Support vector learning with quadratic programming and adaptive step size barrier-projection

A novel efficient technique for extracting valid feature information

“…SVMs were first suggested for classification by Vapnik circa 1960s and have recently become an area of intense research, owing to developments in theory and practical techniques, combined with extensions to regression and density estimation (Drucker, Wu, & Vapnik, 1999;Fan & Palaniswami, 2000;To, Lim, Teo, & Liebelt, 2001;Vapnik, 1998). SVMs were motivated by statistical learning theory; the aim was to solve only the given problem, without solving a more difficult problem as an intermediate step.…”