Separate the test and training set. Design a 2 and 3 layer network. Change the number of neurons in the hidden layers. Interpret how the system is affected when the number of layers changes (2-3), the learning factor changes, the number of iterations changes, and the number of neurons changes. Interpret standard backprobagation and try momentum separately. Turn the results into a table or graph.

You will write the algorithm yourself, the ready library will not be used, the language selection is free.