I tried to generate C++ code for a deep learning neural network in MATLAB. I got the following error

%%This is where the trained network is called. The trained network is given
%%title as net.mat. I am also uploading the trained network for reference. The trained network can be opened using deep leanring netowrk model designer app. 


function label  = classifyX (y) %#codegen 
%CLASSIFYX Classify using SVM Model 
%  CLASSIFYX classifies the measurements in X 
%  using the SVM model in the file mySVM.mat, and then 
%  returns class labels in label.

persistent CompactMdl;
if isempty(CompactMdl)
    CompactMdl = coder.loadDeepLearningNetwork('net.mat');
end
label = predict(CompactMdl,y);

end

Error Message:

Error generating code for network.

Code generation for FeatureInputLayer input is not supported for mkldnn target.

See documentation for a list of supported layers with each target.

My Approach to Error:

I tried going through the webpage for understanding the various functions supported by MKLDNN. And Feauture input Layer is included. But I dont understand why it is still showing likewise.

https://de.mathworks.com/help/coder/ug/networks-and-layers-supported-for-c-code-generation.html

About the model:

The trained networks looks like this. It was working fine with inputs given in MATLAB. But I am facing error duing code generation.

The use case and the inputs matches with the approach descirbed in:

https://de.mathworks.com/help/deeplearning/ug/train-network-on-data-set-of-numeric-features.html

The difference is that I have modelled the input in simulink and created a funciton block with the model.

[Bill Chou]

Hi Swaroop, was pointed to your question by David.

Which release of MATLAB/Simulink are you running? FeatureInput layer was added in R2021a, so just wanted to confirm you're on at least that version.

[Naga Sai Pavan Swaroop Ainapurapu]

@Bill Chou thank you very much for quick response. I am using MATLAB 2020b. Then should i give it a try with R2021a or 2021b version?

[Bill Chou]

Great, that explains it. Yes go ahead and try R2021a or newer, that should resolve the issue.

FYI you can see layer/network support for a specific release of MATLAB by going to:

• https://www.mathworks.com/help/releases/R2020b/coder/ug/networks-and-layers-supported-for-c-code-generation.html
• https://www.mathworks.com/help/releases/R2021a/coder/ug/networks-and-layers-supported-for-c-code-generation.html
• https://www.mathworks.com/help/releases/R2021b/coder/ug/networks-and-layers-supported-for-c-code-generation.html
• https://www.mathworks.com/help/releases/R2022a/coder/ug/networks-and-layers-supported-for-c-code-generation.html

[Naga Sai Pavan Swaroop Ainapurapu]

sure, @Bill Chou. I shall go through and update you in case I needed further support. Thank you :)