Error using function 'Predict/MLFB' (#88.48.221)

I am trying to use a NN, which I pretrained and saved in a .mat file, in simulink (please see the figure attached). My objective is to deploy a block that uses a pretrained network to perform a regression on a random scalar, and then see it on a X,Ygraph.

However, I am getting the following error:

Index exceeds the number of array elements (1). Function 'Predict/MLFB' (#88.48.221), line 2, column 11: "deep.blocks.internal.deepNetwork({in_1}, {size(in_1)}, 'MMT/Predict', 'C:\Users\" Launch diagnostic report.

I am attaching also the files, so that you can launch this very simple model and tell me what is wrong.

Best regards,

Giovanni

Answer 1

[Ruth]

Dear Giovanni,

Thank you for your question and sharing your model.

This is a known issue related to the Predict block that was resolved in R2021a. You can avoid the error by upgrading to MATLAB R2021a or later.

If you are not able to upgrade, you can workaround the issue in R2020b by calling a wrapper predict function from a MATLAB Function Block. This wrapper function loads the network and calls predict. However, this will not be suitable for deployment workflows. I have attached a Simulink model to demonstrate this workflow.

Apologies for any inconvience this may have caused, please let me know if you have any further questions.

Best wishes,

Ruth

[Giovanni]

Dear Ruth,

thank you for your prompt answer. I'll update my version of MATLAB.

Best wishes, Giovanni