3
$\begingroup$

From the literature we know a Gaussian Kernel/RBF has a corresponding Hilbert space with infinite number of dimensions.

Now my question is, specially in the context of a Gaussian Process, can I consider a Gaussian kernel as a linear combination of infinite number of Gaussian distributions?

I'm almost certain that's the case, but I can't find the right reference for it.

$\endgroup$
  • $\begingroup$ "consider a Gaussian kernel as a linear combination of infinite number of Gaussian distributions" Wouldn't this statement imply that a Gaussian mixture in general can be summarized exactly into one Gaussian distribution? Does not sound right to me. $\endgroup$ – wij Jul 7 '17 at 11:12
  • $\begingroup$ No, Gaussian functions are closed under summation only if they all have the same mean, otherwise adding two gaussian functions with different means wouldn't give you a single gaussian function. $\endgroup$ – adrin Jul 8 '17 at 14:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.