Principle of KPCA
KPCA is a traditional PCA algorithm of nonlinear
development. The basic idea of KPCA is to input data
through the kernel method is mapped to feature space F, then
in the feature space F in the principal component analysis to
extract nonlinear features effectively.