The Planning-ahead SMO Algorithm

2013/07/31 by Tobias Glasmachers, Glasmachers, Tobias
Computer Science · #FOS: Computer and information sciences #Machine Learning (cs.LG) #cs.LG

paper · pdf · doi:10.48550/arxiv.1307.8305

arxiv created 2013/07/31 · arxiv updated 2013/08/01

Abstract

The sequential minimal optimization (SMO) algorithm and variants thereof are the de facto standard method for solving large quadratic programs for support vector machine (SVM) training. In this paper we propose a simple yet powerful modification. The main emphasis is on an algorithm improving the SMO step size by planning-ahead. The theoretical analysis ensures its convergence to the optimum. Experiments involving a large number of datasets were carried out to demonstrate the superiority of the new algorithm.

Related

• A Rank-One-Update Method for the Training of Support Vector Machines
• Faster SVM Training via Conjugate SMO
• Dual SVM Training on a Budget
• Training a Support Vector Machine in the Primal
• Iteratively-Reweighted Least-Squares Fitting of Support Vector Machines: A Majorization--Minimization Algorithm Approach