Exploring the Impact of Attacks on Ring AllReduce

Jiayu Wang, Peng Liu, Zehua Guo*, Sen Liu, Chao Yao

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Citation (Scopus)

Abstract

Distributed Machine Learning (DML) is widely used to accelerate the training of the deep learning model. In DML, Parameter-Server (PS) and Ring AllReduce are two typical architectures. Recently, observing that many works address the security problem in PS, whose performance can be greatly degraded by malicious participation during the training process. However, the robustness of Ring AllReduce, which can solve the communication bandwidth problem in PS, to the malicious participant is still unknown. In this paper, we design a series of experiments to explore the security problem in Ring AllReduce, and reveal it can also suffer from the malicious participant.

Original languageEnglish
Title of host publicationProceeding of the 5th Asia-Pacific Workshop on Networking, APNet 2021
PublisherAssociation for Computing Machinery
Pages12-13
Number of pages2
ISBN (Electronic)9781450385879
DOIs
Publication statusPublished - 24 Jun 2021
Event5th Asia-Pacific Workshop on Networking, APNet 2021 - Shenzhen, China
Duration: 24 Jun 202125 Jun 2021

Publication series

NameACM International Conference Proceeding Series

Conference

Conference5th Asia-Pacific Workshop on Networking, APNet 2021
Country/TerritoryChina
CityShenzhen
Period24/06/2125/06/21

Fingerprint

Dive into the research topics of 'Exploring the Impact of Attacks on Ring AllReduce'. Together they form a unique fingerprint.

Cite this