Replication Study of DECAF: Generating Fair Synthetic Data Using Causally-Aware Generative

This repository contains all code required to replicate our replication study of the paper NeurIPS 2021 DECAF: Generating Fair Synthetic Data Using Causally-Aware Generative networks. We built upon the (incomplete) code provided by the authors to repeat the first experiment which involves removing existing bias from real data with existing bias, and the second experiment where synthetically injected bias is added to real data and then removed.

The original code of DECAF paper is here

Prerequisites

Code is compatible with Python version 3.8.* due to explicit version requirements of some of the project dependencies.

We use DVC for storing trained models.

Installation

virtualenv env
source env/bin/activate
pip install -r requirements.txt

Downloading pretrained models

If you want to run the notebooks with pretrained models you need to download them first by running:

dvc pull

Contents

train.py - Training functions for different GAN models, including DECAF, Vanilla_GAN, WGAN_gp, FairGAN.

data.py - Loading data and preprocessing functions for Adult dataset and Credit dataset.

metrics.py - Measuring the Data Quality and Fairness.

experiment_1.ipynb, experiment_2.ipynb - A quick overview of reproduction results.

run_experiment_1.py, show_results.py - Scripts used for generating final results for Experiment 1.