Daily Arxiv

전 세계에서 발간되는 인공지능 관련 논문을 정리하는 페이지 입니다.
본 페이지는 Google Gemini를 활용해 요약 정리하며, 비영리로 운영 됩니다.
논문에 대한 저작권은 저자 및 해당 기관에 있으며, 공유 시 출처만 명기하면 됩니다.

When Generative AI Meets Extended Reality: Enabling Scalable and Natural Interactions

An Explainable Market Integrity Monitoring System with Multi-Source Attention Signals and Transparent Scoring

Can We Trust LLM Detectors?

Embedding Retrofitting: Data Engineering for better RAG

Entropy-Tree: Tree-Based Decoding with Entropy-Guided Exploration

Elsewise: Authoring AI-Based Interactive Narrative with Possibility Space Visualization

Agentic Persona Control and Task State Tracking for Realistic User Simulation in Interactive Scenarios

LLM-based Multimodal Feedback Produces Equivalent Learning and Better Student Perceptions than Educator Feedback

Psychometric Comparability of LLM-Based Digital Twins

Scalable Board Expansion within a General Game System

Structured Hints for Sample-Efficient Lean Theorem Proving

Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning

LLM Prompt Evaluation for Educational Applications

Multimodal Climate Disinformation Detection: Integrating Vision-Language Models with External Knowledge Sources

Controlling Long-Horizon Behavior in Language Model Agents with Explicit State Dynamics

Designing faster mixed integer linear programming algorithm via learning the optimal path

AgriPINN: A Process-Informed Neural Network for Interpretable and Scalable Crop Biomass Prediction Under Water Stress

Grounding Large Language Models in Reaction Knowledge Graphs for Synthesis Retrieval

Deja Vu in Plots: Leveraging Cross-Session Evidence with Retrieval-Augmented LLMs for Live Streaming Risk Assessment

Decoupling Return-to-Go for Efficient Decision Transformer

Natural Language-Driven Global Mapping of Martian Landforms

ICON: Invariant Counterfactual Optimization with Neuro-Symbolic Priors for Text-Based Person Search

EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience

ErrorMap and ErrorAtlas: Charting the Failure Landscape of Large Language Models

Inference-Time Scaling of Verification: Self-Evolving Deep Research Agents via Test-Time Rubric-Guided Verification

VitalDiagnosis: AI-Driven Ecosystem for 24/7 Vital Monitoring and Chronic Disease Management

Creativity in the Age of AI: Rethinking the Role of Intentional Agency

Agentic Confidence Calibration

Off-Policy Actor-Critic with Sigmoid-Bounded Entropy for Real-World Robot Learning

Tabular Incremental Inference

PhysProver: Advancing Automatic Theorem Proving for Physics

Benchmarking Text-to-Python against Text-to-SQL: The Impact of Explicit Logic and Ambiguity

Investigation of the Generalisation Ability of Genetic Programming-evolved Scheduling Rules in Dynamic Flexible Job Shop Scheduling

AgentSM: Semantic Memory for Agentic Text-to-SQL

Improving Methodologies for LLM Evaluations Across Global Languages

Agentic Uncertainty Quantification

From Passive Metric to Active Signal: The Evolving Role of Uncertainty Quantification in Large Language Models

Improving Methodologies for Agentic Evaluations Across Domains: Leakage of Sensitive Information, Fraud and Cybersecurity Threats

Predictive Coding and Information Bottleneck for Hallucination Detection in Large Language Models

Agentic AI Governance and Lifecycle Management in Healthcare

CogToM: A Comprehensive Theory of Mind Benchmark inspired by Human Cognition for Large Language Models

Autonomous Business System via Neuro-symbolic AI

ALIGNAgent: Adaptive Learner Intelligence for Gap Identification and Next-step guidance

From Generative Engines to Actionable Simulators: The Imperative of Physical Grounding in World Models

TransportAgents: a multi-agents LLM framework for traffic accident severity prediction

The Dark Side of AI Transformers: Sentiment Polarization & the Loss of Business Neutrality by NLP Transformers

Tracking the Limits of Knowledge Propagation: How LLMs Fail at Multi-Step Reasoning with Conflicting Knowledge

MiRAGE: A Multiagent Framework for Generating Multimodal Multihop Question-Answer Dataset for RAG Evaluation

Reliability by design: quantifying and eliminating fabrication risk in LLMs. From generative to consultative AI: a comparative analysis in the legal domain and lessons for high-stakes knowledge bases

A tensor network formalism for neuro-symbolic AI

Not Your Typical Sycophant: The Elusive Nature of Sycophancy in Large Language Models

Beyond Prompting: Efficient and Robust Contextual Biasing for Speech LLMs via Logit-Space Integration (LOGIC)

GeMM-GAN: A Multimodal Generative Model Conditioned on Histopathology Images and Clinical Descriptions for Gene Expression Profile Generation

Logic Programming on Knowledge Graph Networks And its Application in Medical Domain

Prometheus Mind: Retrofitting Memory to Frozen Language Models

Replayable Financial Agents: A Determinism-Faithfulness Assurance Harness for Tool-Using LLM Agents

The Paradigm Shift: A Comprehensive Survey on Large Vision Language Models for Multimodal Fake News Detection

Aeon: High-Performance Neuro-Symbolic Memory Management for Long-Horizon LLM Agents

DeepSurvey-Bench: Evaluating Academic Value of Automatically Generated Scientific Survey

Uncovering Latent Bias in LLM-Based Emergency Department Triage Through Proxy Variables

Gated Sparse Attention: Combining Computational Efficiency with Training Stability for Long-Context Language Models

AI/ML in 3GPP 5G Advanced -- Services and Architecture

CookAnything: A Framework for Flexible and Consistent Multi-Step Recipe Image Generation

Dynamic Correction of Erroneous State Estimates via Diffusion Bayesian Exploration

Concept-Guided Backdoor Attack on Vision Language Models

On the Holographic Geometry of Deterministic Computation

Decoding inner speech with an end-to-end brain-to-text neural interface

Physics-informed Neural Operator Learning for Nonlinear Grad-Shafranov Equation

MindEval: Benchmarking Language Models on Multi-turn Mental Health Support

LAET: A Layer-wise Adaptive Ensemble Tuning Framework for Pretrained Language Models

FAST-CAD: A Fairness-Aware Framework for Non-Contact Stroke Diagnosis

Designing LLM-based Multi-Agent Systems for Software Engineering Tasks: Quality Attributes, Design Patterns and Rationale

SONIC: Supersizing Motion Tracking for Natural Humanoid Whole-Body Control

MOSS: Efficient and Accurate FP8 LLM Training with Microscaling and Automatic Scaling

Data-Augmented Deep Learning for Downhole Depth Sensing and Field Validation

Chinese Discharge Drug Recommendation in Metabolic Diseases with Large Language Models

Analysing Moral Bias in Finetuned LLMs through Mechanistic Interpretability

Orders in Chaos: Enhancing Large-Scale MoE LLM Serving with Data Movement Forecasting

Reinforce-Ada: An Adaptive Sampling Framework under Non-linear RL Objectives

TempoControl: Temporal Attention Guidance for Text-to-Video Models

AortaDiff: A Unified Multitask Diffusion Framework For Contrast-Free AAA Imaging

SoREX: Towards Self-Explainable Social Recommendation with Relevant Ego-Path Extraction

The AI Productivity Index (APEX)

Uncovering Grounding IDs: How External Cues Shape Multimodal Binding

Pushing Toward the Simplex Vertices: A Simple Remedy for Code Collapse in Smoothed Vector Quantization

V-CECE: Visual Counterfactual Explanations via Conceptual Edits

Momentum-constrained Hybrid Heuristic Trajectory Optimization Framework with Residual-enhanced DRL for Visually Impaired Scenarios

HARP: Hallucination Detection via Reasoning Subspace Projection

Rethinking Sparse Autoencoders: Select-and-Project for Fairness and Control from Encoder Features Alone

IPA: An Information-Reconstructive Input Projection Framework for Efficient Foundation Model Adaptation

Sparse but Wrong: Incorrect L0 Leads to Incorrect Features in Sparse Autoencoders

ORFuzz: Fuzzing the "Other Side" of LLM Safety -- Testing Over-Refusal

Retro-Expert: Collaborative Reasoning for Interpretable Retrosynthesis

A Survey on Diffusion Language Models

CodeNER: Code Prompting for Named Entity Recognition

ReSem3D: Refinable 3D Spatial Constraints via Fine-Grained Semantic Grounding for Generalizable Robotic Manipulation

VERIRAG: A Post-Retrieval Auditing of Scientific Study Summaries

SustainDiffusion: Optimising the Social and Environmental Sustainability of Stable Diffusion Models

LittleBit: Ultra Low-Bit Quantization via Latent Factorization

Real-Time Execution of Action Chunking Flow Policies

Multi-Agent Deep Research: Training Multi-Agent Systems with M-GRPO

Created by

Haebom

Category

Empty

저자

Haoyang Hong, Jiajun Yin, Yuan Wang, Jingnan Liu, Zhe Chen, Ailing Yu, Ji Li, Zhiling Ye, Hansong Xiao, Yefei Chen, Hualei Zhou, Yun Yue, Minghui Yang, Chunxiao Guo, Junwei Liu, Peng Wei, Jinjie Gu

개요

M-GRPO는 전문 분야 훈련 부족으로 정확도가 저하되는 문제점을 해결하기 위해, 각기 다른 LLM을 사용하는 다중 에이전트 시스템을 위한 hierarchical extension of Group Relative Policy Optimization(GRPO)를 제안합니다. M-GRPO는 메인 에이전트(플래너)와 여러 하위 에이전트(멀티턴 도구 실행자)를 포함하는 수직적 다중 에이전트 시스템을 위해 설계되었으며, 계층적 credit assignment를 유지하면서 메인 및 하위 에이전트 모두에 대해 group-relative advantages를 계산합니다. 또한, 가변적인 하위 에이전트 호출에도 불구하고 고정 크기 배치를 생성하는 trajectory-alignment scheme을 도입하고, 분리된 서버에서 에이전트를 실행하고 최소한의 통계를 공유하는 decoupled training pipeline을 배포하여 확장 가능한 훈련을 가능하게 합니다. M-GRPO는 실제 벤치마크에서 단일 에이전트 GRPO 및 frozen sub-agents를 사용하는 다중 에이전트 GRPO보다 일관되게 우수한 성능을 보이며, 이종 궤적 정렬 및 전문 에이전트 간의 최적화 분리가 도구-증강 추론 작업을 향상시킵니다.

시사점, 한계점

•

다중 에이전트 시스템에서 전문 분야별 훈련의 중요성을 강조하고, 개별 LLM 훈련의 필요성을 제기함

•

계층적 credit assignment를 통해 메인 및 하위 에이전트의 효율적인 학습을 가능하게 함

•

trajectory-alignment scheme을 통해 가변적인 하위 에이전트 호출 문제를 해결함

•

decoupled training pipeline을 통해 분산 환경에서 확장 가능한 학습을 가능하게 함

•

실제 벤치마크를 통해 M-GRPO의 우수성을 입증함

•

논문에서 한계점은 명시적으로 언급되지 않음

Made with Slashpage