MATLAB: Training-progress Plots not supported for compiled EXE

deep learning plot training progress executable compilerMATLAB Compiler

classOptions = trainingOptions('sgdm', …

…

'Plots','training-progress');

I have a Deep Learning script that I would like to deploy as an exe.

I get the error message that the training progress plots are not supported.

2 Questions:

a) How can I make a switch inside my program that determines if it is run from within Matlab or from the Exe?

i.e. if "matlab"

'Plots','training-progress');

else

'none','training-progress');

b) the training progress outputs on my screen. How can I capture these numbers in an array, so that I can plot it myself at the end when running the executable version?

thanks!

Best Answer

To check if the code is run in the MATLAB mode or deployed mode, use ‘isdeployed’ function. Refer to the following link to know more about the function:

https://www.mathworks.com/help/compiler_sdk/ml_code/isdeployed.html

Related Solutions

MATLAB: How to compile Deep learning Neural Network function

Starting R2016b MATLAB release:

You should be able to compile 'trainNetwork' and most command line functions (from both classical and deep learning networks) starting in R2016b.

Functions that cannot be compiled include the deep learning training "plot" function and all user interfaces.

Please refer the 'Neural Network Toolbox' product in this link for information on this:

https://www.mathworks.com/products/compiler/compiler_support.html

Prior to R2016b release:

You can only compile a pre-trained network and use classify function to classify the test data.

So in order to compile the doc example below,

https://www.mathworks.com/help/releases/R2017b/nnet/examples/create-simple-deep-learning-network-for-classification.html

in releases prior to R2016b, please follow the steps below:

1. Run the example code in MATLAB

2. This will create the 'convnet' struct variable in your workspace. This is the pretrained network object. Save this to a mat file like below:

save 'model.mat' convnet

3. Also save the testImageData variable in the workspace to a mat file:

save 'testDigitData.mat' testDigitData

4. Then you can create a MATLAB function like below to be compiled into an executable that used the pretrained network to classify the test data.

function accuracy = runModelFromMATLAB()\nload('model.mat');\nload('testDigitData.mat')\n\nYTest = classify(convnet,testDigitData);\nTTest = testDigitData.Labels;\naccuracy = sum(YTest == TTest)/numel(TTest)\nend

5. Then create executable for this function with MATLAB compiler

MATLAB: How to deploy the Training of a Neural Network in a standalone executable Matlab GUI App

Answering my own question,

Matlab's MCC Compiler actually supports training of Neural Networks now, but not the displaying of the training UI as shown here:

So, the solution was to invoke training without creating this UI as such:

options = trainingOptions('sgdm', 'MaxEpochs',8, 'ValidationData',{XValidation,YValidation}, 'ValidationFrequency',30, 'Verbose',false,'Plots','training-progress');

This needed to be changed to:

options = trainingOptions('sgdm', 'MaxEpochs',1, 'ValidationData',{XValidation,YValidation},'ValidationFrequency',30, 'Verbose',true, 'Plots','none');

Another issue upon execution was with ONNX Runtime Exporting (stalled when trying to create). To solve this, I needed to include the library directory of onnx runtime in the mcc compilation as described here:

https://www.mathworks.com/matlabcentral/answers/500806-how-can-i-use-importonnxnetwork-function-on-a-production-machine-which-only-has-mcr-installed

and as shown here:

I hope this can help someone!

Best Answer

Related Solutions

MATLAB: How to compile Deep learning Neural Network function

MATLAB: How to deploy the Training of a Neural Network in a standalone executable Matlab GUI App

Related Question