-
Train and save a ML model first
-
Run 'centrailzed.py' with appropriate arguments to train and save a centralized ML mode
-
Run 'server.py' and one or more 'client.py' with appropriate arguments to run federated learning, non-simulated, and save a federally trained ML model. The server and clients can be on any machine.
-
Run 'simulate.py' to simulate federated learning, and save a federally trained ML model.
-
Run 'attack.py' with appropriate arguments to execute membership inference attack on a saved ML model
To use secure mode for server-client, run the 'generate.sh' or 'generate.bat' script in the folder 'certificate' to generate the required self-signed certificates
For information about script arguments run python <filename>.py -h