摘抄 Federated learning (aka collaborative learning) is a machine learning technique that trains an algorithm across multiple decentralized edge devices or servers holding local data samples, without exchanging their data samples. 不同于： 传统centralized ML