| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| COMPAS (test) | FR-Train | Accuracy62 | 28 | 4mo ago | |
| Synthetic 1.0 (test) | Ours | Accuracy72.7 | 28 | 4mo ago | |
| COMPAS | LabelBias-DP | DP Disparity-0.1743 | 16 | 4mo ago | |
| Adult | LabelBias-DP | Delta DP-0.1598 | 16 | 4mo ago | |
| Comm | LabelBias-DP | Delta DP0.1731 | 15 | 4mo ago | |
| German Credit (test) | Fair Self-Bounding Learning Algorithm | Equal Opportunity Difference5.15 | 15 | 1mo ago | |
| MEPS | LabelBias-DP | ΔDP-0.5527 | 13 | 4mo ago | |
| COMPAS | AOD-0.2034 | 12 | 3mo ago | ||
| CelebA level-3 snow noise (corrupted) | MSAM | Acc s+85.04 | 8 | 1mo ago | |
| CelebA (clean) | MSAM | Accuracy (s+)86.32 | 8 | 1mo ago | |
| CelebA corrupted level-3 Gaussian noise (test) | Accuracy (s+)97.63 | 8 | 1mo ago | ||
| CelebA clean (test) | MSAM | Accuracy s+98.01 | 8 | 1mo ago | |
| Adult n-m = 2 (sex, race) (Experiment 1) | MIFair | Accuracy84.6 | 8 | 2mo ago | |
| OULAD (test) | AUC69.7 | 7 | 1mo ago | ||
| HMDA (5-fold) | Pareto Non-Dominance Count5 | 7 | 2mo ago | ||
| Synthetic (5-fold) | Agarwal | Pareto Non-Dominance Count5 | 7 | 2mo ago | |
| Adult Income (5-fold) | Pareto Non-Dominance Count5 | 7 | 2mo ago | ||
| German Credit (5-fold) | CEC | Pareto Non-Dominance Count5 | 7 | 2mo ago | |
| COMPAS | YOTO | Inverted Hypervolume0.3661 | 5 | 26d ago | |
| COMPAS | FairBiNN | Hypervolume0.8409 | 5 | 26d ago | |
| Adult | GFB | Hypervolume0.8007 | 5 | 26d ago | |
| UTKFace | FairBiNN | Hypervolume88.76 | 5 | 26d ago | |
| CelebA | GFB | Hypervolume81.98 | 5 | 26d ago | |
| ACSIncome (test) | PCGrad | Macro F169.4 | 4 | 2mo ago | |
| Adult (test) | PCGrad | Macro F117.8 | 4 | 2mo ago |
