[SOLVED] Hardware implementation of AI non-linear functions from MATLAB

[april36]

Hello everybody,
I am a new graduate and currently learning about the implementation of some certain designs onto hardware (specifically from MATLAB/Simulink).
I want to ask if it is possible to implement an Artificial Intelligence block that is designed in MATLAB onto hardware considering especially the non-linear functions of it? If it is possible to implement these non-linear functions onto HW, how can it be done? I would appreciate your help.
Thank you.

 

[FvM]

Implementation in hardware can mean digital signal processing in DSP or FPGA, or analog processing in dedicated nonlinear analog circuits. Please clarify.

 

[crutschow]

What are the non-linear functions?

 

[april36]

Implementation in hardware can mean digital signal processing in DSP or FPGA, or analog processing in dedicated nonlinear analog circuits. Please clarify.

I am sorry, I should have mentioned the implementation of AI non-linear activation functions onto FPGA. I am wondering if it is possible. If so, how?

What are the non-linear functions?

AI non-linear functions such as tanh and sigmoid

 

[FvM]

Sure, suitable implementations are e.g. table with/without piecewise linear interpolation, polynomial approximation or cordic.

 

[april36]

I appreciate your help and will check these.
Last but not least, how would you recommend to implement an already designed AI block from MATLAB/Simulink to an FPGA? There are HDL code generator blocks of Simulink but I would like to hear an expert's opinion about it. If there is a more efficient and a better way.

 

[FvM]

If the code is good enough and not excessively resource hungry, why not?