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

Evaluating Model Robustness and Stability to Dataset Shift #52

Open
nocotan opened this issue May 26, 2021 · 0 comments
Open

Evaluating Model Robustness and Stability to Dataset Shift #52

nocotan opened this issue May 26, 2021 · 0 comments

Comments

@nocotan
Copy link
Member

nocotan commented May 26, 2021

一言でいうと

分布シフトにおける最悪ケース評価の推定量を提案.

論文リンク

http://proceedings.mlr.press/v130/subbaswamy21a/subbaswamy21a.pdf

著者/所属機関

Johns Hopkins University

投稿日付(yyyy/MM/dd)

AISTATS2021

概要

Screen Shot 2021-05-31 at 22 07 02

一般の分布シフトにおけるモデルのロバストネスを評価するため,データセットシフトの最悪ケースにおける誤差の推定量を構築.

新規性・差分

  • DRO手法ではありうるデータセットシフトの最悪ケースを改善するような学習を試みる
  • 本論文では,一致性を持つ最悪ケースの推定量を提案

手法

Screen Shot 2021-05-31 at 22 07 14

K-foldクロスバリデーションにより推定量を構築:

Screen Shot 2021-05-31 at 22 07 20

導出される推定量は以下:

Screen Shot 2021-05-31 at 22 09 02

推定量の性質

Screen Shot 2021-05-31 at 22 10 17

結果

Screen Shot 2021-05-31 at 22 10 53

Screen Shot 2021-05-31 at 22 10 59

コメント

@nocotan nocotan self-assigned this May 26, 2021
@nocotan nocotan changed the title [WIP] Evaluating Model Robustness and Stability to Dataset Shift Evaluating Model Robustness and Stability to Dataset Shift May 31, 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