I am new to the field of kernel analysis and I am trying to calculate sum of squares using multivariate normal kernel. So far I got

The multivariate normal kernel function:

$f(x;H)=\frac{1}{n}\sum K_{H}(x-X_i)$ where $K(z)=ϕ(z)=(2π)^{-p/2} e^{-1/2 z'z}$

Sum of Squares:

SSB=$\sum N_{i}(ϕ(m_i)-ϕ(m))(ϕ(m_i)-ϕ(m))^T$

SSW=$\sum\sum (ϕ(x_{ij})-ϕ(m_{i}))ϕ(x_{ij})-ϕ(m_{i}))^T$.

I tried to plug in $f(x;H)$ into the SSB and SSW, however, it gives me a complicated form that I am not able to simplify.

Is there a way to do this or is there any reference that can help me do it, because I can not find anything.


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.