FCT_GAN

This code is for paper FCT-GAN: Enhancing Table Synthesis via Fourier Transform, which is published in CIKM 2023.

Structure

• Real_Datasets Location of training data

• Fake_Datasets Location of generated data

• model Location of FCT-GAN model. The generator and discriminator structure definition is under model/synthesizier/fctgan_synthesizer.py. The main training logic is also within this file.

Environment

The code is tested in following package version

numpy==1.21.0
torch==1.9.1
pandas==1.2.4
sklearn==0.24.1
dython==0.6.4.post1
scipy==1.4.1

Dataset

We upload the Adult dataset as an example, we are not the owner of this dataset, please check its original link for more information.

Example

Experiment_Script_Adult.ipynb is a demo notebook. Under Real_Datasets, there are three csv files: Adult.csv, Adult_corr.csv and Adult_type.csv. They are corresponding to three column permutation orders of Adult dataset. The ML utility and Statistical similarity evaluations are also in the demo notebook.

To cite this paper, use the following bibtex for now:

@inproceedings{10.1145/3583780.3615202,
author = {Zhao, Zilong and Birke, Robert and Chen, Lydia Y.},
title = {FCT-GAN: Enhancing Global Correlation of Table Synthesis via Fourier Transform},
year = {2023},
isbn = {9798400701245},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3583780.3615202},
doi = {10.1145/3583780.3615202},
booktitle = {Proceedings of the 32nd ACM International Conference on Information and Knowledge Management},
pages = {4450–4454},
numpages = {5},
keywords = {tabular data, gan, fourier transform},
location = {Birmingham, United Kingdom},
series = {CIKM '23}
}