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.

ELV-Halluc: Benchmarking Semantic Aggregation Hallucinations in Long Video Understanding

Quantifying Label-Induced Bias in Large Language Model Self- and Cross-Evaluations

R-4B: Incentivizing General-Purpose Auto-Thinking Capability in MLLMs via Bi-Mode Annealing and Reinforce Learning

EmbodiedOneVision: Interleaved Vision-Text-Action Pretraining for General Robot Control

Language Models and Logic Programs for Trustworthy Financial Reasoning

Occlusion Robustness of CLIP for Military Vehicle Classification

SPGrasp: Spatiotemporal Prompt-driven Grasp Synthesis in Dynamic Scenes

DrugReasoner: Interpretable Drug Approval Prediction with a Reasoning-augmented Language Model

Dynamic Fusion Multimodal Network for SpeechWellness Detection

Agentic AI for Software: thoughts from Software Engineering community

CoViPAL: Layer-wise Contextualized Visual Token Pruning for Large Vision-Language Models

LLM Assertiveness can be Mechanistically Decomposed into Emotional and Logical Components

ONG: Orthogonal Natural Gradient Descent

Tri-Accel: Curvature-Aware Precision-Adaptive and Memory-Elastic Optimization for Efficient GPU Usage

GPT-OSS-20B: A Comprehensive Deployment-Centric Analysis of OpenAI's Open-Weight Mixture of Experts Model

Bridging Generalization and Personalization in Wearable Human Activity Recognition via On-Device Few-Shot Learning

SparK: Query-Aware Unstructured Sparsity with Recoverable KV Cache Channel Pruning

Adaptively Robust LLM Inference Optimization under Prediction Uncertainty

NVIDIA Nemotron Nano 2: An Accurate and Efficient Hybrid Mamba-Transformer Reasoning Model

FLAIR: Frequency- and Locality-Aware Implicit Neural Representations

Hierarchical Evaluation Function: A Multi-Metric Approach for Optimizing Demand Forecasting Models

Learning local and global prototypes with optimal transport for unsupervised anomaly detection and localization

Quantum Flow Matching

BConformeR: A Conformer Based on Mutual Sampling for Unified Prediction of Continuous and Discontinuous Antibody Binding Sites

Preacher: Paper-to-Video Agentic System

UQGNN: Uncertainty Quantification of Graph Neural Networks for Multivariate Spatiotemporal Prediction

Grid2Guide: A* Enabled Small Language Model for Indoor Navigation

ACD-CLIP: Decoupling Representation and Dynamic Fusion for Zero-Shot Anomaly Detection

MAQuA: Adaptive Question-Asking for Multidimensional Mental Health Screening using Item Response Theory

Class Unbiasing for Generalization in Medical Diagnosis

LLM Serving Optimization with Variable Prefill and Decode Lengths

Grid-Agent: An LLM-Powered Multi-Agent System for Power Grid Control

CF3: Compact and Fast 3D Feature Fields

MSC: A Marine Wildlife Video Dataset with Grounded Segmentation and Clip-Level Captioning

A DbC Inspired Neurosymbolic Layer for Trustworthy Agent Design

Convergence Analysis of Aggregation-Broadcast in LoRA-enabled Distributed Fine-Tuning

Exploring the Application of Visual Question Answering (VQA) for Classroom Activity Monitoring

AR-LIF: Adaptive reset leaky integrate-and-fire neuron for spiking neural networks

A Markov Categorical Framework for Language Modeling

Towards Compute-Optimal Many-Shot In-Context Learning

Benchmarking LLM Privacy Recognition for Social Robot Decision Making

Diffusion Models for Time Series Forecasting: A Survey

GPI-Net: Gestalt-Guided Parallel Interaction Network via Orthogonal Geometric Consistency for Robust Point Cloud Registration

ExCyTIn-Bench: Evaluating LLM agents on Cyber Threat Investigation

Bottom-up Domain-specific Superintelligence: A Reliable Knowledge Graph is What We Need

Demographic-aware fine-grained classification of pediatric wrist fractures

Agentic-R1: Distilled Dual-Strategy Reasoning

Driving as a Diagnostic Tool: Scenario-based Cognitive Assessment in Older Drivers from Driving Video

MedVAL: Toward Expert-Level Medical Text Validation with Language Models

NOCTIS: Novel Object Cyclic Threshold based Instance Segmentation

RALLY: Role-Adaptive LLM-Driven Yoked Navigation for Agentic UAV Swarms

Iterative Distillation for Reward-Guided Fine-Tuning of Diffusion Models in Biomolecular Design

Towards Efficient and Accurate Spiking Neural Networks via Adaptive Bit Allocation

Flow-Modulated Scoring for Semantic-Aware Knowledge Graph Completion

TPTT: Transforming Pretrained Transformers into Titans

What Is the Point of Equality in Machine Learning Fairness? Beyond Equality of Opportunity

QGuard:Question-based Zero-shot Guard for Multi-modal LLM Safety

A theoretical framework for self-supervised contrastive learning for continuous dependent data

LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician Validation

Auto prompt sql: a resource-efficient architecture for text-to-sql translation in constrained environments

Labelling Data with Unknown References

FinS-Pilot: A Benchmark for Online Financial RAG System

Diagnosing Reliability in Text-Guided Medical Image Editing

Speeding Up Hyper-Heuristics With Markov-Chain Operator Selection and the Only-Worsening Acceptance Operator

A versatile foundation model for cine cardiac magnetic resonance image analysis tasks

Should I Share this Translation? Evaluating Quality Feedback for User Reliance on Machine Translation

Multiple LLM Agents Debate for Equitable Cultural Alignment

Automated Essay Scoring Incorporating Annotations from Automated Feedback Systems

Can NeRFs See without Cameras?

DiffDecompose: Layer-Wise Decomposition of Alpha-Composited Images via Diffusion Transformers

The challenge of hidden gifts in multi-agent reinforcement learning

How Can I Publish My LLM Benchmark Without Giving the True Answers Away?

Cog-TiPRO: Iterative Prompt Refinement with LLMs to Detect Cognitive Decline via Longitudinal Voice Assistant Commands

From Tokens to Thoughts: How LLMs and Humans Trade Compression for Meaning

Unveil Multi-Picture Descriptions for Multilingual Mild Cognitive Impairment Detection via Contrastive Learning

Toward Real-World Cooperative and Competitive Soccer with Quadrupedal Robot Teams

FreqSelect: Frequency-Aware fMRI-to-Image Reconstruction

ViEEG: Hierarchical Visual Neural Representation for EEG Brain Decoding

One Shot Dominance: Knowledge Poisoning Attack on Retrieval-Augmented Generation Systems

Search and Refine During Think: Autonomous Retrieval-Augmented Reasoning of LLMs

ManipBench: Benchmarking Vision-Language Models for Low-Level Robot Manipulation

Mask-PINNs: Mitigating Internal Covariate Shift in Physics-Informed Neural Networks

ORBIT-2: Scaling Exascale Vision Foundation Models for Weather and Climate Downscaling

FairPO: Robust Preference Optimization for Fair Multi-Label Learning

Automated Parsing of Engineering Drawings for Structured Information Extraction Using a Fine-tuned Document Understanding Transformer

GenTorrent: Scaling Large Language Model Serving with An Overlay Network

Tilus: A Tile-Level GPGPU Programming Language for Low-Precision Computation

Progent: Programmable Privilege Control for LLM Agents

A Rollout-Based Algorithm and Reward Function for Resource Allocation in Business Processes

Agent-Q: Fine-Tuning Large Language Models for Quantum Circuit Generation and Optimization

A Hybrid Fully Convolutional CNN-Transformer Model for Inherently Interpretable Disease Detection from Retinal Fundus Images

More Bang for the Buck: Process Reward Modeling with Entropy-Driven Uncertainty

LATTE-MV: Learning to Anticipate Table Tennis Hits from Monocular Videos

Flip Learning: Weakly Supervised Erase to Segment Nodules in Breast Ultrasound

Optimizing Breast Cancer Detection in Mammograms: A Comprehensive Study of Transfer Learning, Resolution Reduction, and Multi-View Classification

Unmasking Deceptive Visuals: Benchmarking Multimodal Large Language Models on Misleading Chart Question Answering

General Table Question Answering via Answer-Formula Joint Generation

Open-World Skill Discovery from Unsegmented Demonstrations

To See a World in a Spark of Neuron: Disentangling Multi-task Interference for Training-free Model Merging

MOHPER: Multi-objective Hyperparameter Optimization Framework for E-commerce Retrieval System

Diagnosing Reliability in Text-Guided Medical Image Editing

Created by

Haebom

저자

Minghao Liu, Zhitao He, Zhiyuan Fan, Qingyun Wang, Yi R. Fung

개요

MedEBench는 의료 영상에 대한 텍스트 기반 이미지 편집을 평가하기 위한 종합적인 벤치마크입니다. 13개의 해부학적 영역에 걸쳐 70개의 과제를 포함하는 1,182개의 임상적으로 얻은 이미지-프롬프트 삼중항으로 구성됩니다. 편집 정확도, 맥락 보존 및 시각적 품질을 다루는 임상적으로 관련성 있는 평가 프레임워크, 7개의 최첨단 모델에 대한 체계적인 비교, 주의 지점과 ROI 간의 IoU를 사용한 실패 분석 프로토콜을 제공합니다. 수술 결과 시뮬레이션, 개인화된 교육 자료 생성, 환자 의사소통 향상 등 임상적 활용 가능성이 높습니다.

시사점, 한계점

•

시사점:

◦

의료 영상 편집 모델의 체계적인 평가를 위한 표준화된 벤치마크 제공

◦

임상적으로 관련성 있는 평가 지표 (편집 정확도, 맥락 보존, 시각적 품질) 제시

◦

최첨단 모델의 성능 비교 및 실패 패턴 분석을 통한 향후 연구 방향 제시

◦

주의 지점 기반 실패 분석 프로토콜을 통한 모델 개선 방향 제시

◦

의료 영상 편집 기술의 임상적 활용 가능성 증대

•

한계점:

◦

벤치마크에 포함된 데이터셋의 규모 및 다양성에 대한 추가적인 연구 필요

◦

특정 의료 영상 modality에 대한 편향 가능성 존재

◦

실제 임상 환경에서의 일반화 가능성에 대한 추가적인 검증 필요

Made with Slashpage