Preprint
Federated Learning for Mobile Keyboard Prediction
Andrew Hard et al., arXiv, 2018
Read the originalRead on October 17, 2026
- What it is
- A 2018 paper from Google that trains a next-word prediction model for a phone keyboard on the phones themselves (federated learning), so that the typed text is not sent to a server. The authors report that this matched or improved prediction recall compared with training on servers.
- How much weight it carries
- A preprint by the team that built the system, so it describes a design they chose. It concerns next-word prediction, not speech recognition, and it describes the training method, not the policy of any product that uses it today.
Facts taken from this source
- A 2018 study shows that a language model can be trained on client devices without exporting sensitive user data to servers, and that the approach matched or improved prediction recall.
“This work demonstrates the feasibility and benefit of training language models on client devices without exporting sensitive user data to servers.” (Abstract)