What does OUtputs mean in neural network Tollbox ??

Hi,

I'm appliying a neural network for classification of my dataster so i used :

net = patternnet( hiddenLayerSize); to create the network ; [net, tr] = train(net, inputs, targets); for training ; outputs = net(inputs); to test the network;

My question is : What does the outputs matrix represent ?? , i'm triyng to find in which class the test examples have been classified without using the targets; so i need to understand the outputs !!

Can anybody help me please ?

[Greg Heath]

Do not have time for a more helpful answer right now.

Meanwhile, check out some of my code, comment and answers. Search on

 greg patternet

and look at the newest posts first.

Also: There may be something useful in the NEWSGROUP.

Answer 1

[Greg Heath]

For c classes, targets should be columns of the unit matrix eye(c)

 [ c N ]          = size(targets)
 trueclassindices = vec2ind(targets) % 1<= indices <= c
 targets          = ind2vec(trueclassindices) % columns of eye(c)
 ...
 assignedclassindices = vec2ind(outputs)
 err                  = ( assignedclassindices  ~= trueclassindices )
 Nerr                 = sum(err)
 PctErr               = 100*Nerr/N

To find the errors of class i, find the indices of the class i members by using the trueclassindices

 indi = (truclassindices == i) 
 Ni   = length(indi)

Do not have time for a more helpful answer right now.

Meanwhile, check out some of my code, comment and answers. Search on

 greg patternet

and look at the newest posts first.

End statements with semicolons after you understand what they produce

Also: There may be something useful in the NEWSGROUP.

Hope this helps.

Thank you for formally accepting my answer

Greg