GNN-based Neighbor Selection and Resource Allocation for Decentralized Federated Learning

Meng, Chuiyang1; Tang, Ming2; Setayesh, Mehdi1; Wong, Vincent W. S.1

10.1109/GLOBECOM54140.2023.10437919

2023

IEEE Conference on Global Communications (IEEE GLOBECOM) - Intelligent Communications for Shared Prosperity

2334-0983

2576-6813

IEEE CONFERENCE ON GLOBAL COMMUNICATIONS, GLOBECOM

DEC 04-08, 2023

null,Kuala Lumpur,MALAYSIA

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

IEEE

Decentralized federated learning (DFL) enables clients to train a neural network model in a device-to-device (D2D) manner without central coordination. In practical systems, DFL faces challenges due to the dynamic topology changes, time-varying channel conditions, and limited computational capability of devices. These factors can affect the performance of DFL. To address the aforementioned challenges, in this paper, we propose a graph neural network (GNN)-based approach to minimize the total delay on training and improve the learning performance of DFL in D2D wireless networks. In our proposed approach, a multi-head graph attention mechanism is used to capture different features of clients and channels. We design a neighbor selection module which enables each client to select a subset of its neighbors for the participation of model aggregation. We develop a decoder which enables each client to determine its transmit power and CPU frequency. Experimental results show that our proposed algorithm can achieve a lower total delay on training when compared with three baseline schemes. Furthermore, the proposed algorithm achieves similar performance on the testing accuracy when compared with the full participation scheme.

其他

英语

CPCI-S

Engineering ; Telecommunications

Engineering, Electrical & Electronic ; Telecommunications

WOS:001178562001127

Web of Science

1.Univ British Columbia, Dept Elect & Comp Engn, Vancouver, BC, Canada
2.Southern Univ Sci & Technol, Dept Comp Sci & Engn, Shenzhen, Peoples R China

Meng, Chuiyang,Tang, Ming,Setayesh, Mehdi,et al. GNN-based Neighbor Selection and Resource Allocation for Decentralized Federated Learning[C]. 345 E 47TH ST, NEW YORK, NY 10017 USA:IEEE,2023.