Building a CNN architecture

CNN architectures are commonly made by stacking multiple conceptual logical blocks of layers one after another. These logical blocks are all structured the same way, with the same type of layer and layer connections, but they can be different in terms of their parameter configurations, such as the size of the filters, the stride, the type of padding used, and the amount of padding used. The simplest logical convolutional block is a convolutional layer, pooling layer, and activation function, in that order. Padding is a term that’s used to refer to any extra pixels that are added around the input image to preserve its spatial dimensions after convolution. Logical blocks are a way for you to describe and reference the architecture simply and efficiently. They also allow you to build CNN architectures in a depth-wise scalable way without the need to create and set the settings of each layer one by one. Depth is the same as deepness and refers to the...