Feature selection and representation for probabilistic neural network in medical data classification tasks

Abstract

This article presents the study regarding the problem
of feature selection and representation in the training
data sets used for the classification tasks performed by the
probabilistic neural network (PNN). Two methods for feature
representation are proposed. The first one utilizes the nodes of
the decision tree built in the classification problems with the use of
the single decision tree model. In the second method, the principal
component analysis is performed and the principal components
are included in the set of features for the classification tasks.
Depending on the form of the smoothing parameter, different
types of PNN models are explored. The prediction ability of the
PNNs trained on original and reduced data sets is determined
with the use of a 10-fold cross validation procedure.