Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation #16

Open
subinium opened this issue Feb 7, 2021 · 2 comments

Comments

@subinium
Copy link
Owner

subinium commented Feb 7, 2021

@subinium subinium added the xai label Feb 7, 2021
@subinium
Copy link
Owner Author

subinium commented Feb 8, 2021

Introduction

이번 정리는 아래 도움되는 자료의 내용가 확실히 잘 설명해주고 있음

  • 저널 논문은 처음인 느낌(?)
  • image classification 시스템에서 모델의 해석은 중요
  • 모델 결과를 픽셀 단위로 decomposition 하여
  • heatmap style로 visualization!

General Concept

스크린샷 2021-02-09 오후 5 38 06

  • d차원 입력 x의 각 차원의 relevance score를 계산하는 것이 목적
  • f(x) = \sigma_{i=i}^d R_i
    • 각 픽셀의 기여(relevance)하는 값의 합 = 결과
    • 기여도가 양수면 그렇게 예측한 이유, 음수면 그것을 방해한 이유

Layer-wise relevance propagation

  • 각 layer 층에 대해서 기여도의 합은 보존

@subinium
Copy link
Owner Author

subinium commented Feb 8, 2021

@subinium subinium added the TBD label Feb 9, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant