Train Neural Net with 2 dimensional input and 1 dimensional output

Hi,

Despription of the data

I have 2 input arrays (X1 and X2) and 1 target array (T1). Where I want to fit a neural network (feedforwardnet) on; thus a network with 2 inputs and 1 output.

The dimensions of both input and output is 1x13600. The data is corresponding; X1(i) and X2(i) corresponds to T1(i).

Further, I divided this data up in a train-, validation and testset (each with a size of 1x1000).

My question consists mainly out of 2 parts.

• How do I train a neural network with the data given above that is based on the 2 dimensional input and the already divided data sets?

• My second question is a more basic one. But how do you determine what a suitable model is; number of hidden layers, number of neurons, learning algorthm and transferfunction.

Thanks in advance!

Answer 1

[Mahesh Taparia]

Hi,

You can use ‘fitnet’ function to train the network. For more information, you can refer to this link. For second question, you can try with different parameters and different algorithms to find the optimum model.

Hope it will help.