zbMATH — the first resource for mathematics

A modified algorithm for nonconvex support vector classification. (English) Zbl 1157.68434
Gammerman, A. (ed.), Artificial intelligence and applications. Machine learning. As part of the 26th IASTED international multi-conference on applied informatics. Calgary: International Association of Science and Technology for Development (IASTED); Anaheim, CA: Acta Press (ISBN 978-0-88986-710-9/CD-ROM). 46-51 (2008).
Summary: Support vector classifications (SVCs) are widely used as computationally powerful tools for binary classification. As an extension of \(\nu\)-SVC, Perez-Cruz et al. proposed Extended \(\nu\)?-SVC where a nonconvex quadratic programming (QP) problem is formulated and an iterative algorithm is applied to the problem. In the paper, we propose a modification for the existing algorithm of Extended \(\nu\)-SVC, which makes possible to analyze the finite convergence and local optimality of the algorithm. The modification is done so that the algorithm visits only a finite number of basic solutions of the nonconvex QP problem. Though the modification is theoretically rather than practically important, experimental results also show that the modification causes the algorithm to finish faster.
For the entire collection see [Zbl 1154.68012].
68T05 Learning and adaptive systems in artificial intelligence
90C26 Nonconvex programming, global optimization
90C20 Quadratic programming