Enhancing Synthetic Data Generation for Class Imbalance and Privacy Preservation

Synthetic data generation has emerged as a powerful solution to meet the demand for high-quality, diverse, and privacy-preserving data in many domains. Still, there is an open challenge when dealing with class imbalance and privacy preservation in synthetic tabular data generation. Thus, this study introduces two algorithms: balanced Tabular Generative Adversarial Network (b-TGAN) and balanced Tabular Principle Component Analysis (b-TPCA). While b-TGAN proactively tackles class imbalance by incorporating a re-balancing mechanism and leveraging an Autoencoder, b-TPCA offers a privacy-preserving solution by generating synthetic data using statistical properties. Through experiments on five datasets, this study demonstrates the effectiveness of b-TGAN in generating balanced data, particularly in improving the performance on minority classes. b-TPCA also shows promising results, achieving comparable ML utility to the baseline method while enhancing privacy preservation.

I. Introduction

With the advances in artificial intelligence, the performance of machine learning (ML) models is often tied to the volume and quality of data they are trained on. However, obtaining a large amount of real-world data faces many problems. The collection process can be time-consuming and expensive. In addition, privacy concerns, particularly in sensitive domains like healthcare and finance, severely limit data accessibility. In such a context, some regulations such as the General Data Protection Regulation (GDPR) [1] restrict data acquisition and publication. However, the conflict between the necessity for data and the requirement to protect privacy has created a pressing need for innovative solutions. Besides, real-world datasets often face the problems of imbalance and data scarcity in under-representative classes.