Overview

Federated learning allows for 'privacy-preserving' AI. Instead of sending all data to a central server, the model is sent to the data.

The Process

  1. A central server sends a 'base model' to many devices (e.g., smartphones).
  2. Each device trains the model on its local data.
  3. The devices send only the 'updates' (gradients) back to the server.
  4. The server aggregates the updates to improve the global model and sends it back out.

Applications

  • Improving predictive text on mobile keyboards (e.g., Gboard).
  • Medical research across different hospitals without sharing patient records.
  • IoT device optimization.

Related Terms