Fine-Tuning Personalization in Federated Learning to Mitigate Adversarial Clients

However, due to the heterogeneity between the clients' data distributions, the model obtained through the use of FL algorithms may perform poorly on some client's data.