Saved in:
Bibliographic Details
Main Authors: Qu, Linping, Song, Shenghui, Tsui, Chi-Ying, Mao, Yuyi
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2310.16652
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910295787569152
author Qu, Linping
Song, Shenghui
Tsui, Chi-Ying
Mao, Yuyi
author_facet Qu, Linping
Song, Shenghui
Tsui, Chi-Ying
Mao, Yuyi
contents Because of its privacy-preserving capability, federated learning (FL) has attracted significant attention from both academia and industry. However, when being implemented over wireless networks, it is not clear how much communication error can be tolerated by FL. This paper investigates the robustness of FL to the uplink and downlink communication error. Our theoretical analysis reveals that the robustness depends on two critical parameters, namely the number of clients and the numerical range of model parameters. It is also shown that the uplink communication in FL can tolerate a higher bit error rate (BER) than downlink communication, and this difference is quantified by a proposed formula. The findings and theoretical analyses are further validated by extensive experiments.
format Preprint
id arxiv_https___arxiv_org_abs_2310_16652
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle How Robust is Federated Learning to Communication Error? A Comparison Study Between Uplink and Downlink Channels
Qu, Linping
Song, Shenghui
Tsui, Chi-Ying
Mao, Yuyi
Machine Learning
Because of its privacy-preserving capability, federated learning (FL) has attracted significant attention from both academia and industry. However, when being implemented over wireless networks, it is not clear how much communication error can be tolerated by FL. This paper investigates the robustness of FL to the uplink and downlink communication error. Our theoretical analysis reveals that the robustness depends on two critical parameters, namely the number of clients and the numerical range of model parameters. It is also shown that the uplink communication in FL can tolerate a higher bit error rate (BER) than downlink communication, and this difference is quantified by a proposed formula. The findings and theoretical analyses are further validated by extensive experiments.
title How Robust is Federated Learning to Communication Error? A Comparison Study Between Uplink and Downlink Channels
topic Machine Learning
url https://arxiv.org/abs/2310.16652