Optimizing for ROC Curves on Class-Imbalanced Data by Training over a Family of Loss Functions

Open in new window