EFFGAN: Ensembles of fine-tuned federated GANs

TL;DR

EFFGAN introduces an ensemble-based federated GAN framework that effectively learns data distributions across decentralized clients, mitigating client drift and enabling efficient training with many local epochs.

Contribution

The paper proposes EFFGAN, a novel ensemble of fine-tuned federated GANs, to address client drift and improve training efficiency in decentralized data settings.

Findings

EFFGAN outperforms existing methods on benchmark datasets.

It maintains high-quality data distribution learning with many local epochs.

The approach reduces communication costs compared to prior federated GANs.

Abstract

Generative adversarial networks have proven to be a powerful tool for learning complex and high-dimensional data distributions, but issues such as mode collapse have been shown to make it difficult to train them. This is an even harder problem when the data is decentralized over several clients in a federated learning setup, as problems such as client drift and non-iid data make it hard for federated averaging to converge. In this work, we study the task of how to learn a data distribution when training data is heterogeneously decentralized over clients and cannot be shared. Our goal is to sample from this distribution centrally, while the data never leaves the clients. We show using standard benchmark image datasets that existing approaches fail in this setting, experiencing so-called client drift when the local number of epochs becomes to large. We thus propose a novel approach we call EFFGAN: Ensembles of fine-tuned federated GANs. Being an ensemble of local expert generators, EFFGAN is able to learn the data distribution over all clients and mitigate client drift. It is able to train with a large number of local epochs, making it more communication efficient than previous works.