A Novel Heuristic for Building Reduced-Set SVMs Using the Self-Organizing Map

摘要

We introduce a novel heuristic based on the Kohonen's SOM, called Opposite Maps, for building reduced-set SVM classifiers. When applied to the standard SVM (trained with the SMO algorithm) and to the LS-SVM method, the corresponding reduced-set classifiers achieve equivalent (or superior) performances than standard full-set SVMs.