Algorithmic Recourse on ADULT (Acc., VD, CRG)
85.7AccuracyStandard train → MixedPrec
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Standard train → MixedPrecTraining Objective=Standard, Quantization Strategy=MixedPrec2026.05 | 85.7 | 12.1 | 16.2 | |
| R-Margin train → MixedPrecTraining Objective=R-Margin, Quantization Strategy=MixedPrec2026.05 | 85.7 | 10.2 | 13.6 | |
| R-Consistency train → MixedPrecTraining Objective=R-Consistency, Quantization Strategy=MixedPrec2026.05 | 85.7 | 9.7 | 13.1 | |
| CFQ (ours, MixedPrec + CF objective)Training Objective=CF objective, Quantization Strategy=MixedPrec2026.05 | 85.7 | 6.1 | 7.1 |