Kernel PCA Feature Extraction of Event-Related Potentials for Human Signal Detection Performance

Abstract

In this paper, we propose the application of the Kernel PCA techniquefor feature selection in high-dimensional feature space where inputvariables are mapped by a Gaussian kernel. The extracted features areemployed in the regression problem of estimating human signal detectionperformance from brain event-related potentials elicited by taskrelevant signals. We report the superiority of Kernel PCA for featureextraction over linear PCA.