Federated Learning - Private Distributed ML

This talk introduces federated learning for training machine learning models across nodes that cannot or will not share their data. It explains federated averaging, practical distributed-system challenges, security threats, communication costs, privacy guarantees, and when to use the approach.