Using RBF kernel in SVM with high Gamma value. What does this signify?

svm
kernel

#1

I was going through the below-mentioned article:

And here goes the question

  1. Suppose you are using RBF kernel in SVM with high Gamma value. What does this signify?

And the answer in the article is: The model would consider only the points close to the hyperplane for modeling.

Actually, gamma or sigma in RBF is a nice approximation to K in KNN. so if the sigma value is high, then the model would consider even far away points.

You can try plotting using plot(exp(-x^2/(2*sigma^2))) in google.

Any explanations?


#2

Hi @sachinkalsi,

Higher values of Gamma implies that the influence of a single training example will be close while lower values of Gamma implies that the influence will be much more (far). The Gamma parameter works like the inverse of the radius of influence of selected samples. For more details, refer here.