Communication-Efficient Federated Group Distributionally Robust Optimization

Zhishuai Guo, Tianbao Yang

Federated learning faces challenges due to the heterogeneity in data volumes and distributions at different clients, which can compromise model generalization ability to various di…