There's more…

SVMs constitute a class of learning machines recently introduced in the literature. SVMs derive from concepts concerning the statistical theory of learning and present theoretical generalization properties. The theory that governs the functioning mechanisms of SVMs was introduced by Vapnik in 1965 (statistical learning theory), and was more recently perfected, in 1995, by Vapnik himself, and others. SVMs are one of the most widely used tools for pattern classification. Instead of estimating the probability densities of classes, Vapnik suggests directly solving the problem of interest, that is, to determine the decisional surfaces between the classes (classification boundaries).