One-Shot Federated Learning

TL;DR

This paper introduces one-shot federated learning, enabling a central server to learn a global model in a single communication round, significantly improving performance over local models.

Contribution

It proposes a novel one-shot federated learning method based on ensemble learning and knowledge aggregation, achieving substantial performance gains.

Findings

51.5% average relative gain in AUC over local baselines

Within 90.1% of the global ideal performance

Method is effective with a single communication round

Abstract

We present one-shot federated learning, where a central server learns a global model over a network of federated devices in a single round of communication. Our approach - drawing on ensemble learning and knowledge aggregation - achieves an average relative gain of 51.5% in AUC over local baselines and comes within 90.1% of the (unattainable) global ideal. We discuss these methods and identify several promising directions of future work.