Implementation of neural network equation

I am getting different output for neural network : O=purelin(W2*logsig(W1i*Xi+B1)+B2) when i implement in matlab and when i do it mathamatically in excel sheet. How to implement the equations of neural netwok. Actually I want to use this equation without using matlab once network is trained and tested.

Answer 1

[Greg Heath]

The N I-dimensional "I"nputs are normalized from x to xn The N O-dimensional "O"utput targets are normalized from t to tn

The O-dimensional NORMALIZED targets are obtained from

 yn = B2 + LW*logsig( B1 + IW * xn );

The final O-dimensional unnormalized outputs, y, are obtained from yn via the inverse tsetting of the original t to tn normalization of the target.

Hope this helps.

Thank you for formally accepting my answer

Greg

[mohit deshmukh]

How can I get the preprocessing setting of the trained neural network. As I have tried the way it is explained above but still the answer is deviating. I normalised manually.