What is Federated Learning?
Federated Learning — A decentralized approach to training AI where the model is trained across multiple edge devices holding local data.
Federated learning trains AI across multiple devices or organizations without sharing raw data. Each participant trains a local model copy, and only the model updates (not the data) are shared and aggregated. This is critical for healthcare, finance, and other privacy-sensitive industries.
Frequently Asked Questions
How does federated learning protect privacy?
Raw data never leaves its source. Only model weight updates are shared, making it impossible to reconstruct the original training data from the transmitted information.
Where is federated learning used today?
Google uses it for keyboard prediction on Android phones. Hospitals use it to train diagnostic models across institutions without sharing patient data. Banks use it for cross-institution fraud detection.
What are the limitations?
Communication overhead between participants, difficulty handling non-uniform data distributions across participants, and the need for careful aggregation to prevent model poisoning attacks.