federated learning

A machine learning approach that trains an algorithm across multiple decentralized edge devices or servers holding local data samples, without exchanging them.