Applied Federated Learning: Improving Google Keyboard Query Suggestions

TL;DR

This paper demonstrates how federated learning can be effectively applied at a global scale to improve virtual keyboard query suggestions while preserving user privacy, with measurable quality improvements.

Contribution

It presents an end-to-end application of federated learning in a commercial setting, showcasing its benefits for privacy and model performance.

Findings

Quality of query suggestions improved

Federated training metrics align with live deployment results

Enhanced user privacy without sacrificing model accuracy

Abstract

Federated learning is a distributed form of machine learning where both the training data and model training are decentralized. In this paper, we use federated learning in a commercial, global-scale setting to train, evaluate and deploy a model to improve virtual keyboard search suggestion quality without direct access to the underlying user data. We describe our observations in federated training, compare metrics to live deployments, and present resulting quality increases. In whole, we demonstrate how federated learning can be applied end-to-end to both improve user experiences and enhance user privacy.