Finite Newton method for implicit Lagrangian support vector regression

Article type: Research Article

Authors: Balasundaram, S.^{; *} | Kapil,

Affiliations: School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi, India

Correspondence: [*] Corresponding author: School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India. Tel.: +91 11 26704724; Fax: +91 11 26741586; E-mail: balajnu@gmail.com

Abstract: In this paper a finite Newton iterative method of solution for solving the implicit Lagrangian Support Vector Regression (SVR) formulation has been proposed. Unlike solving a quadratic programming problem for the case of the standard SVR the solution of the proposed method is obtained by solving a system of linear equations at each iteration of the algorithm. For the linear or nonlinear SVR the finite termination of the proposed method has been established. The algorithm converges from any starting point and does not need any optimization packages. Experiments have been performed on a number of interesting synthetic and real-world datasets. The results obtained by the proposed method are compared with the standard SVR. Similar or better generalization performance of the proposed method clearly demonstrates its effectiveness and applicability.

Keywords: Implicit Lagrangian support vector machines, Newton method, Support vector regression

DOI: 10.3233/KES-2011-0222

Journal: International Journal of Knowledge-based and Intelligent Engineering Systems, vol. 15, no. 4, pp. 203-214, 2011