Data Synthesis based on Generative Adversarial Networks

By Noseong Park, Mahmoud Mohammadi, Kshitij Gorde, Sushil Jajodia, Hongkyu Park. 2018

615 citations recorded by the source corpus.

Privacy is an important concern for our society where sharing data with partners or releasing data to the public is a frequent occurrence. Some of the techniques that are being used to achieve privacy are to remove identifiers, alter quasi-identifiers, and perturb values. Unfortunately, these approaches suffer from two limitations. First, it has been shown that private information can still be leaked if attackers possess some background knowledge or other information sources. Second, they do not take into account the adverse impact these methods will have on the utility of the released data. In this paper, we propose a method that meets both requirements. Our method, called table-GAN , uses generative adversarial networks (GANs) to synthesize fake tables that are statistically similar to the original table yet do not incur information leakage. We show that the machine learning models trained using our synthetic tables exhibit performance that is similar to that of models trained using the original table for unknown testing cases. We call this property model compatibility . We believe that anonymization/perturbation/synthesis methods without model compatibility are of little value. We used four real-world datasets from four different domains for our experiments and conducted indepth comparisons with state-of-the-art anonymization, perturbation, and generation techniques. Throughout our experiments, only our method consistently shows balance between privacy level and model compatibility.

Subfields: Artificial Intelligence, Sociology and Political Science, Generative Models

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 Sociology and Political Science

In Generative Models

Citations in the corpus

Synthesizing Tabular Data using Generative Adversarial Networks (2018, background): “Finally, perhaps the work that is closest to our work is tableGAN [31] - that is, it tries to solve the problem of generating synthetic data for a tabular dataset 1 .”
A Survey on Data Collection for Machine Learning: A Big Data - AI Integration Perspective (2018, background): “Synthetic Data Generative Adversarial Networks [57], [58], [59], [60], [61], [62] Policies [63], [64] Image [65], [66], [67], [68], [69], [70], [71] Text [72], [73], [74]”
GAN Memory with No Forgetting (2020, background): “Such a property can be used to deliver versatile data augmentation among different domains, which may, for example, benefit a downstream robust classification [61, 7].”
Deep Neural Networks and Tabular Data: A Survey (2021, methodology): “Restrictions apply. the distribution of the distance to closest record (DCR) can be calculated and compared to the respective distances on the test set [142].”
Modeling Tabular data using Conditional GAN (2019, methodology): “Previous models [6, 18] use min-max normalization to normalize continuous values to [−1, 1].”
When Machine Learning Meets Privacy (2020, background): “utility of the dataset can be guaranteed as the statistical similarity of models trained based on synthetic data and realist data has been shown repeatedly in the literature. For example, Park et al. [120] proved the statistical similarity of the generated synthetic tabular data and original data. X”
Synthetic data generation for tabular health records: A systematic review (2022, methodology): “Park 2018 [37] Preserve privacy Numerical Categorical Time-series (a)LA City worker records (b)Records of personal information (c)Records of medical information table-GAN* DCGAN Condensation method k-anonymity + t-closeness DP + δ-disclosure Micro-aggregation Post-randomisation (U)Compare cumulative”
Conditional Wasserstein GAN-based Oversampling of Tabular Data for Imbalanced Learning (2020, background): “table-GAN [21] is an interesting early work that adapts the widely popular Deep Convolutional GAN (DC-GAN) architecture [22] from image to structured data instead of designing an architecture speci cally for tabular data.”
Language Models are Realistic Tabular Data Generators (2022, methodology): “More recent methods for probabilistic tabular data modeling utilize generative adversarial networks (Choi et al., 2017; Park et al., 2018; Mottini et al., 2018; Xu et al., 2019; Koivu et al., 2020) or variational autoencoders (Xu et al., 2019; Ma et al., 2020; Vardhan & Kok, 2020; Darabi & Elor,…”