2021. 5. 8. 16:06ㆍ관심있는 주제/뉴럴넷 질문
좋은 자료로 추후 정리해보기
https://link.medium.com/SMCl0Sdb5fb
An all-MLP Architecture for Vision
As the history of computer vision demonstrates, the availability of larger datasets coupled with increased computational capacity often…
medium.com
https://m.youtube.com/watch?v=KQmZlxdnnuY&feature=youtu.be
https://www.slideshare.net/JinwonLee9/pr317-mlpmixer-an-allmlp-architecture-for-vision
PR-317: MLP-Mixer: An all-MLP Architecture for Vision
Computer Vision 분야에서 CNN은 과연 살아남을 수 있을까요? 안녕하세요 TensorFlow Korea 논문 읽기 모임 PR-12의 317번째 논문 리뷰입니다. 이번에는 Google Research, Brain Team의 MLP-Mixer: An all-MLP Archit…
www.slideshare.net
channel-mixing MLP: Used for communication before different channels, each token is processed independently, that is, each line is used as input;
token-mixingMLP: Used for communication of different airspace positions, each channel legend is processed, that is, each column is used as input.
이 아키텍처를 정형데이터에 적용할 수 있을까부분은 해당 아키텍처에서 각 패치 위치를 얼마나 중요하게 생각하는 지 확인을 해봐야 알 것 같다.