DP-BREM: Differentially-Private and Byzantine-Robust Federated Learning with Client Momentum

This is the code of our work DP-BREM [USENIX Security' 25]

Two ways to run this code

1. Directly run main.py with the required parameters, see get_args() in arg.py for more details.
2. Use set_args() in arg.py to get the object args, then pass it into train_one_run_and_write_to_file(args, filename) in train.py. If you want to process multiple jobs, pass a list of args to sequntial_train_and_write_file(args_list, filename_list).