Preparation data sets
It would be helpful to explain the formation of the
data sets used in this study: first of all, the data was split
into two subsets; development set (60%) and test set
(40%). The test set was equally re-split into two parts
called validation set (20%) and dedicated test set (20%)
in the classification experiments. Development set was
used for feature selection and training of the NN model
while the dedicated test set was used only to examine the
performance of a NN model. Eventually, three data sets
were used as illustrated in Figure 3: development set,
validation set, and dedicated test set.