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.

EndoIR: Degradation-Agnostic All-in-One Endoscopic Image Restoration via Noise-Aware Routing Diffusion

Adaptation and Fine-tuning with TabPFN for Travelling Salesman Problem

CGCE: Classifier-Guided Concept Erasure in Generative Models

EMOD: A Unified EEG Emotion Representation Framework Leveraging V-A Guided Contrastive Learning

Predicting the Future by Retrieving the Past

Quantifying Edits Decay in Fine-tuned LLMs

Retrieval Quality at Context Limit

EGG-SR: Embedding Symbolic Equivalence into Symbolic Regression via Equality Graph

Enhancing Diffusion Model Guidance through Calibration and Regularization

Understanding Cross Task Generalization in Handwriting-Based Alzheimer's Screening via Vision Language Adaptation

Hilbert-Guided Block-Sparse Local Attention

Policy Gradient-Based EMT-in-the-Loop Learning to Mitigate Sub-Synchronous Control Interactions

WAR-Re: Web API Recommendation with Semantic Reasoning

In-depth Analysis on Caching and Pre-fetching in Mixture of Experts Offloading

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

Measuring Model Performance in the Presence of an Intervention

Beyond the Lower Bound: Bridging Regret Minimization and Best Arm Identification in Lexicographic Bandits

When AI Meets the Web: Prompt Injection Risks in Third-Party AI Chatbot Plugins

VLAD-Grasp: Zero-shot Grasp Detection via Vision-Language Models

SymLight: Exploring Interpretable and Deployable Symbolic Policies for Traffic Signal Control

DRAGON: Guard LLM Unlearning in Context via Negative Detection and Reasoning

Sign language recognition from skeletal data using graph and recurrent neural networks

Lived Experience in Dialogue: Co-designing Personalization in Large Language Models to Support Youth Mental Well-being

Language Generation: Complexity Barriers and Implications for Learning

Beyond Redundancy: Diverse and Specialized Multi-Expert Sparse Autoencoder

Compressing Chemistry Reveals Functional Groups

OckBench: Measuring the Efficiency of LLM Reasoning

AdvisingWise: Supporting Academic Advising in Higher Educations Through a Human-in-the-Loop Multi-Agent Framework

Long Grounded Thoughts: Distilling Compositional Visual Reasoning Chains at Scale

TabDistill: Distilling Transformers into Neural Nets for Few-Shot Tabular Classification

Optimizing Diversity and Quality through Base-Aligned Model Collaboration

BrainCSD: A Hierarchical Consistency-Driven MoE Foundation Model for Unified Connectome Synthesis and Multitask Brain Trait Prediction

SSTODE: Ocean-Atmosphere Physics-Informed Neural ODEs for Sea Surface Temperature Prediction

Unveiling the Training Dynamics of ReLU Networks through a Linear Lens

Assessing the Reliability of Large Language Models in the Bengali Legal Context: A Comparative Evaluation Using LLM-as-Judge and Legal Experts

LLMs as Packagers of HPC Software

Report from Workshop on Dialogue alongside Artificial Intelligence

Grounding Foundational Vision Models with 3D Human Poses for Robust Action Recognition

Frequency Matters: When Time Series Foundation Models Fail Under Spectral Shift

Personalized Image Editing in Text-to-Image Diffusion Models via Collaborative Direct Preference Optimization

wa-hls4ml: A Benchmark and Surrogate Models for hls4ml Resource and Latency Estimation

An MLCommons Scientific Benchmarks Ontology

Who Evaluates AI's Social Impacts? Mapping Coverage and Gaps in First and Third Party Evaluations

AI-Enhanced High-Density NIRS Patch for Real-Time Brain Layer Oxygenation Monitoring in Neurological Emergencies

Conformal Prediction-Driven Adaptive Sampling for Digital Twins of Water Distribution Networks

Walking the Schr\"odinger Bridge: A Direct Trajectory for Text-to-3D Generation

Google-MedGemma Based Abnormality Detection in Musculoskeletal radiographs

FlowNet: Modeling Dynamic Spatio-Temporal Systems via Flow Propagation

CoPRIS: Efficient and Stable Reinforcement Learning via Concurrency-Controlled Partial Rollout with Importance Sampling

Fine-Tuning Vision-Language Models for Multimodal Polymer Property Prediction

Elements of Active Continuous Learning and Uncertainty Self-Awareness: a Narrow Implementation for Face and Facial Expression Recognition

Video Text Preservation with Synthetic Text-Rich Videos

C3-Diff: Super-resolving Spatial Transcriptomics via Cross-modal Cross-content Contrastive Diffusion Modelling

Automatic Extraction of Road Networks by using Teacher-Student Adaptive Structural Deep Belief Network and Its Application to Landslide Disaster

Efficient Online Continual Learning in Sensor-Based Human Activity Recognition

In-Context Adaptation of VLMs for Few-Shot Cell Detection in Optical Microscopy

Lookahead Unmasking Elicits Accurate Decoding in Diffusion Language Models

Effective Test-Time Scaling of Discrete Diffusion through Iterative Refinement

Sample-Efficient Language Modeling with Linear Attention and Lightweight Enhancements

Diversified Flow Matching with Translation Identifiability

EVLP:Learning Unified Embodied Vision-Language Planner with Reinforced Supervised Fine-Tuning

Deep one-gate per layer networks with skip connections are universal classifiers

AGRAG: Advanced Graph-based Retrieval-Augmented Generation for LLMs

Automated Invoice Data Extraction: Using LLM and OCR

ConnectomeBench: Can LLMs Proofread the Connectome?

Temporal Sparse Autoencoders: Leveraging the Sequential Nature of Language for Interpretability

Token Is All You Need: Cognitive Planning through Sparse Intent Alignment

Gravity-Awareness: Deep Learning Models and LLM Simulation of Human Awareness in Altered Gravity

Selective Diabetic Retinopathy Screening with Accuracy-Weighted Deep Ensembles and Entropy-Guided Abstention

The Evolution of Probabilistic Price Forecasting Techniques: A Review of the Day-Ahead, Intra-Day, and Balancing Markets

AIRMap - AI-Generated Radio Maps for Wireless Digital Twins

Retracing the Past: LLMs Emit Training Data When They Get Lost

Ming-UniAudio: Speech LLM for Joint Understanding, Generation and Editing with Unified Representation

From Failure Modes to Reliability Awareness in Generative and Agentic AI System

TEMPO: Temporal Multi-scale Autoregressive Generation of Protein Conformational Ensembles

Randomized-MLP Regularization Improves Domain Adaptation and Interpretability in DINOv2

Personalized Chain-of-Thought Summarization of Financial News for Investor Decision Support

Rewiring Human Brain Networks via Lightweight Dynamic Connectivity Framework: An EEG-Based Stress Validation

Production-Grade Local LLM Inference on Apple Silicon: A Comparative Study of MLX, MLC-LLM, Ollama, llama.cpp, and PyTorch MPS

Towards Ecologically Valid LLM Benchmarks: Understanding and Designing Domain-Centered Evaluations for Journalism Practitioners

Predicting Oscar-Nominated Screenplays with Sentence Embeddings

Weightless Neural Networks for Continuously Trainable Personalized Recommendation Systems

Biomedical Hypothesis Explainability with Graph-Based Context Retrieval

DOCUEVAL: An LLM-based AI Engineering Tool for Building Customisable Document Evaluation Workflows

IMDMR: An Intelligent Multi-Dimensional Memory Retrieval System for Enhanced Conversational AI

Customized Retrieval-Augmented Generation with LLM for Debiasing Recommendation Unlearning

AI Brown and AI Koditex: LLM-Generated Corpora Comparable to Traditional Corpora of English and Czech Texts

DigiData: Training and Evaluating General-Purpose Mobile Control Agents

DeepPersona: A Generative Engine for Scaling Deep Synthetic Personas

IterResearch: Rethinking Long-Horizon Agents via Markovian State Reconstruction

Beyond Detection: Exploring Evidence-based Multi-Agent Debate for Misinformation Intervention and Persuasion

AgenticSciML: Collaborative Multi-Agent Systems for Emergent Discovery in Scientific Machine Learning

PADiff: Predictive and Adaptive Diffusion Policies for Ad Hoc Teamwork

Evaluating Online Moderation Via LLM-Powered Counterfactual Simulations

Saliency Map-Guided Knowledge Discovery for Subclass Identification with LLM-Based Symbolic Approximations

Two Heads are Better than One: Distilling Large Language Model Features Into Small Models with Feature Decomposition and Mixture

MENTOR: A Metacognition-Driven Self-Evolution Framework for Uncovering and Mitigating Implicit Risks in LLMs on Domain Tasks

A Theoretical Analysis of Detecting Large Model-Generated Time Series

Boosting Fine-Grained Urban Flow Inference via Lightweight Architecture and Focalized Optimization

Agentic AI Sustainability Assessment for Supply Chain Document Insights

Linguistic Comparison of AI- and Human-Written Responses to Online Mental Health Queries

Created by

Haebom

저자

Koustuv Saha, Yoshee Jain, Munmun De Choudhury

개요

본 논문은 Reddit의 55개 온라인 정신 건강 커뮤니티(OMHC)에서 수집한 24,114개의 게시글과 138,758개의 답변을 사용하여 생성형 AI(GPT-4-Turbo, Llama-3, Mistral-7B)의 정신 건강 지원 능력을 평가한 연구이다. AI 응답과 인간의 응답을 심리언어학 및 어휘 의미론적 측정을 기반으로 비교 분석하여 AI 응답이 더 장황하고, 가독성이 높으며, 분석적으로 구조화되어 있지만, 인간의 상호작용에 내재된 언어적 다양성과 개인적 서술이 부족함을 밝혔다. 질적 분석을 통해 AI 응답의 중립성과 상호 확인 과정의 부재와 같은 특징을 확인하고 보완적인 통찰을 얻었다. 결론적으로, 생성형 AI를 OMHC에 통합하는 윤리적 및 실질적 함의를 논의하며, AI의 확장성과 신속성과 인간 연결의 진정성, 사회적 상호 작용성, 전문성을 균형 있게 고려하는 프레임워크를 제안한다.

시사점, 한계점

•

시사점:

◦

생성형 AI가 OMHC에 대한 즉각적이고 개인화된 응답을 제공할 수 있는 잠재력을 보여줌.

◦

AI 응답의 장점(가독성, 구조화)과 단점(언어적 다양성 부족, 개인적 서술 부족)을 명확히 제시.

◦

AI 통합 시 윤리적 및 실질적 함의에 대한 논의를 제공.

◦

인간 상호작용의 중요성을 강조하며, AI와 인간 지원의 상호 보완적 역할을 제안.

•

한계점:

◦

특정 플랫폼(Reddit)의 데이터에 기반하여 일반화에 한계가 있을 수 있음.

◦

AI 모델의 성능 평가에 사용된 측정 지표의 한계.

◦

AI 응답의 장기적 효과 및 심리적 영향에 대한 추가 연구 필요.

◦

AI와 인간 지원의 최적 통합 방식에 대한 구체적인 지침 부족.

Made with Slashpage