Federated Machine Learning (FML) is a subset of DML that focuses on training machine learning models across decentralized devices holding local data, such as smartphones or IoT devices, without transferring this data to a central server. Instead, each device trains a local model on its data and shares only the model updates (e.g., gradients) with a central server, which aggregates these updates to build a global model. This methodology enhances data privacy and security, as raw data remains on local devices.