Differentially Private Generative Adversarial Networks for Time Series, Continuous, and Discrete Open Data

By Lorenzo Frigerio, Anderson Santana de Oliveira, Laurent Gomez, Patrick Duverger. 2019

118 citations recorded by the source corpus.

Open data plays a fundamental role in the 21st century by stimulating economic growth and by enabling more transparent and inclusive societies. However, it is always difficult to create new high-quality datasets with the required privacy guarantees for many use cases. In this paper, we developed a differential privacy framework for privacy preserving data publishing using Generative Adversarial Networks. It can be easily adapted to different use cases, from the generation of time-series, to continuous, and discrete data. We demonstrate the efficiency of our approach on real datasets from the French public administration and classic benchmark datasets. Our results maintain both the original distribution of the features and the correlations among them, at the same time providing a good level of privacy.

Subfields: Artificial Intelligence, Electrical and Electronic Engineering, Sociology and Political Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Electrical and Electronic Engineering

In Sociology and Political Science

Citations in the corpus

DP-CGAN: Differentially Private Synthetic Data and Label Generation (2019, methodology): “Finally, in DPGAN frameworks [43, 18] the discriminator gradients are clipped and perturbed by adding Gaussian noise to gradients of the discriminator loss, while in our framework, Gaussian noise is added to the accumulation of clipped gradients of discriminator loss on real data and clipped gradien”
Machine Learning for Synthetic Data Generation: a Review (2023, unknown): “Regulated data such as clinical and genomics data in raw may not be shared and one solution is to share synthesized data instead.”
Synthetic Data for Deep Learning (2019, background): “[182] extend the DPGAN framework to continuous, categorical, and time series data.”