CISPA
Browse
29_margctgan_a_marginally_better_.pdf (607.36 kB)

MargCTGAN: A “Marginally” Better CTGAN for the Low Sample Regime

Download (607.36 kB)
conference contribution
posted on 2024-03-26, 09:00 authored by Tejumade Afonja, Dingfan Chen, Mario Fritz
The potential of realistic and useful synthetic data is significant. However, current evaluation methods for synthetic tabular data generation predominantly focus on downstream task usefulness, often neglecting the importance of statistical properties. This oversight becomes particularly prominent in low sample scenarios, accompanied by a swift deterioration of these statistical measures. In this paper, we address this issue by conducting an evaluation of three popular synthetic tabular data generators based on their marginal distribution, column-pair correlation, joint distribution and downstream task utility performance across high to low sample regimes. The popular model shows strong utility, but underperforms in low sample settings in terms of utility. To overcome this limitation, we propose that adds feature matching of de-correlated marginals, which results in a consistent improvement in downstream utility as well as statistical properties of the synthetic data.

History

Primary Research Area

  • Trustworthy Information Processing

Name of Conference

German Conference on Pattern Recognition (GCPR)

Volume

14264

Page Range

524-537

Publisher

Springer Nature

Open Access Type

  • Not Open Access

BibTeX

@inproceedings{Afonja:Chen:Fritz:2024, title = "MargCTGAN: A “Marginally” Better CTGAN for the Low Sample Regime", author = "Afonja, Tejumade" AND "Chen, Dingfan" AND "Fritz, Mario", year = 2024, month = 3, pages = "524--537", publisher = "Springer Nature", issn = "1611-3349", doi = "10.1007/978-3-031-54605-1_34" }