Can anyone who has worked with both Matlab and Mathematica provide some comparison of how the Matlab Neural Network Toolbox compare to that of Mathematica.

Mathematica as far as I understand is a beautiful interface for MXNet and has its encoder and decoder in Wolfram Language that automates the data preparation for the network ingestion and back. So is Matlab doing something different and implementing its Neural Network completely from scratch in its own language and not using any open source implementation? If the answer is no then what open source implementation they are using?

  • 1
    $\begingroup$ It seems that you already know MMA part of the question. Is it on topic then? $\endgroup$
    – Kuba
    Feb 5, 2019 at 20:20
  • $\begingroup$ @Kuba Please don't down vote the question. Since many people using mathematica are multilingual, this is the ideal place to get some comparison. It is good to see what Matlab folks are doing since i see many jobs in deep learning asking for Matlab skills and mathematica share is still minimal. I am quite sure there would be many people here that can provide great insight. $\endgroup$
    – user13892
    Feb 5, 2019 at 20:26
  • $\begingroup$ Also while answering this question reveal more about the Mathematica side of things as well. $\endgroup$
    – user13892
    Feb 5, 2019 at 20:29
  • 1
    $\begingroup$ Unhappily, the MATLAB exchange didn't take off on area51. $\endgroup$ Feb 6, 2019 at 11:30
  • 1
    $\begingroup$ It seems like they may be using their own, ONNX-compliant, infrastructure: mathworks.com/help/deeplearning/ug/… $\endgroup$
    – Carl Lange
    Feb 6, 2019 at 12:54


Browse other questions tagged or ask your own question.