I am reading this article: https://www.sciencedirect.com/science/article/pii/S2210650221000249
There, a multi layered particle swarm optimization of CNN parameters is presented. First step (layer) is to determine the architecture of the CNN and the second is to determine hyperparameters of each layer. On this image, all parameters (and their ranges) which are being determined are presented:
As you can see, there are three parameters regarding the architecture of the CNN, which are :
- Number of convolutional layers
- Number of pooling layers
- Number of fully connected layers
What I am confused about is (and it is not written in the article) when these parameters are determined, how to build an actual CNN out of this. I understand that all fully connected layers are at the end of the CNN. I also understand that Convolutional layer is usually followed by a Pooling layer. But what I don't understand is how to determine the actual architecture if we have a different amount of Convolutional and Pooling layers. What is the order of those layers if that is the case?
