How to write neural networks in form of a function???

 I want to use the ann function for optimization using genetic algorithm.. Thanks in advance

Answer 1

[Greg Heath]

 h =  tanh(b1 + IW*x);

 y = b2 + LW*h;

[Bharath]

I am getting error as : Inner matrix dimensions must agree

Basically I want to do multi objective optimization of 2 ann functions in ga. My neural network for both is Multi Layer Feed Forward with 16 hidden neurons, 1 hidden layer, transfer fn's : logsig, purelin

[Greg Heath]

I cannot help with code I cannot see.

I apologize for not being a mind reader.

Answer 2

[Greg Heath]

It is more efficient to have bipolar inputs, tansig hidden nodes and purelin outputs. Bipolar inputs and outputs can be obtained using ZSCORE, MAPMINMAX and/or MAPSTD. Since the corresponding equations are relatively straightforward, the following equations are for normalized quantities:

 [ I N ] = size(xn)
 [ O N ] = size(tn)
 [ O N ] = size(yn)
 [ H N ] = size(h)

 [ H 1 ] = size(B1)
 [ H I ] = size(IW)
 [ O 1 ] = size(B2)
 [ O H ] = size(LW)

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

Hope this helps.

Greg