FedGAN: Federated Generative Adversarial Networks for Distributed Data

TL;DR

FedGAN enables training generative adversarial networks across distributed, non-i.i.d. data sources efficiently, with theoretical convergence guarantees and reduced communication, suitable for privacy-sensitive applications.

Contribution

This paper introduces FedGAN, a novel federated GAN training algorithm with proven convergence and communication efficiency for distributed, non-i.i.d. data.

Findings

FedGAN converges similarly to centralized GANs across various datasets.

It reduces communication costs compared to traditional distributed GANs.

FedGAN demonstrates robustness to decreased communication frequency.

Abstract

We propose Federated Generative Adversarial Network (FedGAN) for training a GAN across distributed sources of non-independent-and-identically-distributed data sources subject to communication and privacy constraints. Our algorithm uses local generators and discriminators which are periodically synced via an intermediary that averages and broadcasts the generator and discriminator parameters. We theoretically prove the convergence of FedGAN with both equal and two time-scale updates of generator and discriminator, under standard assumptions, using stochastic approximations and communication efficient stochastic gradient descents. We experiment FedGAN on toy examples (2D system, mixed Gaussian, and Swiss role), image datasets (MNIST, CIFAR-10, and CelebA), and time series datasets (household electricity consumption and electric vehicle charging sessions). We show FedGAN converges and has similar performance to general distributed GAN, while reduces communication complexity. We also show its robustness to reduced communications.