Federated Learning

A machine learning approach where models are trained across multiple decentralized devices or servers holding local data, without exchanging raw data. Federated learning preserves privacy by keeping data on-device while sharing only model updates. Apple uses federated learning to improve Siri without accessing user recordings.