FFPDG: Fast, Fair and Private Data Generation

TL;DR

This paper introduces FFPDG, a novel data generation method that is fast, fair, private, and flexible, addressing biases and computational costs in synthetic data creation, with proven effectiveness through theoretical and empirical evaluations.

Contribution

The paper presents a new data generation approach that improves fairness, privacy, and speed over existing methods, with demonstrated theoretical and empirical benefits.

Findings

Models trained on generated data perform well on real tasks.

The method ensures fairness and privacy in synthetic data.

It reduces computational resources compared to GAN-based methods.

Abstract

Generative modeling has been used frequently in synthetic data generation. Fairness and privacy are two big concerns for synthetic data. Although Recent GAN [\cite{goodfellow2014generative}] based methods show good results in preserving privacy, the generated data may be more biased. At the same time, these methods require high computation resources. In this work, we design a fast, fair, flexible and private data generation method. We show the effectiveness of our method theoretically and empirically. We show that models trained on data generated by the proposed method can perform well (in inference stage) on real application scenarios.