NeurIPS 2020

Robust Federated Learning: The Case of Affine Distribution Shifts

Meta Review

The paper studies robust federating averaging under affine shifts. All the reviewers agree that the paper has novel and interesting result. For camera ready, please take reviewers' feedback into account. In particular, a key weakness of the work is it's restriction to the affine setting. Comments on why the setting is interesting and if/how the result can be extended to more general setting would be useful.