Dec, 2023
因子辅助的个性化异构数据联邦学习优化
Factor-Assisted Federated Learning for Personalized Optimization with Heterogeneous Data
Feifei Wang, Huiyun Tang, Yang Li
TL;DRFederated learning aims to protect data privacy, and this paper proposes FedSplit, a personalized federated learning framework addressing the challenge of data heterogeneity by splitting hidden elements into shared and personalized groups, optimizing with a novel objective function. Additionally, factor analysis is introduced to decouple hidden elements, resulting in a practically implemented model referred to as FedFac, which demonstrates superior prediction performance compared to other state-of-the-art federated learning methods on real datasets.