How could I set C and gamma value for SVM in Matlab

257 Views Asked by At

I trt to train a pre-designed SVM with RBF kernel. I want to fix the C and gamma value before train. I use

Mdl = fitcsvm(Xapp,Yapp,'KernelFunction','rbf','KernelScale', 1,'BoxConstraint', 1,...);

But after training, the C(BoxConstraint) and gamma(KernelScale) are changed. How could I fix them?

There are two links show how could they change the parameters. But I don't know if they could fix them until train finished. 1 2

1

There are 1 best solutions below

0
On

Set 'OptimizeHyperparameters' as 'none' instead of 'auto'