This repository contains the implementation of Differentially Private Conditional GAN(DP-CGAN). There are two different Conditional GANs(CGAN) that we made them differentially private: CGAN and Advanced CGAN. CGAN referes to a Vanilla CGAN architecture in which both generator and discriminator consist of two fully connected layers while in Advanced CGAN the generator and discriminator architectures are exactly the same as in infoGAN.
Here is the link to our slides for DP-CGAN:
If you use this code, please cite the following paper:
@inproceedings{torkzadehmahani2019dp,
title={DP-CGAN: Differentially Private Synthetic Data and Label Generation},
author={Torkzadehmahani, Reihaneh and Kairouz, Peter and Paten, Benedict},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops},
pages={0--0},
year={2019}
}