KDRcpp is a R/C++ port of the KDR Matlab implementation by K. Fukumizu.
KDR is a supervised dimension reduction method that uses RKHS to express conditional independance and find the dimension reduction subspace. It involves solving a non-trivial optimization problem.
Currently only Gaussian kernels are supported.
The package is not available on CRAN. To install, run:
# install package devtools if necessary
devtools::install_github('aschmu/KDRcpp')