Papers

New on arXiv

Tue 28 Oct 2025

418 papers

SpatialTraceGen: High-Fidelity Traces for Efficient VLM Spatial Reasoning DistillationLLM reasoning → Spatial and embodied reasoning +2
Semi-Supervised Preference Optimization with Limited FeedbackLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
See the Speaker: Crafting High-Resolution Talking Faces from Speech with Prior Guidance and Region RefinementDiffusion models → Audio, speech and music
Non-myopic Matching and Rebalancing in Large-Scale On-Demand Ride-Pooling Systems Using Simulation-Informed Reinforcement LearningDeep reinforcement learning methods → RL for energy, networks and operations
The Kinetics of Reasoning: How Chain-of-Thought Shapes Learning in Transformers?LLM reasoning → Prompting and chain-of-thought reasoning
BlackboxNLP-2025 MIB Shared Task: Improving Circuit Faithfulness via Better Edge SelectionMechanistic interpretability → Circuits and mechanistic analysis
Automating Benchmark DesignLLM agents and tool use → General LLM agent frameworks and benchmarks
Secure Retrieval-Augmented Generation against Poisoning AttacksRetrieval-augmented generation → Privacy and security of RAG
StorageXTuner: An LLM Agent-Driven Automatic Tuning Framework for Heterogeneous Storage SystemsLLM agents and tool use → Agent memory systems
Aligning Large Language Models with Procedural Rules: An Autoregressive State-Tracking Prompting for In-Game TradingLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Emergence of Minimal Circuits for Indirect Object Identification in Attention-Only TransformersMechanistic interpretability → Circuits and mechanistic analysis
Emergent Coordinated Behaviors in Networked LLM Agents: Modeling the Strategic Dynamics of Information OperationsLLM agents and tool use → Social simulation and multi-agent debate
Enhancing Hierarchical Reinforcement Learning through Change Point Detection in Time SeriesDeep reinforcement learning methods → Deep RL algorithms and representations
LRT-Diffusion: Calibrated Risk-Aware Guidance for Diffusion PoliciesDiffusion models → Diffusion and flow policies for robotics +2
SCOUT: A Lightweight Framework for Scenario Coverage Assessment in Autonomous DrivingMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection +1
Finding Culture-Sensitive Neurons in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Can Aha Moments Be Fake? Towards Quantifying Decorative and True Thinking in Chain-of-ThoughtLLM reasoning → Prompting and chain-of-thought reasoning
SemCoT: Accelerating Chain-of-Thought Reasoning through Semantically-Aligned Implicit TokensLLM reasoning → Prompting and chain-of-thought reasoning
OrchVis: Hierarchical Multi-Agent Orchestration for Human OversightLLM agents and tool use → General LLM agent frameworks and benchmarks
RiddleBench: A New Generative Reasoning Benchmark for LLMsLLM reasoning → Reasoning benchmarks and evaluation
Seeing Through the MiRAGE: Evaluating Multimodal Retrieval Augmented GenerationRetrieval-augmented generation → Multimodal RAG
Deep Reinforcement Learning Approach to QoSAware Load Balancing in 5G Cellular Networks under User Mobility and Observation UncertaintyDeep reinforcement learning methods → RL for energy, networks and operations
Scheduling Your LLM Reinforcement Learning with Reasoning TreesLLM reasoning → Reinforcement learning for reasoning
The Generation Phases of Flow Matching: a Denoising PerspectiveDiffusion models → Flow matching, time series and science
Ming-Flash-Omni: A Sparse, Unified Architecture for Multimodal Perception and GenerationMixture-of-experts → Multimodal MoE models
SafeEditor: Unified MLLM for Efficient Post-hoc T2I Safety EditingLLM safety and security → Vision-language and text-to-image safety
Perception, Understanding and Reasoning, A Multimodal Benchmark for Video Fake News DetectionLLM reasoning → Multimodal reasoning +1
ProofSketch: Efficient Verified Reasoning for Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
CT-Less Attenuation Correction Using Multiview Ensemble Conditional Diffusion Model on High-Resolution Uncorrected PET ImagesDiffusion models → Medical imaging and segmentation
Conflict Adaptation in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
MASPRM: Multi-Agent System Process Reward ModelLLM reasoning → Reinforcement learning for reasoning +1
From Narrative to Action: A Hierarchical LLM-Agent Framework for Human Mobility GenerationLLM agents and tool use → Multimodal and embodied agents
Generative View StitchingDiffusion models → Video generation
Uniform Discrete Diffusion with Metric Path for Video GenerationDiffusion models → Video generation
Routing Matters in MoE: Scaling Diffusion Transformers with Explicit Routing GuidanceDiffusion models → Diffusion foundations and sampling +1
Agent Data Protocol: Unifying Datasets for Diverse, Effective Fine-tuning of LLM AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
Tongyi DeepResearch Technical ReportLLM agents and tool use → Web, search, and research agents
Greedy Sampling Is Provably Efficient for RLHFLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
AgentFold: Long-Horizon Web Agents with Proactive Context ManagementLLM agents and tool use → Agent memory systems
ParallelMuse: Agentic Parallel Thinking for Deep Information SeekingLLM agents and tool use → Web, search, and research agents
AgentFrontier: Expanding the Capability Frontier of LLM Agents with ZPD-Guided Data SynthesisLLM agents and tool use → General LLM agent frameworks and benchmarks +1
Repurposing Synthetic Data for Fine-grained Search Agent SupervisionLLM reasoning → Reinforcement learning for reasoning +2
STAR-Bench: Probing Deep Spatio-Temporal Reasoning as Audio 4D IntelligenceLLM reasoning → Audio-language model reasoning +1
InFeR: Informed Failure Resilience in Learned Visual Navigation ControlOffline RL and imitation learning from data → Language-conditioned policies and VLAs
Dissecting Role Cognition in Medical LLMs via Neuronal AblationLLM reasoning → Clinical and biomedical applications
Learning to Drive Safely with Hybrid OptionsDeep reinforcement learning methods → RL for autonomous driving and traffic
OrchDAG: Complex Tool Orchestration in Multi-Turn Interactions with Plan DAGsLLM agents and tool use → Tool learning and function calling
Evolving Interactive Diagnostic Agents in a Virtual Clinical EnvironmentLLM agents and tool use → Medical and scientific domain agents
Optimizing Retrieval for RAG via Reinforcement LearningRetrieval-augmented generation → Agentic RAG, agents and memory +1
Advancing site-specific disease and pest management in precision agriculture: From reasoning-driven foundation models to adaptive, feedback-based learningLLM reasoning → Clinical and biomedical applications
FunReason-MT Technical Report: Advanced Data Synthesis Solution for Real-world Multi-Turn Tool-useLLM reasoning → Reasoning for agents, planning and code +2
OpenReward: Learning to Reward Long-form Agentic Tasks via Reinforcement LearningLLM agents and tool use → Reinforcement learning for LLM agents +1
ReForm: Reflective Autoformalization with Prospective Bounded Sequence OptimizationLLM reasoning → Mathematical reasoning
ReplicationBench: Can AI Agents Replicate Astrophysics Research Papers?LLM agents and tool use → Web, search, and research agents
Towards Quadrupedal Jumping and Walking for Dynamic Locomotion using Reinforcement LearningDeep reinforcement learning methods → RL for robot locomotion and manipulation
OSWorld-MCP: Benchmarking MCP Tool Invocation In Computer-Use AgentsLLM reasoning → Reasoning for agents, planning and code +1
Dual-Mind World Models: A General Framework for Learning in Dynamic Wireless NetworksWorld models → World models for language agents +1
Latent Sketchpad: Sketching Visual Thoughts to Elicit Multimodal Reasoning in MLLMsLLM reasoning → Multimodal reasoning +1
Sample-efficient and Scalable Exploration in Continuous-Time RLDeep reinforcement learning methods → RL theory and sample complexity
Mitigating Hallucination in Large Language Models (LLMs): An Application-Oriented Survey on RAG, Reasoning, and Agentic SystemsLLM reasoning → Hallucination detection and mitigation +1
Adaptive Surrogate Gradients for Sequential Reinforcement Learning in Spiking Neural NetworksDeep reinforcement learning methods → Spiking and neuromorphic RL
Affordance Representation and Recognition for Autonomous AgentsWorld models → World models for language agents +1
Rethinking Visual Intelligence: Insights from Video PretrainingLLM reasoning → Multimodal reasoning
SPARTA: Evaluating Reasoning Segmentation Robustness through Black-Box Adversarial Paraphrasing in Text Autoencoder Latent SpaceLLM reasoning → Multimodal reasoning +1
Law in Silico: Simulating Legal Society with LLM-Based AgentsLLM agents and tool use → Social simulation and multi-agent debate
Can LLMs Write Faithfully? An Agent-Based Evaluation of LLM-generated Islamic ContentLLM agents and tool use → Web, search, and research agents
Human-Level Reasoning: A Comparative Study of Large Language Models on Logical and Abstract ReasoningLLM reasoning → Reasoning benchmarks and evaluation
Fill in the Blanks: Accelerating Q-Learning with a Handful of Demonstrations in Sparse Reward SettingsDeep reinforcement learning methods → Deep RL algorithms and representations
SynthWorlds: Controlled Parallel Worlds for Disentangling Reasoning and Knowledge in Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Comprehensive and Efficient Distillation for Lightweight Sentiment Analysis ModelsKnowledge distillation → Knowledge distillation for LLMs
OS-Sentinel: Towards Safety-Enhanced Mobile GUI Agents via Hybrid Validation in Realistic WorkflowsLLM agents and tool use → Agent threats, guardrails and governance
Metadata-Driven Retrieval-Augmented Generation for Financial Question AnsweringRetrieval-augmented generation → Legal and financial RAG
APTBench: Benchmarking Agentic Potential of Base LLMs During Pre-TrainingLLM agents and tool use → General LLM agent frameworks and benchmarks
Improving LLM Reasoning via Dependency-Aware Query Decomposition and Logic-Parallel Content ExpansionLLM reasoning → Knowledge- and tool-augmented reasoning
Adaptive Knowledge Transferring with Switching Dual-Student Framework for Semi-Supervised Medical Image SegmentationKnowledge distillation → Distillation for medical imaging
Automatically Benchmarking LLM Code Agents through Agent-Driven Annotation and EvaluationLLM agents and tool use → Software engineering coding agents
VDSAgents: A PCS-Guided Multi-Agent System for Veridical Data Science AutomationLLM agents and tool use → Web, search, and research agents
What do vision-language models see in the context? Investigating multimodal in-context learningMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Critique-RL: Training Language Models for Critiquing through Two-Stage Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning
Cybersecurity AI Benchmark (CAIBench): A Meta-Benchmark for Evaluating Cybersecurity AI AgentsLLM agents and tool use → Agent threats, guardrails and governance
Retrieval- and Argumentation-Enhanced Multi-Agent LLMs for Judgmental Forecasting (Extended Version with Supplementary Material)LLM agents and tool use → Social simulation and multi-agent debate
Lookahead Tree-Based Rollouts for Enhanced Trajectory-Level Exploration in Reinforcement Learning with Verifiable RewardsLLM reasoning → Reinforcement learning for reasoning
Verifying Large Language Models' Reasoning Paths via Correlation Matrix RankLLM reasoning → Reasoning benchmarks and evaluation
ViPER: Empowering the Self-Evolution of Visual Perception Abilities in Vision-Language ModelMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP ToolsLLM agents and tool use → Tool learning and function calling
SALS: Sparse Attention in Latent Space for KV cache CompressionEfficient LLM inference and serving → KV cache optimization
Survey and Tutorial of Reinforcement Learning Methods in Process Systems EngineeringDeep reinforcement learning methods → RL for energy, networks and operations
DeshadowMamba: Deshadowing as 1D Sequential SimilarityState-space sequence models (S4, Mamba) → Mamba for image restoration
Can LLMs Translate Human Instructions into a Reinforcement Learning Agent's Internal Emergent Symbolic Representation?LLM agents and tool use → Reinforcement learning for LLM agents
From Memorization to Reasoning in the Spectrum of Loss CurvatureLLM reasoning → Reasoning benchmarks and evaluation
Enabling Near-realtime Remote Sensing via Satellite-Ground Collaboration of Large Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Towards Transparent Reasoning: What Drives Faithfulness in Large Language Models?LLM reasoning → Clinical and biomedical applications
PaTaRM: Bridging Pairwise and Pointwise Signals via Preference-Aware Task-Adaptive Reward ModelingLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
SCOPE: Saliency-Coverage Oriented Token Pruning for Efficient Multimodel LLMsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Beyond Inference Intervention: Identity-Decoupled Diffusion for Face AnonymizationDiffusion models → Image generation, text-to-image and editing
Blindfolded Experts Generalize Better: Insights from Robotic Manipulation and VideogamesOffline RL and imitation learning from data → Imitation learning methods
MGA: Memory-Driven GUI Agent for Observation-Centric InteractionMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
BLM1: A Boundless Large Model for Cross-Space, Cross-Task, and Cross-Embodiment LearningLLM reasoning → Spatial and embodied reasoning +1
Enhancing Vision-Language Models for Autonomous Driving through Task-Specific Prompting and Spatial ReasoningLLM reasoning → Chain-of-thought reasoning for driving +2
BMGQ: A Bottom-up Method for Generating Complex Multi-hop Reasoning Questions from Semi-structured DataLLM reasoning → Knowledge- and tool-augmented reasoning

All 418