In this section, we compare classification results of the selected CNN I performing alone and of several instances of the CNN I combined in a single model (which is referred as an “ensemble” model below). We also measure the impact of the size of the training data on the performance of a single CNN I.