Parametric non-parallel support vector machines for pattern classification

Abstract

This paper proposes Parametric non-parallel support vector machines for binary pattern classification. Through an intelligent redesigning of the Support vector machine optimisation, not only do we bring noise resilience into the model, but also retain its sparsity. Our model exhibits properties similar to Support vector machines, hence many SVM related learning algorithms can be extended to make it scalable for large scale problems. Experimental results on several benchmark UCI datasets validate our claims.