What is det curve?

A DET curve (Detection Error Tradeoff curve) is a graphical representation of the performance of a binary classifier that varies the decision threshold. The curve plots the false positive rate (FPR) on the x-axis against the false negative rate (FNR) on the y-axis for different threshold values.

By adjusting the decision threshold, a classifier can trade off between FPR and FNR. A low threshold results in fewer false negatives but more false positives, while a high threshold results in fewer false positives but more false negatives. The DET curve visualizes the trade-off between these errors.

The steepness of the curve indicates the sensitivity of the classifier to changes in the decision threshold. A steeper curve indicates that small changes in the threshold can result in a significant change in classification performance.

DET curves are often used in biometric authentication research, as they can help evaluate the performance of different biometric systems with varying threshold values. They can also be used in other fields where binary classification is applied, such as fraud detection or spam filtering.