Federated learning is a distributed, on-device computation framework that enables training global models without exporting sensitive user data to servers. In this work, we describe methods to extend the federation framework to evaluate strategies for personalization of global models. We present tools to analyze the effects of personalization and evaluate conditions under which personalization yields desirable models. We report on our experiments personalizing a language model for a virtual keyboard for smartphones with a population of tens of millions of users. We show that a significant fraction of users benefit from personalization.
@article{arxiv.1910.10252,
title = {Federated Evaluation of On-device Personalization},
author = {Kangkang Wang and Rajiv Mathews and Chloé Kiddon and Hubert Eichner and Françoise Beaufays and Daniel Ramage},
journal= {arXiv preprint arXiv:1910.10252},
year = {2019}
}