[mlpack-git] [mlpack] Mean shift clustering (#388)

[Shangtong Zhang]

In http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/TUZEL1/MeanShift.pdf,
It seems kernel must be radially symmetric.
(using the notation in the paper above) What I use to calculate mean shift vector is the derivative of  k(s) rather than the derivative of K(x). 
For Gaussian kernel (u=1), dk(s)/ds = -0.5 * exp(-0.5 * s), dK(x)/dx = -x * exp(-0.5 * x). 
But for the function Gradient() in Gaussian Kernel class, I think it should implement dK(x)/dx.
It may be strange for Gradient() to implement dk(s)/ds.
I think it should have another name, but I really don't know what name is appropriate.


---
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/pull/388#issuecomment-72311878
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20150131/a4d706a6/attachment.html>