What formula is used when setting net.perfor​mParam.nor​​malizatio​n = 'standard' in combination with MSE

Hi,

does someone know how the error of the neural network is computed when setting up these command lines:

net.performParam.normalization = 'standard'
net.performFcn = 'mse'

In the matlab documentation there is only written, that it normalizes errors between -2 and 2, corresponding to normalizing outputs and targets between -1 and 1. But i can't figure out the formular being used.

I will be very thankful if someone can tell me the formular.

Answer 1

[Greg Heath]

It is just a straightforward linear transformation. For the target, t:

tn = a*t +b

-1 = a*tmin + b

1 = a*tmax + b

Just solve for a and b. Then plug in

t = (tn-b)/a

Hope this helps

Thank you for formally accepting my answer

Greg