Fairness-aware Synthetic Data Generation on Adult dataset
0.92PrecisionOriginal data D
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Original data Dsource=original biased data2021.10 | 0.92 | 0.936 | 0.807 | 0.116 | 0.18 | |
| DECAF-DPfairness_constraint=Demographic Parity2021.10 | 0.781 | 0.881 | 0.672 | 0.001 | 0.001 | |
| DECAF-NDfairness_constraint=No Debiasing2021.10 | 0.78 | 0.92 | 0.781 | 0.152 | 0.198 | |
| DECAF-FTUfairness_constraint=Fairness through Unawareness2021.10 | 0.763 | 0.925 | 0.765 | 0.004 | 0.054 | |
| DECAF-CFfairness_constraint=Counterfactual Fairness2021.10 | 0.743 | 0.875 | 0.769 | 0.003 | 0.039 | |
| WGAN-GPtype=Wasserstein GAN with Gradient Penalty2021.10 | 0.683 | 0.914 | 0.798 | 0.12 | 0.189 | |
| FairGANtype=Fairness-aware GAN2021.10 | 0.681 | 0.814 | 0.766 | 0.009 | 0.097 | |
| GANtype=Standard Generative Adversarial Network2021.10 | 0.607 | 0.439 | 0.567 | 0.023 | 0.089 |