Federated learning

This refers to a machine learning strategy that trains an algorithm across various devices or servers that hold local data samples, without data exchange. The model is locally trained on each device, and only the model updates are shared, ensuring privacy and enabling collaborative learning without sharing raw data.