Skip to content

Reproducing kernel Hilbert space regularization, also known as Gaussian process regression or kriging in statistics.

License

Notifications You must be signed in to change notification settings

RoyCCWang/PatchMixtureKriging.jl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

44 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PatchMixtureKriging

Build Status Coverage

This package is deprecated. Please do not use.

Legacy notes

To install, start Julia REPL, press ], and run the command add PatchMixtureKriging. You might need a few more packages to run the scripts in the example folder; see the first few lines to see which packages are required for each file.

Example folder: IBB1D.jl is a 1D regression example using the iterated Brownian bridge kernel.

In the example folder, mixGP.jl: The binary splitting tree partitioning (BSP) scheme from Park et. al's "Patchwork Kriging for Large-scale Gaussian Process Regression" (https://jmlr.org/papers/v19/17-042.html) is used to subdivide the input domain into smaller overlapping regions; see patchGP_partitioning.jl for a visualization for a 2D domain.

mixGP.jl: using the BSP, a separate local Gaussian process regression is run for each region. The final solution is a input-varying convex combination (i.e. the weights depend on the input) of the local regression results. A write-up of the details is coming soon.

About

Reproducing kernel Hilbert space regularization, also known as Gaussian process regression or kriging in statistics.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages