Federated Learning for Training Brain-Computer Interfaces

Federated learning is a privacy-preserving Machine Learning technique that could unlock the participation of a large number of distributed collaborating clients fostering the development of effective models that need personal and sensitive data to be trained and tested. In this paper, we present a preliminary evaluation of such cutting-edge techniques that may leverage the future adoption of innovative wearable devices. Experimental activities concern the recognition of motor imagery (MI) through EEG signals to build brain-computer interfaces (BCIs).