How to pass parameters to a custom kernel function in fitcsvm?

1 view (last 30 days)
Sepp
Sepp on 4 May 2016
Edited: Sepp on 4 May 2016
Hello
I'm using the fitcsvm method of Matlab for training a SVM classifier. Using the name-value pair 'KernelFunction', 'myKernel' it is possible to provide a custom kernel implementation. The kernel must be defined in a .m file in the following way
function G = kernel(U,V).
The only way I see for passing parameters to the kernel (like gamma for the Gaussian kernel) is to use global variables which is not so nice in the context of parallel execution.
Is there another possibility for passing parameters to the kernel?

Sign in to answer this question.

Answers (0)

Categories

Find more on Statistics and Machine Learning Toolbox in Help Center and File Exchange

Tags

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!