auroc
LibAUC: A Deep Learning Library for X-Risk Optimization
Confound-matched controls for classifier claims: 1:1 matching, bootstrap AUROC, anchored verdicts, and a refusal when the control is vacuous.
