This action is done by default at the end of training. The profile of the output
errors are plotted against number of epochs, e.g. the mean square error and
the maximum square error as seen in Figure 2.
At this point the main script terminates, but the network definition and the trained values of the network are retained by MATLAB for further analysis.