Federated learning trains a shared model across many devices, such as phones, without collecting their data in one place.
In plain words
Normally you gather all the data on a server and train there. In federated learning, each device improves the model using its own data and sends back only the improvement. A central server combines those updates into a better shared model. The raw examples never leave the devices.
A quick example
A phone keyboard wants better word suggestions. Each phone learns a little from what its owner types, then sends only the model update to the server. The server averages updates from many phones into a better model and sends it back out, without ever receiving anyone's messages.
Where you'll meet it
- Keyboards and speech features that learn on your phone.
- Training on sensitive data that should not leave where it lives.
Related
Sources
- Docsfederated learning, Google Machine Learning Glossary · read 29 Sept 2026
- PaperCommunication-Efficient Learning of Deep Networks from Decentralized Data, McMahan et al. (arXiv) · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.