Efficient Non-linear PCA

We present Kernel PCA with the Nyström method by deriving PCA for the Nyström randomized subsampling method. We provide a confidence bound to measure its statistical accuracy and apply it to the regression task to create Nyström principal component regression.