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

Autopruner: An end-to-end trainable filter pruning method for efficient deep model inference #18

Open
nocotan opened this issue Feb 17, 2021 · 0 comments
Assignees

Comments

@nocotan
Copy link
Member

nocotan commented Feb 17, 2021

一言でいうと

フィルタレベルのPruningをEnd-to-Endで行う学習可能な手続きを提案.

論文リンク

https://arxiv.org/pdf/1805.08941.pdf

著者/所属機関

Jian-Hao Luo, Jianxin Wu
National Key Laboratory for Novel Software Technology

投稿日付(yyyy/MM/dd)

2018/05/23

概要

Screen Shot 2021-02-17 at 10 52 02

既存のフィルタレベルのPruningはフィルタの選択とfine-tuningの2-stepから構成されていた.
これに対して提案手法はこれをEnd-to-Endで行い,かつ自動でこれらの決定を行うフレームワークを提案.

新規性・差分

  • 単一モデルでEnd-to-Endで学習可能なフィルタレベルのPruning:Pruning先のフィルタ選択とモデルのfine-tuningをEnd-to-Endで行う学習可能なフレームワークを提案.
  • 適応的な圧縮レートと複数レイヤーの圧縮:最適な圧縮レートが自動で決定される仕組みを提案
  • 良好な実験結果

手法

Screen Shot 2021-02-17 at 10 52 25

Pruning対象のネットワークにAutoPrunerレイヤーを追加する方式でPruningを提案.

結果

Screen Shot 2021-02-17 at 10 52 45

Screen Shot 2021-02-17 at 10 53 04

コメント

Attentionっぽい.

@nocotan nocotan self-assigned this Feb 17, 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