[공지사항]을 빙자한 안부와 근황

Show more

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.

EarthCrafter: Scalable 3D Earth Generation via Dual-Sparse Latent Diffusion

SFNet: A Spatial-Frequency Domain Deep Learning Network for Efficient Alzheimer's Disease Diagnosis

Solving nonconvex Hamilton--Jacobi--Isaacs equations with PINN-based policy iteration

EndoControlMag: Robust Endoscopic Vascular Motion Magnification with Periodic Reference Resetting and Hierarchical Tissue-aware Dual-Mask Contro

SegQuant: A Semantics-Aware and Generalizable Quantization Framework for Diffusion Models

APTx Neuron: A Unified Trainable Neuron Architecture Integrating Activation and Computation

Artificial Intelligence for Green Hydrogen Yield Prediction and Site Suitability using SHAP-Based Composite Index: Focus on Oman

NeuroHD-RA: Neural-distilled Hyperdimensional Model with Rhythm Alignment

Photonic Fabric Platform for AI Accelerators

Feature-Enhanced TResNet for Fine-Grained Food Image Classification

Frequency-Dynamic Attention Modulation for Dense Prediction

Spatial Frequency Modulation for Semantic Segmentation

Fragment size density estimator for shrinkage-induced fracture based on a physics-informed neural network

SToFM: a Multi-scale Foundation Model for Spatial Transcriptomics

Advanced U-Net Architectures with CNN Backbones for Automated Lung Cancer Detection and Segmentation in Chest CT Images

Infinite Video Understanding

Fairness Evaluation of Large Language Models in Academic Library Reference Services

Cautious Next Token Prediction

How Well Does GPT-4o Understand Vision? Evaluating Multimodal Foundation Models on Standard Computer Vision Tasks

SpecCLIP: Aligning and Translating Spectroscopic Measurements for Stars

Prompt Guidance and Human Proximal Perception for HOT Prediction with Regional Joint Loss

Data-Driven Exploration for a Class of Continuous-Time Indefinite Linear--Quadratic Reinforcement Learning Problems

GTA: Grouped-head latenT Attention

Modeling Public Perceptions of Science in Media

LoX: Low-Rank Extrapolation Robustifies LLM Safety Against Fine-tuning

EXGnet: a single-lead explainable-AI guided multiresolution network with train-only quantitative features for trustworthy ECG arrhythmia classification

MRI-CORE: A Foundation Model for Magnetic Resonance Imaging

KVCache Cache in the Wild: Characterizing and Optimizing KVCache Cache at a Large Cloud Provider

Unified Sparse-Matrix Representations for Diverse Neural Architectures

Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start

Robot Operation of Home Appliances by Reading User Manuals

Deep Video Discovery: Agentic Search with Tool Use for Long-form Video Understanding

ORL-LDM: Offline Reinforcement Learning Guided Latent Diffusion Model Super-Resolution Reconstruction

Application of YOLOv8 in monocular downward multiple Car Target detection

RoBridge: A Hierarchical Architecture Bridging Cognition and Execution for General Robotic Manipulation

Monitoring digestate application on agricultural crops using Sentinel-2 Satellite imagery

Machine Learning-Based Modeling of the Anode Heel Effect in X-ray Beam Monte Carlo Simulations

DMS-Net:Dual-Modal Multi-Scale Siamese Network for Binocular Fundus Image Classification

Mapping Industry Practices to the EU AI Act's GPAI Code of Practice Safety and Security Measures

Parasite: A Steganography-based Backdoor Attack Framework for Diffusion Models

Attention-Based Multiscale Temporal Fusion Network for Uncertain-Mode Fault Diagnosis in Multimode Processes

AirCache: Activating Inter-modal Relevancy KV Cache Compression for Efficient Large Vision-Language Model Inference

Towards Detecting Persuasion on Social Media: From Model Development to Insights on Persuasion Strategies

ICCO: Learning an Instruction-conditioned Coordinator for Language-guided Task-aligned Multi-robot Control

Att-Adapter: A Robust and Precise Domain-Specific Multi-Attributes T2I Diffusion Adapter via Conditional Variational Autoencoder

A Deep Learning Approach for Augmenting Perceptional Understanding of Histopathology Images

ORANSight-2.0: Foundational LLMs for O-RAN

AlignDistil: Token-Level Language Model Alignment as Adaptive Policy Distillation

An Efficient and Precise Training Data Construction Framework for Process-supervised Reward Model in Mathematical Reasoning

Koel-TTS: Enhancing LLM based Speech Generation with Preference Alignment and Classifier Free Guidance

RAPID-Net: Accurate Pocket Identification for Binding-Site-Agnostic Docking

Graph Neural Networks for O-RAN Mobility Management: A Link Prediction Approach

Optimizing Privacy-Utility Trade-off in Decentralized Learning with Generalized Correlated Noise

Can We Generate Images with CoT? Let's Verify and Reinforce Image Generation Step by Step

RALAD: Bridging the Real-to-Sim Domain Gap in Autonomous Driving with Retrieval-Augmented Learning

Alleviating Seasickness through Brain-Computer Interface-based Attention Shift

NVS-SQA: Exploring Self-Supervised Quality Representation Learning for Neurally Synthesized Scenes without References

Onto-LLM-TAMP: Knowledge-oriented Task and Motion Planning using Large Language Models

A Survey of Event Causality Identification: Taxonomy, Challenges, Assessment, and Prospects

Flexible Coded Distributed Convolution Computing for Enhanced Straggler Resilience and Numerical Stability in Distributed CNNs

Vascular Segmentation of Functional Ultrasound Images using Deep Learning

Language model developers should report train-test overlap

BadHMP: Backdoor Attack against Human Motion Prediction

The FIX Benchmark: Extracting Features Interpretable to eXperts

Cross-domain Multi-step Thinking: Zero-shot Fine-grained Traffic Sign Recognition in the Wild

Impact of Stickers on Multimodal Sentiment and Intent in Social Media: A New Task, Dataset and Baseline

Constructing Optimal Noise Channels for Enhanced Robustness in Quantum Machine Learning

Multi-Level Explanations for Generative Language Models

Enhancing Sequential Recommender with Large Language Models for Joint Video and Comment Recommendation

Towards Efficient Generative Large Language Model Serving: A Survey from Algorithms to Systems

From DDMs to DNNs: Using process data and models of decision-making to improve human-AI interactions

ACMP: Allen-Cahn Message Passing with Attractive and Repulsive Forces for Graph Neural Networks

Deliberative Searcher: Improving LLM Reliability via Reinforcement Learning with constraints

Automated planning with ontologies under coherence update semantics (Extended Version)

Working with AI: Measuring the Occupational Implications of Generative AI

LEGO Co-builder: Exploring Fine-Grained Vision-Language Modeling for Multimodal LEGO Assembly Assistants

A Community-driven vision for a new Knowledge Resource for AI

Tournament of Prompts: Evolving LLM Instructions Through Structured Debates and Elo Ratings

Turing Test 2.0: The General Intelligence Threshold

Agent RL Scaling Law: Agent RL with Spontaneous Code Execution for Mathematical Problem Solving

More is Less: The Pitfalls of Multi-Model Synthetic Preference Data in DPO Safety Alignment

From Hypothesis to Publication: A Comprehensive Survey of AI-Driven Research Support Systems

A novel approach to navigate the taxonomic hierarchy to address the Open-World Scenarios in Medicinal Plant Classification

Balans: Multi-Armed Bandits-based Adaptive Large Neighborhood Search for Mixed-Integer Programming Problem

Learning Neural Strategy-Proof Matching Mechanism from Examples

LLM as a code generator in Agile Model Driven Development

Conflict Detection for Temporal Knowledge Graphs:A Fast Constraint Mining Algorithm and New Benchmarks

Large Learning Rates Simultaneously Achieve Robustness to Spurious Correlations and Compressibility

Pretraining on the Test Set Is No Longer All You Need: A Debate-Driven Approach to QA Benchmarks

Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains

Ultra3D: Efficient and High-Fidelity 3D Generation with Part Attention

Yume: An Interactive World Generation Model

Flow Matching Meets Biology and Life Science: A Survey

On the Interaction of Compressibility and Adversarial Robustness

AI Telephone Surveying: Automating Quantitative Data Collection with an AI Interviewer

From Feedback to Checklists: Grounded Evaluation of AI-Generated Clinical Notes

CASCADE: LLM-Powered JavaScript Deobfuscator at Google

How Should We Meta-Learn Reinforcement Learning Algorithms?

Vision Transformer attention alignment with human visual perception in aesthetic object evaluation

PRIX: Learning to Plan from Raw Pixels for End-to-End Autonomous Driving

Streaming 4D Visual Geometry Transformer

Created by

Haebom

저자

Dong Zhuo, Wenzhao Zheng, Jiahe Guo, Yuqi Wu, Jie Zhou, Jiwen Lu

개요

본 논문은 비디오로부터 4D 시공간 기하 정보를 인지하고 재구성하는 과제를 다룬다. 실시간 상호작용 애플리케이션을 위해, 자기회귀적 대규모 언어 모델과 유사한 철학을 공유하는 스트리밍 4D 시각 기하 트랜스포머를 제안한다. 인과 트랜스포머 구조를 사용하여 입력 시퀀스를 온라인 방식으로 처리하는 단순하고 효율적인 설계를 탐구한다. 시간적 인과적 어텐션을 사용하고 과거 키와 값을 암시적 메모리로 캐싱하여 효율적인 스트리밍 장기 4D 재구성을 가능하게 한다. 이 설계는 높은 품질의 공간 일관성을 유지하면서 과거 정보를 점진적으로 통합하여 실시간 4D 재구성을 처리할 수 있다. 효율적인 훈련을 위해, 밀집 양방향 시각 기하 기반 트랜스포머(VGGT)로부터 인과 모델로 지식 증류를 제안한다. 추론을 위해, 본 모델은 대규모 언어 모델 분야에서 최적화된 효율적인 어텐션 연산자(예: FlashAttention)의 마이그레이션을 지원한다. 다양한 4D 기하 인지 벤치마크에 대한 광범위한 실험은 제안된 모델이 경쟁력 있는 성능을 유지하면서 온라인 시나리오에서 추론 속도를 높여 확장 가능하고 상호 작용적인 4D 비전 시스템을 위한 길을 열었다는 것을 보여준다. 코드는 https://github.com/wzzheng/StreamVGGT 에서 이용 가능하다.

GitHub - wzzheng/StreamVGGT: Code for Streaming 4D Visual Geometry Transformer

Code for Streaming 4D Visual Geometry Transformer. Contribute to wzzheng/StreamVGGT development by creating an account on GitHub.

시사점, 한계점

•

시사점:

◦

실시간 4D 시각 기하 정보 재구성을 위한 효율적인 스트리밍 아키텍처 제안.

◦

인과적 트랜스포머와 암시적 메모리 활용을 통한 장기 의존성 처리 개선.

◦

지식 증류 기법을 통한 효율적인 모델 훈련.

◦

대규모 언어 모델의 최적화된 어텐션 연산자 활용을 통한 추론 속도 향상.

◦

확장 가능하고 상호 작용적인 4D 비전 시스템 구축 가능성 제시.

•

한계점:

◦

제안된 모델의 성능이 밀집 양방향 VGGT에 비해 얼마나 뒤쳐지는지에 대한 명확한 정량적 비교 부족.

◦

다양한 유형의 비디오 데이터에 대한 일반화 성능에 대한 추가적인 분석 필요.

◦

실제 응용 분야에서의 실시간 성능 및 안정성에 대한 평가 부족.

Made with Slashpage