On the effectiveness of partial variance reduction in federated learning with heterogeneous data

Open in new window