Why do we have to normalize the input for an artificial neural network?

Question

It is a principal question, regarding the theory of neural networks:

Why do we have to normalize the input for a neural network?

I understand that sometimes when for example the input values are non-numerical a certain transformation must be performed, but when we have a numerical input? Why the numbers must be in a certain interval?

What will happen if the data is not normalized?

Answer 1

Normalization is required so that all values lie in a comparable range. Normalization makes model training faster because it converts large values into small values so that training takes less time.

Neural networks are expensive to train than most of machine learning algorithms, so that input data should be comprised of small values and lie in between a particular range e.g. (-1, 1).