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

Choosing the Sample with Lowest Loss makes SGD Robust #9

Open
nocotan opened this issue Jan 1, 2021 · 0 comments
Open

Choosing the Sample with Lowest Loss makes SGD Robust #9

nocotan opened this issue Jan 1, 2021 · 0 comments
Assignees

Comments

@nocotan
Copy link
Member

nocotan commented Jan 1, 2021

一言でいうと

各ステップで現在の損失が一番小さいようなサンプルを選択して更新するMin-k Loss SGD (MKL-SGD)を提案.

論文リンク

https://arxiv.org/pdf/2001.03316.pdf

著者/所属機関

Vatsal Shah, Xiaoxia Wu, Sujay Sanghavi (UT Austin)

投稿日付(yyyy/MM/dd)

2020/01/10

概要

  • 各ステップで現在の損失が一番小さいようなサンプルを選択して更新するMin-k Loss SGD (MKL-SGD)を提案.
  • 提案手法は計算コストが小さく,線形収束が可能で,外れ値に対してロバスト.

Screen Shot 2021-01-01 at 16 11 20

新規性・差分

ノイズにロバストなSGDの亜種を提案.

手法

Screen Shot 2021-01-01 at 17 11 27

Screen Shot 2021-01-01 at 16 11 49

結果

Screen Shot 2021-01-01 at 16 12 19

Screen Shot 2021-01-01 at 16 12 30

コメント

@nocotan nocotan self-assigned this Jan 2, 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