Abstract: Kernel principal component analysis is presented for kernel feature selection and High dimensional feature extraction to show kernel adaptations for nonlinear features selection of medical image data sets (MIDS). The proposed kernel principal component analysis extracts the salient features from a sample of unclassified patterns by use of a kernel. The kernel principal component analysis iteratively constructs a linear subspace of a high-dimensional feature space by exploiting a variance condition for the nonlinearly transformed samples. The resulting kernel subspace can be first chosen and then be processed for composite kernel subspace through the efficient combination representations used for further reconstruction and classification based on support vector machine.

Keywords: Support Vector Machine, Principal component analysis, data-dependent kernel, nonlinear subspace.