Figure 2.

Network structure of the model used for binary classification. The network consists of six pairs of convolutional layers (with kernel size 3 × 3) and max-pooling layers. The feature maps are flattened and reduced to an output of size two. In total the network contains 7210 parameters.