Title
Federated Learning with Adaptive Communication Compression Under Dynamic Bandwidth and Unreliable Networks
Abstract
•We propose a cloud-edge-clients architecture Cecilia for federated learning.•We propose a algorithm called ACFL to integrate into Cecilia.•ACFL can adaptively compress shared information according to network conditions.•The theoretical convergence of ACFL is analyzed without data distribution assumptions.•We conduct extensive experiments to validate the superiority of ACFL.
Year
DOI
Venue
2020
10.1016/j.ins.2020.05.137
Information Sciences
Keywords
DocType
Volume
Federated learning,Communication compression,Dynamic bandwidth,Unreliable networks
Journal
540
ISSN
Citations 
PageRank 
0020-0255
1
0.36
References 
Authors
0
6
Name
Order
Citations
PageRank
Xiongtao Zhang132.10
Xiaomin Zhu2921100.31
Ji Wang314012.56
Hui Yan471.80
Huangke Chen533916.53
Weidong Bao6236.49