Training support vector machines using Gilbert´s algorithm

Support vector machines are classifiers designed around the computation of an optimal separating hyperplane. This hyperplane is typically obtained by solving a constrained quadratic programming problem, but may also be located by solving a nearest point problem. Gilbert´s algorithm can be used to solve this nearest point problem but is unreasonably slow. In this paper we present a modified version of Gilbert´s algorithm for the fast computation of the support vector machine hyperplane. We then compare our algorithm with the nearest point algorithm and with sequential minimal optimization.