MATLAB: Finding two layers to replace in googlenet

Question

Dear Altruists,

I was trying this example: https://www.mathworks.com/help/deeplearning/examples/train-deep-learning-network-to-classify-new-images.html for deep learning using googlenet.But its showing me error which is "Undefined function or variable 'findLayersToReplace' " in this two lines

[learnableLayer,classLayer] = findLayersToReplace(lgraph);
[learnableLayer,classLayer] 

I have tried this lines in the command prompt

edit(fullfile(matlabroot,'examples','nnet','main','freezeWeights.m'))
edit(fullfile(matlabroot,'examples','nnet','main','createLgraphUsingConnections.m'))
edit(fullfile(matlabroot,'examples','nnet','main','findLayersToReplace.m'))

But i'm not getting any fruitfull result…

Please help me creating the function of findLayersToReplace.

Thanks in advance.

Answer 1

For R2018a, you can follow this tutorial. Essentially it shows you what the findLayersToReplace function was doing.

For most transfer learning applications, we remove the final fully connected layer, softmax layer and classification layer.

These are then replaced with your own fc, softmax and classification layers.

https://www.mathworks.com/help/releases/R2018a/nnet/examples/transfer-learning-using-googlenet.html