A novel projection twin support vector machine for binary classification

Abstract

Based on the recently proposed projection twin support vector machine (PTSVM) and projection twin support vector machine with regularization term (RPTSVM), we propose a novel projection twin support vector machine (NPTSVM) for binary classification problems. Our proposed NPTSVM seeks two optimal projection directions simultaneously by solving a single quadratic programming problem, and the projected samples of one class are well separated from those of another class to some extent. Similar to RPTSVM, the singularity of matrix is avoided and the structural risk minimization principle is implemented in our NPTSVM. In addition, in our NPTSVM, we also discuss the nonlinear classification scenario which is not covered in PTSVM. The experimental results on several artificial and publicly available benchmark datasets show the feasibility and effectiveness of the proposed method.