Fig. 1.

Schematic of the federated learning framework. The model is trained in a distributed manner: the institutions periodically communicate the local updates with a central server to learn a global model; the central server aggregates the updates and sends back the parameters of the updated global model