"Calculation of class confidences of Neural net -operator"

Hi everybody,

I have read about neural nets that basically there are different ways of calculating the class confidences. Among others, the entropy of the normalized network's output may be used etc...
Which one is the method that is implemented in RapidMiner (6.0.008)?
If possible please also provide a reference to a publication where the method is explained, since I need this for my thesis

Many thanks in advance and have a nice evening!