On-device federated learning with fuzzy logic based client selection
Zhaoyang Du, Celimuge Wu, Tsutomu Yoshinage, Lei Zhong, Yusheng Ji · 2022
With the rapid development of IoT, more and more advanced sensor devices are used to collect and process a large number of datasets. In order to protect user privacy while training machine learning models with multiple datasets, federated learning (FL) is introduced. Due to the heterogeneity of different IoT devices, the client selection method may greatly impact the performance of federated learning. In this paper, we introduce an enhanced version of our previous approach to address the client selection issue in a realistic FL environment. The proposed method considers the number of local data, computing capability, and network resources of each client participating in FL based on fuzzy logic. Then, the most suitable clients are selected for federated learning to get higher performance. Moreover, we use Raspberry Pi and laptops as FL clients and verify the superiority of the proposed method through extensive experiments.