Calculation process of artificial neural networks by nntool

It was analyzed by using the nntool.

for example

Number of Neuron 6 Number of Hidden layers 3 Transfer Function Tansig

Then, in nntool, iw {1,1} lw {2,1} bias1. I got the results.

I want to know that calculation to process of nntool using excel

Answer 1

[Greg Heath]

I don't know what problem you are trying to solve. However,

1 hidden layer with tansig transfer functions is a universal approximator.

The only reason you would consider 2 hidden layers is if the required number of hidden nodes is so high that it requires many more unknown weights than you have training equations.

In 30 years of experience with NNs, I have never encountered 3 hidden layers.

Again, 99% of the time only 1 hidden layer is necessary.

Hope this helps.

Greg