Convolutional layers

Since the CNN section was covered in Chapter 2, Deep Learning Basics, you should know in which context CNNs are commonly used. In that section, we have mentioned that each layer of the same CNN can have a different implementation. The first three sections of this chapter describe possible layer implementations in detail, starting from the convolutional layers. But first, let's recap the process by which CNN perceive images. They perceive images as volumes (3D objects) and not as bi-dimensional canvases (having width and height only). The reason is the following: digital color images have a Red-Blue-Green (RGB) encoding and it is the mixing of these colors that produces the spectrum that can be perceived by human eyes. This also means that CNNs ingest images as three separate layers of color, one on top of the other. This translates into receiving a color...