Code Generation for Prediction of Machine Learning Model

Question

Inna Pelloso 2022-7-26

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1768410-code-generation-for-prediction-of-machine-learning-model

回答： Fred Smith 2022-8-24

Hi,

I have auto-generated C++ code using Matlab Coder for a machine learning model, that has a 1x27 double as input. Is there anyway to change the way some of the C++ is generated? Specifically, is there any way to #define the number of inputs (27 in this case) and just use the definition inside the auto-generated code (as opposed to the way 27 is being referenced multiple times, since it is breaking my API)?

Any help would be appreciated!

IP

static void main_prediction_mdl()
{
  real_T dv[27];
  real_T prediction_mdl;
  // Initialize function 'prediction_mdl' input arguments.
  // Initialize function input argument 'X'.
  // Call the entry-point 'prediction_mdl'.
  argInit_1x27_real_T(dv);
  prediction_mdl = prediction_mdl(dv);
}

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Fred Smith 2022-8-24

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1768410-code-generation-for-prediction-of-machine-learning-model#answer_1032410

Hi Inna,

If you make the input to the network variable-sized that will produce generic code. You can use coder.varsize to make a fixed-size input variable-sized.

In general there is no support for producing fixed-size code whose size is parametrized by a #define in MATLAB Coder today.

Cheers,

Fred