This code is used to obtain the results from our paper, FedGT: Identification of Malicious Clients in Federated Learning with Secure Aggregation, Marvin Xhemrishi, Johan Östman, Antonia Wachter-Zeh, and Alexandre Graell i Amat, accepted for publication at IEEE Transactions on Forensics and Information Security.
The code in this repo allows for simulating FedGT and some other aggregation techniques on several types of attacks. The datasets supported are MNIST, CIFAR10 and ISIC2019. In order to run different datasets, please change line 71 of main_threaded.py as follows:
cfg_path = "./cfg_files/cfg_[DATASET].toml"
where DATASET can be cifar, isic or mnist.
The hyperparameters can be set at the folder cfg_files and three different .toml file (one per dataset)
After the hyperparameters are set and line 71 has been change accordingly, run it as:
python main_threaded.py