Algorithmic Recourse on COMPAS
68.7AccuracyCFQ (ours, MixedPrec + CF objective)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CFQ (ours, MixedPrec + CF objective)Training Objective=CF objective, Quantization Strategy=MixedPrec2026.05 | 68.7 | 0.074 | 0.083 | |
| Standard train → MixedPrecTraining Objective=Standard, Quantization Strategy=MixedPrec2026.05 | 68.6 | 0.136 | 0.178 | |
| R-Margin train → MixedPrecTraining Objective=R-Margin, Quantization Strategy=MixedPrec2026.05 | 68.6 | 0.117 | 0.154 | |
| R-Consistency train → MixedPrecTraining Objective=R-Consistency, Quantization Strategy=MixedPrec2026.05 | 68.6 | 0.112 | 0.149 |