Definition
Federated learning is a machine learning approach that trains an algorithm across multiple decentralized edge devices or servers holding local data samples, without exchanging the data itself. Instead of pooling raw data into a central location, the model is sent to the data, trained locally, and then the model updates are aggregated. This method is crucial for privacy preservation and reducing communication overhead, enabling AI development on sensitive datasets. For instance, a smartphone keyboard can learn user typing patterns and vocabulary without sending personal messages to a central server. Unlike traditional centralized learning where all data is collected, federated learning keeps data localized. This concept is relevant for cybersecurity professionals and AI practitioners, appearing in domains related to data privacy and distributed systems on certifications like CompTIA Security+.