A Novel Method for Solving Universum Twin Bounded Support Vector Machine in the Primal Space

In supervised learning, the Universum, a third class that is not a part of either class in the classification task, has proven to be useful. In this study we propose (N 𝔘 TBSVM), a Newton-based approach for solving in the primal space the optimization problems related to Twin Bounded Support Vector Machines with Universum data ( 𝔘 TBSVM). In the N 𝔘 TBSVM, the constrained programming problems of 𝔘 TBSVM are converted into unconstrained optimization problems, and a generalization of Newton’s method for solving the unconstrained problems is introduced. Numerical experiments on synthetic, UCI, and NDC data sets show the ability and effectiveness of the proposed N 𝔘 TBSVM. We apply the suggested method for gender detection from face images, and compare it with other methods.