What is the reason for effect of gamma parameter in non-linear classification using Support Vector Machines?




The kernel equations in Support Vector Machines are given by :

In the RBF kernel, parameter ‘gamma’ is present.
I found it online that as we decrease the value of gamma, the variance of a model increases and it starts overfitting the data. Can anyone please share a resource explaining why this happens in terms of the kernel function mathematically?

Any help would be appreciated!


Hi Corporate_Cowboy,

You can refer to this link

Hope this will help you to understand.

closed #3