RBPK & EuDiC : A Narrative on SVM Efficiency Improvement

This book discusses the kernel function selection problem of SVM and its solution towards increasing the efficiency of SVM classifiers. It demonstrates that by combining the good characteristics of two kernel functions, it is possible to have a generalized kernel. It also discusses that the SVM can be made selection parameter invariant, if it is possible to avoid the kernel function for the classification of unseen samples during the classification phase. In a nutshell, this book gives a new kernel function, RBPK, and a classifier, EuDiC, in context of SVM.