Incremental learning of oblique projection-based kernel discriminator: Proof and example

• 1、Institute of Intelligent Information Processing, Guizhou University, Guiyang 550025

Abstract：The problem of classifier design may be set in the framework of function approximation.Previously, we proposed to solve the problem in a Reproducing Kernel Hilbert Space (RKHS) continuously defined onthe pattern feature space, adopted oblique projection to discriminate a pattern class, called the target class,from other classes, by obliquely projecting a pattern feature vector onto the subspace spanned by the training patternfeatures of the target class, along the subspace spanned by those of other classes, and provided an incrementallearning algorithm for online training and sparsification of the discriminator. In this manuscript, we provide the proofsof the related theorems and an application example in incremental restoration of signals using two-channel samples.

Keywords： Kernel-based nonlinear classifier kernel discriminator oblique projection incremental learning signal restoration