Reimplementation of FedAWS

This is a reimplementation of FedAWS (Federated Learning with Only Positive Labels, ICML 2020) in PyTorch. The reimplementation algorithm comes from FedRepo.

Running Tips

Just use python train_lab.py then you can run FedAWS under cifar-10 dataset. You can change hyperparameters in train_lab.py.

Personal Homepage

• Homepage

Datasets

Only cifar-10 is on this repo. But you can also use cifar-100 and FaMNIST datasets after you download them.

Acknowledgement

This repo is based on FedRepo. Thanks for the author's great work.