Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators

About

The wide-spread availability of rich data has fueled the growth of machine learning applications in numerous domains. However, growth in domains with highly-sensitive data (e.g., medical) is largely hindered as the private nature of data prohibits it from being shared. To this end, we propose Gradient-sanitized Wasserstein Generative Adversarial Networks (GS-WGAN), which allows releasing a sanitized form of the sensitive data with rigorous privacy guarantees. In contrast to prior work, our approach is able to distort gradient information more precisely, and thereby enabling training deeper models which generate more informative samples. Moreover, our formulation naturally allows for training GANs in both centralized and federated (i.e., decentralized) data scenarios. Through extensive experiments, we find our approach consistently outperforms state-of-the-art approaches across multiple metrics (e.g., sample quality) and datasets.

Dingfan Chen, Tribhuvanesh Orekondy, Mario Fritz• 2020

Related benchmarks

TaskDatasetResultRank
Image GenerationCelebA 64 x 64 (test)
FID433.6
203
Image GenerationMNIST
FID61.34
44
Image GenerationFashion MNIST
FID131.3
38
Image GenerationCelebA 32x32 (test)
FID290
17
Differentially Private Image SynthesisMNIST
FID47.7
16
Differentially Private Image SynthesisCIFAR-10
FID194.4
16
Differentially Private Image SynthesisF-MNIST
FID97.2
16
Differentially Private Image SynthesisCelebA
FID290
16
Differentially Private Image SynthesisCAMELYON
FID291.8
16
Image ClassificationMNIST (test)
Accuracy (ε=10)80.75
14
Showing 10 of 17 rows

Other info

Follow for update