Federated Generative Privacy

TL;DR

FedGP introduces a federated learning framework using GANs to generate privacy-preserving artificial data, effectively reducing information disclosure risks while maintaining high-quality data for model training.

Contribution

This work presents FedGP, a novel federated GAN-based approach for privacy-preserving data sharing and model protection in federated learning.

Findings

FedGP generates high-quality labeled data for supervised learning.

FedGP significantly reduces vulnerability to model inversion attacks.

Empirical results demonstrate effective privacy preservation without sacrificing data utility.

Abstract

In this paper, we propose FedGP, a framework for privacy-preserving data release in the federated learning setting. We use generative adversarial networks, generator components of which are trained by FedAvg algorithm, to draw privacy-preserving artificial data samples and empirically assess the risk of information disclosure. Our experiments show that FedGP is able to generate labelled data of high quality to successfully train and validate supervised models. Finally, we demonstrate that our approach significantly reduces vulnerability of such models to model inversion attacks.