBench: Benchmarking Vision Language Models on Image Distortion IdentificationMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Insights into Security-Related AI-Generated Pull RequestsLLM agents and tool use → Software engineering coding agents
Visual Reasoning through Tool-supervised Reinforcement LearningLLM reasoning → Multimodal reasoning +2
Infection-Reasoner: A Compact Vision-Language Model for Wound Infection Classification with Evidence-Grounded Clinical ReasoningLLM reasoning → Clinical and biomedical applications +1
Generalization and Membership Inference Attack a Practical PerspectiveLLM safety and security → Membership inference attacks
Behavioral Transfer in AI Agents: Evidence and Privacy ImplicationsLLM agents and tool use → Social simulation and multi-agent debate
SceneOrchestra: Efficient Agentic 3D Scene Synthesis via Full Tool-Call Trajectory GenerationLLM agents and tool use → Multimodal and embodied agents
MMCORE: MultiModal COnnection with Representation Aligned Latent EmbeddingsDiffusion models → Image generation, text-to-image and editing
A Reproducibility Study of Metacognitive Retrieval-Augmented GenerationRetrieval-augmented generation → RAG robustness and adaptive retrieval
A Data-Free Membership Inference Attack on Federated Learning in Hardware AssuranceLLM safety and security → Federated unlearning and poisoning
DR-Venus: Towards Frontier Edge-Scale Deep Research Agents with Only 10K Open DataLLM agents and tool use → Web, search, and research agents
Wan-Image: Pushing the Boundaries of Generative Visual IntelligenceDiffusion models → Image generation, text-to-image and editing
Rethinking Reinforcement Fine-Tuning in LVLM: Convergence, Reward Decomposition, and GeneralizationLLM reasoning → Multimodal reasoning +2
ChipCraftBrain: Validation-First RTL Generation via Multi-Agent OrchestrationLLM agents and tool use → Software engineering coding agents
If you're waiting for a sign... that might not be it! Mitigating Trust Boundary Confusion from Visual Injections on Vision-Language Agentic SystemsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Environmental Understanding Vision-Language Model for Embodied AgentLLM reasoning → Spatial and embodied reasoning +2
Expert Upcycling: Shifting the Compute-Efficient Frontier of Mixture-of-ExpertsMixture-of-experts → MoE large language models
AnyRecon: Arbitrary-View 3D Reconstruction with Video Diffusion ModelDiffusion models → 3D, point cloud and scene generation
UniT: Toward a Unified Physical Language for Human-to-Humanoid Policy Learning and World ModelingWorld models → Humanoid and legged locomotion world models
FASTER: Value-Guided Sampling for Fast RLLLM reasoning → Reinforcement learning for reasoning
VLA Foundry: A Unified Framework for Training Vision-Language-Action ModelsVision-language-action models → VLA architectures and generalization
Discovering a Shared Logical Subspace: Steering LLM Logical Reasoning via Alignment of Natural-Language and Symbolic ViewsLLM reasoning → Reasoning benchmarks and evaluation
SpanVLA: Efficient Action Bridging and Learning from Negative-Recovery Samples for Vision-Language-Action ModelLLM reasoning → Chain-of-thought reasoning for driving +2
Unveiling Fine-Grained Visual Traces: Evaluating Multimodal Interleaved Reasoning Chains in Multimodal STEM TasksLLM reasoning → Multimodal reasoning +1
A-MAR: Agent-based Multimodal Art Retrieval for Fine-Grained Artwork UnderstandingLLM agents and tool use → Multimodal and embodied agents
Mask World Model: Predicting What Matters for Robust Robot Policy LearningWorld models → Embodied and robot world models +1
IR-Flow: Bridging Discriminative and Generative Image Restoration via Rectified FlowDiffusion models → Inverse problems and image restoration
MMControl: Unified Multi-Modal Control for Joint Audio-Video GenerationDiffusion models → Audio, speech and music
MedFlowSeg: Flow Matching for Medical Image Segmentation with Frequency-Aware AttentionDiffusion models → Medical imaging and segmentation
ECLASS-Augmented Semantic Product Search for Electronic ComponentsLLM agents and tool use → Web, search, and research agents
Pause or Fabricate? Training Language Models for Grounded ReasoningLLM reasoning → Multimodal reasoning
FEPLB: Exploiting Copy Engines for Nearly Free MoE Load Balancing in Distributed TrainingMixture-of-experts → Efficient MoE training and inference systems
Safety-Critical Contextual Control via Online Riemannian Optimization with World ModelsWorld models → Latent dynamics models and model-based RL
SafetyALFRED: Evaluating Safety-Conscious Planning of Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
CoInteract: Physically-Consistent Human-Object Interaction Video Synthesis via Spatially-Structured Co-GenerationDiffusion models → Video generation +1
Time Series Augmented Generation for Financial ApplicationsLLM reasoning → Reasoning for agents, planning and code
A Self-Evolving Framework for Efficient Terminal Agents via Observational Context CompressionEfficient LLM inference and serving → Efficient reasoning and agent inference
RF-HiT: Rectified Flow Hierarchical Transformer for General Medical Image SegmentationDiffusion models → Medical imaging and segmentation
Multi-modal Reasoning with LLMs for Visual Semantic ArithmeticLLM reasoning → Multimodal reasoning
On Reasoning-Centric LLM-based Automated Theorem ProvingLLM reasoning → Mathematical reasoning
Taming Actor-Observer Asymmetry in Agents via Dialectical AlignmentLLM agents and tool use → General LLM agent frameworks and benchmarks
DT2IT-MRM: Debiased Preference Construction and Iterative Training for Multimodal Reward ModelingMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
Mesh Memory Protocol: Semantic Infrastructure for Multi-Agent LLM SystemsLLM agents and tool use → Agent memory systems
Integrating Anomaly Detection into Agentic AI for Proactive Risk Management in Human ActivityLLM agents and tool use → Agent threats, guardrails and governance
Cyber Defense Benchmark: Agentic Threat Hunting Evaluation for LLMs in SecOpsLLM agents and tool use → Agent threats, guardrails and governance
Assessing VLM-Driven Semantic-Affordance Inference for Non-Humanoid Robot MorphologiesMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
ReaLB: Real-Time Load Balancing for Multimodal MoE InferenceMixture-of-experts → Efficient MoE training and inference systems
Seeing Candidates at Scale: Multimodal LLMs for Visual Political Communication on InstagramMultimodal LLMs / vision-language models → Social bias and fairness in VLMs
CoDA: Towards Effective Cross-domain Knowledge Transfer via CoT-guided Domain AdaptationLLM reasoning → Reasoning benchmarks and evaluation
EVPO: Explained Variance Policy Optimization for Adaptive Critic Utilization in LLM Post-TrainingLLM reasoning → Reinforcement learning for reasoning
LePREC: Reasoning as Classification over Structured Factors for Assessing Relevance of Legal IssuesLLM reasoning → Reasoning for agents, planning and code
Involuntary In-Context Learning: Exploiting Few-Shot Pattern Completion to Bypass Safety Alignment in GPT-5.4LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
Do LLMs Game Formalization? Evaluating Faithfulness in Logical ReasoningLLM reasoning → Reasoning benchmarks and evaluation
Unsupervised Confidence Calibration for Reasoning LLMs from a Single GenerationLLM reasoning → Test-time compute and efficient thinking
What Makes an LLM a Good Optimizer? A Trajectory Analysis of LLM-Guided Evolutionary SearchLLM agents and tool use → General LLM agent frameworks and benchmarks
VCE: A zero-cost hallucination mitigation method of LVLMs via visual contrastive editingMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
HP-Edit: A Human-Preference Post-Training Framework for Image EditingDiffusion models → Image generation, text-to-image and editing +1
Lost in Translation: Do LVLM Judges Generalize Across Languages?Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
GRASPrune: Global Gating for Budgeted Structured Pruning of Large Language ModelsEfficient LLM inference and serving → Early exit and edge inference
Does Self-Consistency Improve the Recall of Encyclopedic Knowledge?LLM reasoning → Test-time compute and efficient thinking
Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture the Flag ChallengesLLM agents and tool use → Agent threats, guardrails and governance
DASH-KV: Accelerating Long-Context LLM Inference via Asymmetric KV Cache HashingEfficient LLM inference and serving → KV cache optimization
PLaMo 2.1-VL Technical ReportMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Large Language Models Exhibit Normative ConformityLLM agents and tool use → Social simulation and multi-agent debate
Rethinking Scale: Deployment Trade-offs of Small Language Models under Agent ParadigmsLLM reasoning → Reasoning for agents, planning and code +1
TEMPO: Scaling Test-time Training for Large Reasoning ModelsLLM reasoning → Test-time compute and efficient thinking
Explicit Trait Inference for Multi-Agent CoordinationLLM agents and tool use → Social simulation and multi-agent debate
HarDBench: A Benchmark for Draft-Based Co-Authoring Jailbreak Attacks for Safe Human-LLM Collaborative WritingLLM safety and security → Jailbreaking
Multimodal embodiment-aware navigation transformerDiffusion models → Diffusion and flow policies for robotics
DR-MMSearchAgent: Deepening Reasoning in Multimodal Search AgentsLLM agents and tool use → Multimodal and embodied agents

All 433