stand for a theoretical PDF from which we
can generate an unlimited amount of data.
When another theoretical class
we can determine the classification performance of the
optimal Neyman-Pearson classifier:
which can be compared with the performance of the classifier
that uses the projected PDFs in place of