federated learning

Noun

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