Artificial Neural Networks (ANNs) are now extremely widespread tools in various technologies. In the simplest application, ANNs provide a feedforward architecture for connections between neurons. The feedforward neural network is the first and simplest type of ANN devised. In the presence of basic hypotheses that interact with some problems, the intrinsic unidirectional structure of feedforward networks is strongly limiting. However, it is possible to start from it and create networks in which the results of computing one unit affect the computational process of another. It is evident that algorithms that manage the dynamics of these networks must meet new convergence criteria.
In this chapter, we'll go over the main ANN architectures, such as convolutional NNs, recurrent NNs, and long short-term memory (LSTM). We'll explain...