Personalized federated learning (PFL) has become a hot research topic that can learn a personalized learning model for each client. Existing PFL models prefer to aggregate similar clients with similar data distribution to improve the performance of learning models. However, similarity-based PFL methods may exacerbate the class imbalance problem. In this article, we propose a novel dynamic affinity-based PFL (DA-PFL) model to alleviate the class imbalanced problem during federated learning. Specifically, we build an affinity metric from a complementary perspective to guide which clients should be aggregated. We then design a dynamic aggregation strategy that adjusts client aggregation based on the affinity metric in each round, thereby reducing the risk of class imbalance. Extensive experiments demonstrate that the proposed DA-PFL model can significantly improve the accuracy of each client in four real-world datasets with state-of-the-art comparison methods.
Yang et al. (Wed,) studied this question.