Personalized and Private Peer-to-Peer Machine Learning

TL;DR

This paper presents a decentralized, asynchronous machine learning algorithm that ensures privacy, achieves provable convergence, and outperforms previous methods in personalized, privacy-preserving peer-to-peer settings.

Contribution

Introduces a novel decentralized, asynchronous algorithm with differential privacy guarantees for personalized machine learning in peer-to-peer networks.

Findings

Algorithm achieves provable convergence rate.

Differential privacy effectively protects personal data.

Significant performance improvements over non-private and isolated models.

Abstract

The rise of connected personal devices together with privacy concerns call for machine learning algorithms capable of leveraging the data of a large number of agents to learn personalized models under strong privacy requirements. In this paper, we introduce an efficient algorithm to address the above problem in a fully decentralized (peer-to-peer) and asynchronous fashion, with provable convergence rate. We show how to make the algorithm differentially private to protect against the disclosure of information about the personal datasets, and formally analyze the trade-off between utility and privacy. Our experiments show that our approach dramatically outperforms previous work in the non-private case, and that under privacy constraints, we can significantly improve over models learned in isolation.