Equation From a simple feedforward neural network

Hi every one, I'm working on a code which provide the full equation from a FeedForword Neural Network (FNN) in a text file consedering all weights and biases : I have already the trained FNN stored (the "net" file), the first step is to see if I have the same result when using :

load Net17 net 
input = [12,0.2]; % an input example
output = net(input');

and when I do this :

 IW = net.IW{1,1} ;
 b1 = net.b{1};
 b2 = net.b{2};
 LW = net.LW{2,1};
 y = b2 + LW * tansig( IW * input' +b1 );

Note that my net is a simple FNN with one hiden layer; 2 input neurons, 17 hiden neurons (for this example) and one output neuron.

I don't know if a did a mistake but the result is diffrent :

output = 353.3947
y =  -7.8709

any suggestions ??? Thanks...

References :

http://fr.mathworks.com/help/nnet/ug/multilayer-neural-network-architecture.html?refresh=true http://fr.mathworks.com/help/nnet/ref/setwb.html http://fr.mathworks.com/help/nnet/ref/tansig.html http://fr.mathworks.com/matlabcentral/answers/165233-neural-network-how-does-neural-network-calculate-output-from-net-iw-net-lw-net-b

Answer 1

[Greg Heath]

You forgot that the net uses the default MAPMINAX to normalize the input and target before training and, then, to denormalize the output.

Hope this helps.

Thank you for formally accepting my answer

Greg

[Mohamed BENALLAL]

Thank you very much !! it's working now !!