Architecture of the convolutional neural network, inspired from LeNet (LeCun et al. 1998), and comprised of three convolutional layers with 11 × 11, 7 × 7, and 3 × 3 kernel sizes, and 32, 64, and 128 filters, respectively, followed by three fully connected hidden layers with 2048, 512, and 32 neurons. ReLU activations were applied between each layers. Max-pooling layers with 2 × 2 kernel sizes and stride = 2 were inserted after the first two convolutional layers, and dropout of 0.5 was used before the fully connected layers.