Abstract: One of the major drawbacks of federated learning (FL) is data imbalance and uneven reliability of clients, which adversely impacts model performance and generalization ability. Among these ...