일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | ||||||
2 | 3 | 4 | 5 | 6 | 7 | 8 |
9 | 10 | 11 | 12 | 13 | 14 | 15 |
16 | 17 | 18 | 19 | 20 | 21 | 22 |
23 | 24 | 25 | 26 | 27 | 28 |
- deep learning
- contrastive learning
- Meta AI
- Computer Vision
- active learning
- Segment Anything 리뷰
- Stable Diffusion
- Segment Anything 설명
- Data-centric AI
- Self-supervised learning
- Data-centric
- iclr 2024
- CVPR
- ai 최신 논문
- 논문 리뷰
- iclr spotlight
- VLM
- cvpr 2024
- ssl
- Computer Vision 논문 리뷰
- cvpr 논문 리뷰
- Prompt Tuning
- Segment Anything
- iclr 논문 리뷰
- deep learning 논문 리뷰
- Prompt란
- Multi-modal
- ICLR
- 자기지도학습
- 논문리뷰
- Today
- Total
목록Computer Vision 논문 리뷰 (4)
Study With Inha
Entropy is not Enough for Test-time Adaptation:From the Perspective of Disentangled FactorsICLR 2024, Spotlight논문 링크: https://arxiv.org/abs/2403.07366github 링크: https://github.com/Jhyun17/DeYO프로젝트 페이지: https://whitesnowdrop.github.io/DeYO SOCIAL MEDIA TITLE TAGSOCIAL MEDIA DESCRIPTION TAG TAGwhitesnowdrop.github.io Introduction잘 학습된 모델이라 하더라도 distribution shift가 일어난 새로운 환경에서에서 robustness가 부족한 모..
AAAI 2024 accepted,Entropic Open-set Active LearningPaper Link: https://arxiv.org/abs/2312.14126GitHub: https://github.com/bardisafa/EOAL GitHub - bardisafa/EOAL: [AAAI 2024] An Implementation of Entropic Open-set Active Learning[AAAI 2024] An Implementation of Entropic Open-set Active Learning - bardisafa/EOALgithub.com IntroductionActive Learning (AL)은 레이블이 없는 dataset에서 가장 유익한 sample을 선별한 뒤, 해..
Google Research, NeurIPS 2023 acceptedStableRep: Synthetic Images from Text-to-Image ModelsMake Strong Visual Representation Learners논문 링크: https://arxiv.org/pdf/2306.00984.pdf StableRep은 NeurIPS 2023에 accept된 논문으로 LG AI Research에서 정리한 NeurIPS 2023 주요 연구주제에 선정된 논문이다.LG AI 리서치 블로그: https://www.lgresearch.ai/blog/view?seq=379 [NeurIPS 2023] 주요 연구 주제와 주목할 만한 논문 소개 - LG AI Research BLOGNeurIPS 2023,..
ECCV 2022, Visual Prompt Tuning, M. Jia et al. 논문 링크: https://arxiv.org/pdf/2203.12119.pdf 1. Introduction 최근 GPT 계열 모델과 같이 대규모 데이터와 대규모 모델을 활용한 딥러닝 연구가 많아졌다. 그러한 데이터의 경우 엔비디아나 구글과 같이 엄청난 컴퓨팅 파워를 가지고 있는 대기업이 아닌 일반인들은 Pretrain된 모델을 Fine-tuning하는 것도 어려운 상황에 이르렀다 :( 따라서 본 논문에서는 비전 분야에서 대규모 Transformer 모델을 효율적으로 활용하기 위한 새로운 fine-tuning 방법인 Visual Prompt Tuning (VPT)을 제안한다. 이는 기존의 fine-tuning 방법보다 더 ..