Gaussian Process Flavours is a collection of algorithms based on Gaussian Process (GP) Regression implemented in Matlab. Codes are general enough to be used also in different libraries.
Algorithm | Code | Description |
---|---|---|
Kernel Eigenvalue Decomposition | KernelDecomposition.m | GP kernel eigenvalue decomposition based on Mercer's Theorem |
Fast Approximate GPIS | FAGP.m | GP Implicit Surface algorithm relying on kernel eigenvalue decomposition to speed up computation |
Logarithmic GPIS in 2D | LogGPIS2D.m | GP Implicit Surface providing accurate description of the distance from obstacles exploiting gradient information |
Online Recursive GPIS | recursiveGPIS.m | Online computation of GP suitable for data streams |
The mathematical description can be found in this pdf file.