Definition
A performance measurement for classification models that penalizes false confidence, quantifying the difference between predicted probabilities and the true binary labels.
In short: A performance measurement for classification models that penalizes false confidence, quantifying the difference between predicted probabilities and the true binary labels.