Papers

New on arXiv

Wed 17 Dec 2025

345 papers

SmartChunk Retrieval: Query-Aware Chunk Compression with Planning for Efficient Document RAGRetrieval-augmented generation → Multimodal RAG
Recursive Knowledge Synthesis for Multi-LLM Systems: Stability Analysis and Tri-Agent Audit FrameworkLLM agents and tool use → General LLM agent frameworks and benchmarks
Modeling the Mental World for Embodied AI: A Comprehensive ReviewWorld models → World models for navigation
SoliReward: Mitigating Susceptibility to Reward Hacking and Annotation Noise in Video Generation Reward ModelsLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
On-device Large Multi-modal Agent for Human Activity RecognitionLLM agents and tool use → Multimodal and embodied agents
EdgeFlex-Transformer: Transformer Inference for Edge DevicesMixture-of-experts → Vision mixture-of-experts
OpComm: A Reinforcement Learning Framework for Adaptive Buffer Control in Warehouse Volume ForecastingDeep reinforcement learning methods → RL for energy, networks and operations
Simulation-Driven Railway Delay Prediction: An Imitation Learning ApproachOffline RL and imitation learning from data → Imitation and offline RL for driving
The Deleuzian Representation HypothesisMechanistic interpretability → Sparse autoencoder training and evaluation
High-Performance Self-Supervised Learning by Joint Training of Flow MatchingDiffusion models → Flow matching, time series and science
Hard Negative Sample-Augmented DPO Post-Training for Small Language ModelsLLM reasoning → Reinforcement learning for reasoning +1
Tiny, On-Device Decision Makers with the MiniConv LibraryDeep reinforcement learning methods → RL for robot locomotion and manipulation
SuperFlow: Training Flow Matching Models with RL on the FlyDiffusion models → Diffusion and flow policies for robotics
CoVAR: Co-generation of Video and Action for Robotic Manipulation via Multi-Modal DiffusionDiffusion models → Diffusion and flow policies for robotics
Provably Extracting the Features from a General SuperpositionMechanistic interpretability → Sparse autoencoder training and evaluation
Are vision-language models ready to zero-shot replace supervised classification models in agriculture?Multimodal LLMs / vision-language models → Agricultural vision-language models
BRAID: Bounded Reasoning for Autonomous Inference and DecisionsLLM reasoning → Prompting and chain-of-thought reasoning
Seeing is Believing (and Predicting): Context-Aware Multi-Human Behavior Prediction with Vision Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
The Perceptual Observatory Characterizing Robustness and Grounding in MLLMsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Small Language Models for Efficient Agentic Tool Calling: Outperforming Large Models with Targeted Fine-tuningLLM agents and tool use → General LLM agent frameworks and benchmarks
R4: Retrieval-Augmented Reasoning for Vision-Language Models in 4D Spatio-Temporal SpaceLLM reasoning → Spatial and embodied reasoning +2
SALVE: Sparse Autoencoder-Latent Vector Editing for Mechanistic Control of Neural NetworksMechanistic interpretability → Sparse autoencoder training and evaluation
City Navigation in the Wild: Exploring Emergent Navigation from Web-Scale Knowledge in MLLMsLLM reasoning → Spatial and embodied reasoning +1
Scalable Agentic Reasoning for Designing Biologics Targeting Intrinsically Disordered ProteinsLLM agents and tool use → Medical and scientific domain agents
DSO: Direct Steering Optimization for Bias MitigationMultimodal LLMs / vision-language models → Social bias and fairness in VLMs
A Unification of Discrete, Gaussian, and Simplicial DiffusionDiffusion models → Diffusion foundations and sampling
Leveraging Spreading Activation for Improved Document Retrieval in Knowledge-Graph-Based RAG SystemsRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
VET Your Agent: Towards Host-Independent Autonomy via Verifiable Execution TracesLLM agents and tool use → Agent threats, guardrails and governance
Seeing Beyond Words: Self-Supervised Visual Learning for Multimodal Large Language ModelsLLM reasoning → Multimodal reasoning +1
Optimizing Agentic Language Model Inference via Speculative Tool CallsLLM agents and tool use → Software engineering coding agents
A Neurosymbolic Approach to Loop Invariant Generation via Weakest Precondition ReasoningLLM reasoning → Reasoning benchmarks and evaluation
CodeMem: Architecting Reproducible Agents via Dynamic MCP and Procedural MemoryLLM agents and tool use → Agent memory systems
DiffusionVL: Translating Any Autoregressive Models into Diffusion Vision Language ModelsMultimodal LLMs / vision-language models → Image editing and quality assessment +1
Dynamic Rebatching for Efficient Early-Exit Inference with DREXEfficient LLM inference and serving → Early exit and edge inference
End-to-End Training for Autoregressive Video Diffusion via Self-ResamplingDiffusion models → Video generation
VLIC: Vision-Language Models As Perceptual Judges for Human-Aligned Image CompressionMultimodal LLMs / vision-language models → Visual token pruning and efficiency +2
Skyra: AI-Generated Video Detection via Grounded Artifact ReasoningLLM reasoning → Multimodal reasoning
mimic-video: Video-Action Models for Generalizable Robot Control Beyond VLAsVision-language-action models → Human-video and latent-action pretraining
BashArena: A Control Setting for Highly Privileged AI AgentsLLM agents and tool use → Agent threats, guardrails and governance
Can LLMs Guide Their Own Exploration? Gradient-Guided Reinforcement Learning for LLM ReasoningLLM reasoning → Reinforcement learning for reasoning
Explaining the Reasoning of Large Language Models Using Attribution GraphsLLM reasoning → Reasoning benchmarks and evaluation
Stepwise Think-Critique: Interleaved Reasoning and Self-Critique in a Single LLMLLM reasoning → Test-time compute and efficient thinking
SoFlow: Solution Flow Models for One-Step Generative ModelingDiffusion models → Flow matching, time series and science
Characterizing Mamba's Selective Memory using Auto-EncodersState-space sequence models (S4, Mamba) → Core SSM architectures for language
VTCBench: Can Vision-Language Models Understand Long Context with Vision-Text Compression?Multimodal LLMs / vision-language models → Visual token pruning and efficiency
A Teacher-Student Perspective on the Dynamics of Learning Near the Optimal PointKnowledge distillation → General knowledge distillation methods
Corrective Diffusion Language ModelsDiffusion models → Discrete and text diffusion +1
IMKD: Intensity-Aware Multi-Level Knowledge Distillation for Camera-Radar FusionKnowledge distillation → Detection and dense prediction distillation
GRAN-TED: Generating Robust, Aligned, and Nuanced Text Embedding for Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
CTkvr: KV Cache Retrieval for Long-Context LLMs via Centroid then Token IndexingEfficient LLM inference and serving → KV cache optimization
DeX-Portrait: Disentangled and Expressive Portrait Animation via Explicit and Latent Motion RepresentationsDiffusion models → Audio, speech and music
Autonomous Pressure Control in MuVacAS via Deep Reinforcement Learning and Deep Learning Surrogate ModelsDeep reinforcement learning methods → RL for energy, networks and operations
Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode SupervisionLLM reasoning → Mathematical reasoning
How Do Semantically Equivalent Code Transformations Impact Membership Inference on LLMs for Code?LLM safety and security → Membership inference attacks
Copyright Infringement Risk Reduction via Chain-of-Thought and Task Instruction PromptingLLM reasoning → Multimodal reasoning
Double Horizon Model-Based Policy OptimizationDeep reinforcement learning methods → RL theory and sample complexity
Step-GUI Technical ReportLLM agents and tool use → GUI and mobile agents
FM-EAC: Feature Model-based Enhanced Actor-Critic for Multi-Task Control in Dynamic EnvironmentsDeep reinforcement learning methods → RL theory and sample complexity
MiVLA: Towards Generalizable Vision-Language-Action Model with Human-Robot Mutual Imitation Pre-trainingVision-language-action models → Human-video and latent-action pretraining
EUBRL: Epistemic Uncertainty Directed Bayesian Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
Bilateral Spatial Reasoning about Street Networks: Graph-based RAG with Qualitative Spatial RepresentationsLLM reasoning → Spatial and embodied reasoning
SCOPE: Prompt Evolution for Enhancing Agent EffectivenessLLM agents and tool use → General LLM agent frameworks and benchmarks
Image Complexity-Aware Adaptive Retrieval for Efficient Vision-Language ModelsMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Dual-Density Inference for Efficient Language Model ReasoningLLM reasoning → Test-time compute and efficient thinking
Adversarial versification in portuguese as a jailbreak operator in LLMsLLM safety and security → Jailbreaking
Expand and Prune: Maximizing Trajectory Diversity for Effective GRPO in Generative ModelsLLM reasoning → Reinforcement learning for reasoning
Exploring User Acceptance and Concerns toward LLM-powered Conversational Agents in Immersive Extended RealityLLM agents and tool use → Social simulation and multi-agent debate
Bits for Privacy: Evaluating Post-Training Quantization via Membership InferenceLLM safety and security → Membership inference attacks
A Masked Reverse Knowledge Distillation Method Incorporating Global and Local Information for Image Anomaly DetectionKnowledge distillation → Detection and dense prediction distillation
KD360-VoxelBEV: LiDAR and 360-degree Camera Cross Modality Knowledge Distillation for Bird's-Eye-View SegmentationKnowledge distillation → Detection and dense prediction distillation
SynthSeg-Agents: Multi-Agent Synthetic Data Generation for Zero-Shot Weakly Supervised Semantic SegmentationLLM agents and tool use → Multimodal and embodied agents +1
Graph Contextual Reinforcement Learning for Efficient Directed Controller SynthesisDeep reinforcement learning methods → RL for games and combinatorial optimization
Well Begun, Half Done: Reinforcement Learning with Prefix Optimization for LLM ReasoningLLM reasoning → Reinforcement learning for reasoning
Generative Preprocessing for Image Compression with Pre-trained Diffusion ModelsDiffusion models → Diffusion for compression and wireless
VLA-AN: An Efficient and Onboard Vision-Language-Action Framework for Aerial Navigation in Complex EnvironmentsVision-language-action models → Aerial and drone VLA
Assessing the Visual Enumeration Abilities of Specialized Counting Architectures and Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Intersectional Fairness in Vision-Language Models for Medical Image Disease ClassificationMultimodal LLMs / vision-language models → Medical vision-language models
CangLing-KnowFlow: A Unified Knowledge-and-Flow-fused Agent for Comprehensive Remote Sensing ApplicationsLLM agents and tool use → Multimodal and embodied agents
DEER: Draft with Diffusion, Verify with Autoregressive ModelsLLM reasoning → Test-time compute and efficient thinking +2
MCP-SafetyBench: A Benchmark for Safety Evaluation of Large Language Models with Real-World MCP ServersLLM agents and tool use → Agent threats, guardrails and governance
EagleVision: A Dual-Stage Framework with BEV-grounding-based Chain-of-Thought for Spatial IntelligenceLLM reasoning → Spatial and embodied reasoning +1
Explainable Action Form Assessment by Exploiting Multimodal Chain-of-Thoughts ReasoningLLM reasoning → Multimodal reasoning
Offline Multi-Task Multi-Objective Data-Driven Evolutionary Algorithm with Language Surrogate Model and Implicit Q-LearningDeep reinforcement learning methods → RL for games and combinatorial optimization
Beyond Majority Voting: Towards Fine-grained and More Reliable Reward Signal for Test-Time Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning +1
MCPZoo: A Large-Scale Dataset of Runnable Model Context Protocol Servers for AI AgentLLM agents and tool use → Agent threats, guardrails and governance
An Efficient Gradient-Based Inference Attack for Federated LearningLLM safety and security → Federated unlearning and poisoning
Automatic Reward Shaping from Multi-Objective Human HeuristicsDeep reinforcement learning methods → Deep RL algorithms and representations
QoS-Aware Hierarchical Reinforcement Learning for Joint Link Selection and Trajectory Optimization in SAGIN-Supported UAV Mobility ManagementDeep reinforcement learning methods → RL for robot navigation and UAV control
Beyond Fast and Slow: Cognitive-Inspired Elastic Reasoning for Large Language ModelsLLM reasoning → Test-time compute and efficient thinking
Quantifying Return on Security Controls in LLM SystemsLLM safety and security → Prompt injection and agent security
The Semantic Illusion: Certified Limits of Embedding-Based Hallucination Detection in RAG SystemsRetrieval-augmented generation → Hallucination detection in RAG
SGM: Safety Glasses for Multimodal Large Language Models via Neuron-Level DetoxificationMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
Agentic AI for ISAC: Analysis, Framework, and Case StudyLLM agents and tool use → Agentic AI for wireless networks
LADY: Linear Attention for Autonomous Driving Efficiency without TransformersEnd-to-end autonomous driving → VLMs for autonomous driving
Spectral Representation-based Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Beyond Accuracy: A Geometric Stability Analysis of Large Language Models in Chess EvaluationLLM reasoning → Reasoning for agents, planning and code
ISS Policy : Scalable Diffusion Policy with Implicit Scene SupervisionDiffusion models → Diffusion and flow policies for robotics +1
DreamPRM-Code: Function-as-Step Process Reward Model with Label Correction for LLM CodingLLM reasoning → Reinforcement learning for reasoning +1
Adaptive Partitioning and Learning for Stochastic Control of Diffusion ProcessesDeep reinforcement learning methods → Adversarial robustness and RL in finance
Imitation Game: Reproducing Deep Learning Bugs Leveraging an Intelligent AgentLLM agents and tool use → Software engineering coding agents

All 345