Fig. 1: Overall view of the proposed ProxyFL.
From: Decentralized federated learning through proxy model sharing
ProxyFL is a communication-efficient, decentralized federated learning method where each client (e.g., hospital) maintains a private model, a proxy model, and private data. During distributed training, the client communicates with others only by exchanging their proxy model which enables data and model autonomy. After training, a client’s private model can be used for inference.
