how can I access the outputs of the hidden layers in a neural network?

I have a 2 hidden layer network. I trained it using a set of input output data but after training I want to access the outputs of the hidden layers for applying SVD on the hidden layer output. Please let me know how can I do it.

Thank You.

[mahdis mbioeng]

is there any way to access the outputs of hidden layers with this? net.outputs{i}.processParams

Answer 1

[Greg Heath]

Take the I-H-O net and create an I-H net with no hidden layer.

See the section in the documentation on custom nets.

ADDITION:

Or, you can just write the code. If there is no normalization

IW = cell2mat(net.IW); b1 = cell2mat(net.b{1});

h = tansig( IW*x + repmat( b1,H,N ) );

Hope this helps.

Thank you for formally accepting my answer.

Greg

[Preeti Prajapati]

Thank you very much sir.

[Image Analyst]

You can thank him by "Accepting" his answer.

[Sriharsha]

what are X,H,N here?

[Greg Heath]

input matrix

number of hidden nodes

number of data points

HTH

GREG

[Sriharsha]

Thanks