StandardizeKernel: Standardize a Kernel Matrix

In SeojinBang/MKKC: Multiple Kernel K-means Clustering

Description

StandardizeKernel is used to standardize kernel matrices obtained from multiple views so that the views are comparable to each other. A kernel matrix x is centered and then scaled by dividing by its trace.

Usage

StandardizeKernel(x, center = TRUE, scale = TRUE)

Arguments

x	A kernel matrix to be standardized.
center	Logical. If TRUE, the kernel matrix x will be centered. The default is TRUE
scale	Logical. If TRUE, the kernel matrix x will be scaled. The default is TRUE.

Value

This function returns the standardized kernel matrix. If both arguments center and scale are TRUE, the kernel matrix will be centered first and then scaled.

References

\insertRef

scholkopf1998nonlinearMKKC

\insertRef

bang2018mkkcMKKC

Examples

x <- diag(3); x
StandardizeKernel(x, center = TRUE, scale = FALSE)   # centered
StandardizeKernel(x, center = FALSE, scale = TRUE)   # scaled
StandardizeKernel(x, center = TRUE, scale = TRUE)    # centered and scaled

SeojinBang/MKKC documentation built on Sept. 18, 2019, 1:42 p.m.