Papers

New on arXiv

Sat 21 Mar 2026

181 papers

Beyond Static Visual Tokens: Structured Sequential Visual Chain-of-Thought ReasoningLLM reasoning → Multimodal reasoning +1
Distilled Large Language Model-Driven Dynamic Sparse Expert Activation MechanismMixture-of-experts → Vision mixture-of-experts
T-MAP: Red-Teaming LLM Agents with Trajectory-aware Evolutionary SearchLLM agents and tool use → Agent threats, guardrails and governance +1
Graphs RAG at Scale: Beyond Retrieval-Augmented Generation With Labeled Property Graphs and Resource Description Framework for Complex and Unknown Search SpacesRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Causal Direct Preference Optimization for Distributionally Robust Generative RecommendationLLM alignment / RLHF / preference optimisation → Preference alignment for recommenders and RAG
Detection of adversarial intent in Human-AI teams using LLMsLLM safety and security → Prompt injection and agent security
DiscoUQ: Structured Disagreement Analysis for Uncertainty Quantification in LLM Agent EnsemblesLLM agents and tool use → Social simulation and multi-agent debate
Understanding Contextual Recall in Transformers: How Finetuning Enables In-Context Reasoning over Pretraining KnowledgeLLM reasoning → Reasoning benchmarks and evaluation
Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language ModelsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks
User Preference Modeling for Conversational LLM Agents: Weak Rewards from Retrieval-Augmented InteractionRetrieval-augmented generation → Conversational and speech RAG +2
AC4A: Access Control for AgentsLLM agents and tool use → Agent threats, guardrails and governance
Causally-Guided Diffusion for Stable Feature SelectionDiffusion models → Diffusion foundations and sampling
Profit is the Red Team: Stress-Testing Agents in Strategic Economic InteractionsLLM agents and tool use → Agent threats, guardrails and governance +1
Do LLM-Driven Agents Exhibit Engagement Mechanisms? Controlled Tests of Information Load, Descriptive Norms, and Popularity CuesLLM agents and tool use → Social simulation and multi-agent debate
Mitigating Shortcut Reasoning in Language Models: A Gradient-Aware Training ApproachLLM reasoning → Test-time compute and efficient thinking
LLM Router: Rethinking Routing with Prefill ActivationsLLM reasoning → Test-time compute and efficient thinking
MemoNoveltyAgent: A Historical Research Memory-Aware Agent Workflow for Paper Novelty AssessmentLLM agents and tool use → Web, search, and research agents
TAFG-MAN: Timestep-Adaptive Frequency-Gated Latent Diffusion for Efficient and High-Quality Low-Dose CT Image DenoisingDiffusion models → Medical imaging and segmentation
EruDiff: Refactoring Knowledge in Diffusion Models for Advanced Text-to-Image SynthesisDiffusion models → Image generation, text-to-image and editing
Predictive Regularization Against Visual Representation Degradation in Multimodal Large Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
RLVR Training of LLMs Does Not Improve Thinking Ability for General QA: Evaluation Method and a Simple SolutionLLM reasoning → Reinforcement learning for reasoning
ME-IQA: Memory-Enhanced Image Quality Assessment via Re-RankingLLM reasoning → Multimodal reasoning +1
Memory-Efficient Fine-Tuning Diffusion Transformers via Dynamic Patch Sampling and Block SkippingDiffusion models → Diffusion foundations and sampling
Subjective-Graph LLM Agents for Simulating Uncertainty in Classroom Social PerceptionLLM agents and tool use → Social simulation and multi-agent debate
CTCal: Rethinking Text-to-Image Diffusion Models via Cross-Timestep Self-CalibrationDiffusion models → Image generation, text-to-image and editing
Mamba Learns in Context: Structure-Aware Domain Generalization for Multi-Task Point Cloud UnderstandingState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
Reasoning Topology Matters: Network-of-Thought for Complex Reasoning TasksLLM reasoning → Prompting and chain-of-thought reasoning
Premier: Personalized Preference Modulation with Learnable User Embedding in Text-to-Image GenerationLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
mmWave-Diffusion:A Novel Framework for Respiration Sensing Using Observation-Anchored Conditional Diffusion ModelDiffusion models → 3D, point cloud and scene generation
Clinical Cognition Alignment for Gastrointestinal Diagnosis with Multimodal LLMsMultimodal LLMs / vision-language models → Medical vision-language models
SWE-Next: Scalable Real-World Software Engineering Tasks for AgentsLLM agents and tool use → Software engineering coding agents
AI-Driven Multi-Agent Simulation of Stratified Polyamory Systems: A Computational Framework for Optimizing Social Reproductive EfficiencyLLM agents and tool use → Social simulation and multi-agent debate
PAVE: Premise-Aware Validation and Editing for Retrieval-Augmented LLMsRetrieval-augmented generation → RAG robustness and adaptive retrieval
Towards Intelligent Geospatial Data Discovery: a knowledge graph-driven multi-agent framework powered by large language modelsLLM agents and tool use → Multimodal and embodied agents
ROI-Driven Foveated Attention for Unified Egocentric Representations in Vision-Language-Action SystemsVision-language-action models → Human-video and latent-action pretraining
Attention in Space: Functional Roles of VLM Heads for Spatial ReasoningLLM reasoning → Spatial and embodied reasoning +1
StageCraft: Execution Aware Mitigation of Distractor and Obstruction Failures in VLA ModelsVision-language-action models → Generalist VLA models and benchmarks
Speedup Patch: Learning a Plug-and-Play Policy to Accelerate Embodied ManipulationWorld models → Embodied and robot world models
Diffusion Model for Manifold Data: Score Decomposition, Curvature, and Statistical ComplexityDiffusion models → Diffusion foundations and sampling
Hear Both Sides: Efficient Multi-Agent Debate via Diversity-Aware Message RetentionLLM reasoning → Reasoning for agents, planning and code +1
AEGIS: From Clues to Verdicts -- Graph-Guided Deep Vulnerability Reasoning via Dialectics and Meta-AuditingLLM reasoning → Reasoning for agents, planning and code
ACRFence: Preventing Semantic Rollback Attacks in Agent Checkpoint-RestoreLLM agents and tool use → Agent threats, guardrails and governance
Reasoning Traces Shape Outputs but Models Won't Say SoLLM reasoning → Prompting and chain-of-thought reasoning
Beyond Token Eviction: Mixed-Dimension Budget Allocation for Efficient KV Cache CompressionEfficient LLM inference and serving → KV cache optimization
Towards Practical World Model-based Reinforcement Learning for Vision-Language-Action ModelsWorld models → Embodied and robot world models +1
Bayesian Learning in Episodic Zero-Sum GamesLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
MKA: Memory-Keyed Attention for Efficient Long-Context ReasoningLLM reasoning → Test-time compute and efficient thinking
Improving Diffusion Generalization with Weak-to-Strong Segmented GuidanceDiffusion models → Diffusion foundations and sampling
Generative Diffusion Model for Risk-Neutral Derivative PricingDiffusion models → Flow matching, time series and science
Context Cartography: Toward Structured Governance of Contextual Space in Large Language Model SystemsLLM reasoning → Spatial and embodied reasoning
GraphDiffMed: Knowledge-Constrained Differential Attention with Pharmacological Graph Priors for Medication Recommendation
Exploring Data Augmentation and Resampling Strategies for Transformer-Based Models to Address Class Imbalance in AI Scoring of Scientific Explanations in NGSS Classroom
Clinical Note Bloat Reduction for Efficient LLM Use
IWLV-Ramayana: A Sarga-Aligned Parallel Corpus of Valmiki's Ramayana Across Indian Languages
Can Large Language Models Reliably Extract Physiology Index Values from Coronary Angiography Reports?
Alignment midtraining for animals
TRACE: Traceroute-based Internet Route change Analysis with Ensemble Learning
Experimental Design for Missing Physics
Interpretable Battery Aging without Extra Tests via Neural-Assisted Physics-based Modelling
Learning to Remember: Attentive Reinforcement Learning for Edge Serverless Autoscaling
Ordinal Semantic Segmentation Applied to Medical and Odontological Images
Implicit Turn-Wise Policy Optimization for Proactive User-LLM Interaction
Errors in AI-Assisted Retrieval of Medical Literature: A Comparative Study
Cloud-Edge Collaborative Large Models for Robust Photovoltaic Power Forecasting
Neutrino Oscillation Parameter Estimation Using Structured Hierarchical Transformers
Problems with Chinchilla Approach 2: Systematic Biases in IsoFLOP Parabola Fits
Cyber Deception for Mission Surveillance via Hypergame-Theoretic Deep Reinforcement Learning
From Causal Discovery to Dynamic Causal Inference in Neural Time Series
GraPHFormer: A Multimodal Graph Persistent Homology Transformer for the Analysis of Neuroscience Morphologies
Hard labels sampled from sparse targets mislead rotation invariant algorithms
Learning to Aggregate Zero-Shot LLM Agents for Corporate Disclosure Classification
Integrative Learning of Dynamically Evolving Multiplex Graphs and Nodal Attributes Using Neural Network Gaussian Processes with an Application to Dynamic Terrorism Graphs
Beyond Expression Similarity: Contrastive Learning Recovers Functional Gene Associations from Protein Interaction Structure
Before the Tool Call: Deterministic Pre-Action Authorization for Autonomous AI Agents
gUFO: A Gentle Foundational Ontology for Semantic Web Knowledge Graphs
Two Approaches to Direct Estimation of Riesz Representers
MOELIGA: a multi-objective evolutionary approach for feature selection with local improvement
Stability of Sequential and Parallel Coordinate Ascent Variational Inference
Active Inference for Physical AI Agents -- An Engineering Perspective
Deep Adaptive Rate Allocation in Volatile Heterogeneous Wireless Networks
Discriminative Representation Learning for Clinical Prediction
Democratizing AI: A Comparative Study in Deep Learning Efficiency and Future Trends in Computational Processing
Enhancing LIME using Neural Decision Trees
LLM-ODE: Data-driven Discovery of Dynamical Systems with Large Language Models
Bayesian Scattering: A Principled Baseline for Uncertainty on Image Data
The Hidden Puppet Master: Predicting Human Belief Change in Manipulative LLM Dialogues
Data-Driven Weak-form Discovery of Stochastic Systems
Unfolding with a Wasserstein Loss
Natural Gradient Descent for Online Continual Learning
The data heat island effect: quantifying the impact of AI data centers in a warming world
Spectral-Sphere-Constrained Hyper-Connections
Auto-differentiable data assimilation: Co-learning of states, dynamics, and filtering algorithms
Scene Graph-guided SegCaptioning Transformer with Fine-grained Alignment for Controllable Video Segmentation and Captioning
Characterizing the onset and offset of motor imagery during passive arm movements induced by an upper-body exoskeleton
RubricRAG: Towards Interpretable and Reliable LLM Evaluation via Domain Knowledge Retrieval for Rubric Generation
Incentive-Aware Federated Averaging with Performance Guarantees under Strategic Participation
ReLaMix: Residual Latency-Aware Mixing for Delay-Robust Financial Time-Series Forecasting
Semantic Sections: An Atlas-Native Feature Ontology for Obstructed Representation Spaces
Discrete Coefficients and Open Invariant Covers in the Homology of Ample Groupoids
Restoring Neural Network Plasticity for Faster Transfer Learning

All 181