FedOVA: One-vs-All Training Method for Federated Learning with Non-IID Data

Zhu，Yuanshao1; Markos，Christos1,2; Zhao，Ruihui3; Zheng，Yefeng3; Yu，James J.Q.1

10.1109/IJCNN52387.2021.9533409

2021-07-18

International Joint Conference on Neural Networks (IJCNN)

2161-4393

978-1-6654-4597-9

Proceedings of the International Joint Conference on Neural Networks

2021-July

1-7

JUL 18-22, 2021

null,null,ELECTR NETWORK

345 E 47TH ST, NEW YORK, NY 10017 USA

IEEE

Federated Learning (FL) is a privacy-oriented framework that allows distributed edge devices to jointly train a shared global model without transmitting their sensed data to centralized servers. FL aims to balance the naturally conflicting objectives of obtaining massive amounts of data while protecting sensitive information. However, the data stored locally on each edge device are typically not independent and identically distributed (non-IID). Such data heterogeneity poses a severe statistical challenge for the optimization and convergence of the global model. In response to this issue, we propose Federated One-vs-All (FedOVA), an efficient FL algorithm that first decomposes a multi-class classification problem into more straightforward binary classification problems and then combines their respective outputs using ensemble learning. Experiments on several public datasets show that FedOVA achieves higher accuracy and faster convergence than federated averaging and data sharing. Furthermore, our approach can support practical settings with a large number of clients (up to 1000 clients) in FL.

Training Fluctuations Neural networks Distributed databases Speech recognition Collaborative work Data models

第一 ; 通讯

英语

CPCI-S ; EI

General Program of Guangdong Basic and Applied Basic Research Foundation[2019A1515011032]

Computer Science ; Engineering

Computer Science, Artificial Intelligence ; Computer Science, Hardware & Architecture ; Engineering, Electrical & Electronic

WOS:000722581700122

20214110996071

2-s2.0-85116476672

Scopus

1.Southern University of Science and Technology,Department of Computer Science and Engineering,China
2.Faculty of Engineering and Information Technology,University of Technology Sydney,
3.Tencent Jarvis Lab,

Zhu，Yuanshao,Markos，Christos,Zhao，Ruihui,et al. FedOVA: One-vs-All Training Method for Federated Learning with Non-IID Data[C]. 345 E 47TH ST, NEW YORK, NY 10017 USA:IEEE,2021:1-7.