Generating Differentially Private Datasets Using GANs

41 citations recorded by the source corpus.

How researchers see this paper

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

Privacy in Deep Learning: A Survey (2020, background): “Triastcyn & Faltings present a technique that generates synthetic datasets that still have statistical properties of the real data while providing differential privacy guarantees with respect to this data [61].”
LOGAN: Membership Inference Attacks Against Generative Models (2017, methodology): “We also evaluate our attack against a recently proposed technique for (ε, δ)-Differentially Private GANs [62], where Gaussian noise [16] is injected in the discriminator forward pass during training.”
Synthetic Data for Deep Learning (2019, methodology): “Data release approaches (here we use the terminology from [598]) that perform synthetic data generation have the following advantages:”
Security and Privacy Issues in Deep Learning (2018, methodology): “Depending on where the noise is added, such approaches can be divided into three groups: gradient-level [Abadi et al. 2016b; McMahan et al. 2018; Xie et al. 2018; Acs et al. 2018; Yu et al. 2019], objective-level [Chaudhuri and Monteleoni 2009; Phan et al. 2016, 2017a,b] and label-level [Papernot et”