Federated Learning
A machine learning approach where models are trained across multiple decentralized devices or servers without sharing raw data, preserving privacy.
How It Works
Each device trains a local model on its own data. Only the model updates (gradients or weights) are sent to a central server, which aggregates them into a global model. Raw data never leaves the device.
Applications
Mobile keyboard prediction (Google Gboard), healthcare AI trained across hospitals without sharing patient records, financial fraud detection across banks, and any scenario where data cannot be centralized due to privacy or regulation.
Challenges
Non-IID data (each device has different data distributions), communication overhead, handling stragglers (slow devices), and security against adversarial participants who might try to poison the model.