Write a multi hidden layered $($with number of hidden layers as parameter$)$ neural network program with m input variables and n output variables. The size m$,$ n and number of perceptrons in hidden layers are parameterized. Initially use random number function to give weights to the edges. Generate your test data $($meaningfully$)$ using a simple algorithm, and then test it out on neural network. There must be at least $100$ test data for training.