Fig. 1.

The architecture of the neural network models used for DeepCov. The number of intermediate convolutional layers and the size of the convolutional filters in these layers is variable. The three values in brackets in the right-hand cells describe the dimensionality of the data output from each layer. The values are given in the order: (number of feature channels, width, height). The width and height describe the spatial dimensions of the data and are both equal to the number of columns in the input alignment, m