Machine Learning - Artificial Neural Networks
It was the neural networks in the human brain that inspired artificial neural networks. Taking into account the complexity of the human brain, scientists and engineers developed an architecture that fits into our digital world of binary computers. The diagram below illustrates one typical architecture.
We have an input layer with many sensors that collect data from the outside world. On the right hand side, we have an output layer, which predicts the outcome. There are several layers hidden between the two layers. As each layer increases complexity in training the network, its results would improve in most situations. We will now discuss several types of architectures that have been designed.
ANN Architectures
Below is a diagram showing several ANN architectures that have been developed over time and are in use today.
Source:
https://towardsdatascience.com/the-mostly-complete-chart-of-neural-networks-explained-3fb6f2367464
A neural network architecture is developed for a specific type of application. Therefore, if you want to use one for your machine learning application, you will have to design your own or use one of the existing architectures. There is no single guideline that says you should use a particular network architecture. The type of application you ultimately choose depends on your application needs.