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.

Phare: A Safety Probe for Large Language Models

TCC-Bench: Benchmarking the Traditional Chinese Culture Understanding Capabilities of MLLMs

FALCON: False-Negative Aware Learning of Contrastive Negatives in Vision-Language Pretraining

Real-Time Verification of Embodied Reasoning for Generative Skill Acquisition

DexGarmentLab: Dexterous Garment Manipulation Environment with Generalizable Policy

REI-Bench: Can Embodied Agents Understand Vague Human Instructions in Task Planning?

Creating General User Models from Computer Use

Two Minds Better Than One: Collaborative Reward Modeling for LLM Alignment

Super-Resolution Generative Adversarial Networks based Video Enhancement

Superposition Yields Robust Neural Scaling

AdaptCLIP: Adapting CLIP for Universal Visual Anomaly Detection

On the Well-Posedness of Green's Function Reconstruction via the Kirchhoff-Helmholtz Equation for One-Speed Neutron Diffusion

CXMArena: Unified Dataset to benchmark performance in realistic CXM Scenarios

Ultrasound Report Generation with Multimodal Large Language Models for Standardized Texts

Accelerating Chain-of-Thought Reasoning: When Goal-Gradient Importance Meets Dynamic Skipping

STORYANCHORS: Generating Consistent Multi-Scene Story Frames for Long-Form Narratives

M3G: Multi-Granular Gesture Generator for Audio-Driven Full-Body Human Motion Synthesis

Concept-Level Explainability for Auditing & Steering LLM Responses

Generative Pre-trained Autoregressive Diffusion Transformer

FACET: Force-Adaptive Control via Impedance Reference Tracking for Legged Robots

Video-Enhanced Offline Reinforcement Learning: A Model-Based Approach

Learn to Think: Bootstrapping LLM Reasoning Capability Through Graph Representation Learning

Non-stationary Diffusion For Probabilistic Time Series Forecasting

ReGraP-LLaVA: Reasoning enabled Graph-based Personalized Large Language and Vision Assistant

RM-R1: Reward Modeling as Reasoning

Edge-Cloud Collaborative Computing on Distributed Intelligence and Model Optimization: A Survey

Synthesize-on-Graph: Knowledgeable Synthetic Data Generation for Continue Pre-training of Large Language Models

FreqKV: Frequency Domain Key-Value Compression for Efficient Context Window Extension

Vision Transformers in Precision Agriculture: A Comprehensive Survey

MF-LLM: Simulating Population Decision Dynamics via a Mean-Field Large Language Model Framework

What's Pulling the Strings? Evaluating Integrity and Attribution in AI Training and Inference through Concept Shift

UniversalRAG: Retrieval-Augmented Generation over Corpora of Diverse Modalities and Granularities

VCM: Vision Concept Modeling Based on Implicit Contrastive Learning with Vision-Language Instruction Fine-Tuning

GSFF-SLAM: 3D Semantic Gaussian Splatting SLAM via Feature Field

SPC: Evolving Self-Play Critic via Adversarial Games for LLM Reasoning

WASP: Benchmarking Web Agent Security Against Prompt Injection Attacks

OptimAI: Optimization from Natural Language Using LLM-Powered AI Agents

Process Reward Models That Think

MARFT: Multi-Agent Reinforcement Fine-Tuning

BrainPrompt: Multi-Level Brain Prompt Enhancement for Neurological Condition Identification

Dynamic Early Exit in Reasoning Models

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment

Leakage and Interpretability in Concept-Based Models

Evaluating Menu OCR and Translation: A Benchmark for Aligning Human and Automated Evaluations in Large Vision-Language Models

Probabilistic Stability Guarantees for Feature Attributions

CoT-RAG: Integrating Chain of Thought and Retrieval-Augmented Generation to Enhance Reasoning in Large Language Models

Low-hallucination Synthetic Captions for Large-Scale Vision-Language Model Pre-training

MCP Guardian: A Security-First Layer for Safeguarding MCP-Based AI System

Explicit and Implicit Representations in AI-based 3D Reconstruction for Radiology: A Systematic Review

OVERLORD: Ultimate Scaling of DataLoader for Multi-Source Large Foundation Model Training

A simulation-heuristics dual-process model for intuitive physics

Mimic In-Context Learning for Multimodal Tasks

ProtoECGNet: Case-Based Interpretable Deep Learning for Multi-Label ECG Classification with Contrastive Learning

Digital Twin Catalog: A Large-Scale Photorealistic 3D Object Digital Twin Dataset

Large Language Models Could Be Rote Learners

SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning

Prot42: a Novel Family of Protein Language Models for Target-aware Protein Binder Generation

FAST: Federated Active Learning with Foundation Models for Communication-efficient Sampling and Training

Beyond Conventional Transformers: The Medical X-ray Attention (MXA) Block for Improved Multi-Label Diagnosis Using Knowledge Distillation

Resource Allocation for RIS-Assisted CoMP-NOMA Networks using Reinforcement Learning

Effectively Controlling Reasoning Models through Thinking Intervention

ImF: Implicit Fingerprint for Large Language Models

ReaRAG: Knowledge-guided Reasoning Enhances Factuality of Large Reasoning Models with Iterative Retrieval Augmented Generation

Severing Spurious Correlations with Data Pruning

NaFM: Pre-training a Foundation Model for Small-Molecule Natural Products

CP-NCBF: A Conformal Prediction-based Approach to Synthesize Verified Neural Control Barrier Functions

Detecting LLM-Generated Peer Reviews

SocialJax: An Evaluation Suite for Multi-agent Reinforcement Learning in Sequential Social Dilemmas

Unifying Text Semantics and Graph Structures for Temporal Text-attributed Graphs with Large Language Models

Logic-in-Frames: Dynamic Keyframe Search via Visual Semantic-Logical Verification for Long Video Understanding

HICD: Hallucination-Inducing via Attention Dispersion for Contrastive Decoding to Mitigate Hallucinations in Large Language Models

UC-MOA: Utility-Conditioned Multi-Objective Alignment for Distributional Pareto-Optimality

dFLMoE: Decentralized Federated Learning via Mixture of Experts for Medical Data Analysis

Revisiting Backdoor Attacks on Time Series Classification in the Frequency Domain

Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models

Language-Enhanced Representation Learning for Single-Cell Transcriptomics

Semantic Shift Estimation via Dual-Projection and Classifier Reconstruction for Exemplar-Free Class-Incremental Learning

MoSE: Hierarchical Self-Distillation Enhances Early Layer Embeddings

Flexible Prefrontal Control over Hippocampal Episodic Memory for Goal-Directed Generalization

Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation

Bayesian Inverse Problems Meet Flow Matching: Efficient and Flexible Inference via Transformers

PINN-DT: Optimizing Energy Consumption in Smart Building Using Hybrid Physics-Informed Neural Networks and Digital Twin Framework with Blockchain Security

FANformer: Improving Large Language Models Through Effective Periodicity Modeling

DPZV: Elevating the Tradeoff between Privacy and Utility in Zeroth-Order Vertical Federated Learning

UniTok: A Unified Tokenizer for Visual Generation and Understanding

AutoBS: Autonomous Base Station Deployment with Reinforcement Learning and Digital Network Twins

Overcoming Dependent Censoring in the Evaluation of Survival Models

Can RLHF be More Efficient with Imperfect Reward Models? A Policy Coverage Perspective

AI-Powered Bayesian Inference

Yes, Q-learning Helps Offline In-Context RL

Culture-TRIP: Culturally-Aware Text-to-Image Generation with Iterative Prompt Refinement

The Hidden Strength of Disagreement: Unraveling the Consensus-Diversity Tradeoff in Adaptive Multi-Agent Systems

FairKV: Balancing Per-Head KV Cache for Fast Multi-GPU Inference

Multi-Faceted Multimodal Monosemanticity

Ranking Joint Policies in Dynamic Games using Evolutionary Dynamics

JL1-CD: A New Benchmark for Remote Sensing Change Detection and a Robust Multi-Teacher Knowledge Distillation Framework

KL Penalty Control via Perturbation for Direct Preference Optimization

Pre-training Auto-regressive Robotic Models with 4D Representations

To Think or Not to Think: Exploring the Unthinking Vulnerability in Large Reasoning Models

Collaborative Deterministic-Probabilistic Forecasting for Real-World Spatiotemporal Systems

MetaDE: Evolving Differential Evolution by Differential Evolution

Created by

Haebom

저자

Minyang Chen, Chenchen Feng, and Ran Cheng

개요

본 논문은 차분 진화(DE) 알고리즘의 하이퍼파라미터 최적화 문제를 해결하기 위해 메타 차원에서 DE 자체를 이용하여 하이퍼파라미터와 전략을 진화시키는 MetaDE 방법론을 제안한다. MetaDE는 특수한 파라미터화 기법을 통해 DE의 파라미터와 전략을 동적으로 수정하며, GPU 가속 컴퓨팅 프레임워크를 활용하여 계산 효율을 높인다. CEC2022 벤치마크 및 로봇 제어(진화적 강화 학습) 실험을 통해 MetaDE의 성능을 검증하고, 소스 코드를 공개하였다.

시사점, 한계점

•

시사점:

◦

DE 알고리즘의 하이퍼파라미터 최적화 문제에 대한 새로운 접근 방식 제시

◦

메타 차원의 DE를 이용한 자동화된 하이퍼파라미터 최적화 및 문제 해결 과정 구현

◦

GPU 가속을 통한 계산 효율 향상

◦

CEC2022 벤치마크 및 로봇 제어 실험에서 우수한 성능 검증

◦

소스 코드 공개를 통한 재현성 및 확장성 확보

•

한계점:

◦

제안된 MetaDE의 일반적인 성능과 한계에 대한 더 폭넓은 실험적 분석 필요

◦

특정 문제 유형에 대한 최적화된 파라미터 설정 및 전략 발견에 대한 추가 연구 필요

◦

다른 최적화 알고리즘과의 비교 분석 필요

◦

GPU 가속에 대한 의존성으로 인한 접근성 제약 가능성

Made with Slashpage