Characteristic showing relation among probability of two errors: False Acceptance vs False Rejection, when threshold of acceptance/rejection is changeable. Is suitable for binary classification or detection tasks. In this type of characteristic axis are no linear. More FAR and FRR error are similiar to gaussian more characteristic is linear .
Those plots are sufficient to place well known EER (Equal Error Rate) and minDCF (minimal Decision Cost Function) - error when we assume cost of error.
File main.py presents example of use scripts. In the second part
of file main.py include FRR vs FAR - cumulative distributions
of two types of errors well discribe ROC curve.
