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

One-Bit Over-the-Air Aggregation for Communication-Efficient Federated Edge Learning: Design and Convergence Analysis #9

Open
mitsuhiko-nozawa opened this issue May 28, 2021 · 0 comments

Comments

@mitsuhiko-nozawa
Copy link
Contributor

mitsuhiko-nozawa commented May 28, 2021

一言でいうと

one-bit broadband digital aggregation (OBDA)で多数のエッジデバイスが同時に通信できるように、コミュニケーションボトルネックを改善

論文リンク

https://arxiv.org/pdf/2001.05713.pdf

著者/所属機関

Guangxu Zhu, Yuqing Du, Deniz G ¨und ¨uz, and Kaibin Huang

Shenzhen Research Institute of Big Data, Shenzhen, China
The University of Hong Kong, Hong Kong
Imperial College of London, London, U.K.

投稿日付(yyyy/MM/dd)

IEEE 2020/11/26

概要

Federated Learningにおいて、同時にパラメータサーバーにアクセスする際に問題となるネットワークボトルネックを、勾配の符号を取る量子化手法と、集約(averaging, voting)された勾配を無線通信時に推定することで改善

新規性・差分

従来の研究はアナログ変調に関するものが多かったが、現在普及しているデジタル変調で利用できるようにした

手法

image
各エッジデバイスで定期的に計算された勾配の符号を、エッジサーバー(パラメータサーバー)に同時に通信、集約とボーティング後、再びエッジデバイスに通信することを繰り返す

結果

1つのエッジサーバー、100個のエッジデバイスを使ってMNISTとCifar-10を学習
MNISTでは6層ネットワーク、Cifar-10ではResNet18
3つのwireless hostilities下で測定
収束率は異なるがどのhostility下でも収束はしており、デバイス数が増えると精度が向上している(図右)

image
image

コメント

@mitsuhiko-nozawa mitsuhiko-nozawa self-assigned this May 28, 2021
@mitsuhiko-nozawa mitsuhiko-nozawa changed the title [WIP] One-Bit Over-the-Air Aggregation for Communication-Efficient Federated Edge Learning: Design and Convergence Analysis One-Bit Over-the-Air Aggregation for Communication-Efficient Federated Edge Learning: Design and Convergence Analysis Jun 1, 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