Summary
2007 International Symposium on Nonlinear Theory and its Applications
2007
Session Number:18PM2-B
Session:
Number:18PM2-B-2
A NewWorking Set Selection for Decomposition-Type SVM Learning Algorithms
Norikazu Takahashi, Masashi Kuranoshita, Yusuke Kawazoe, Jun Guo, Jun’ichi Takeuchi,
pp.280-283
Publication Date:2007/9/16
Online ISSN:2188-5079
DOI:10.34385/proc.41.18PM2-B-2
PDF download (80.5KB)
Summary:
Decomposition methods are efficient iterative techniques for solving large quadratic programming (QP) problems arising in support vector machines. In each step, the decomposition method chooses a set of a small number of variables called the working set and then solves the QP problem with respect to those selected variables. In this paper, we propose a new working set selection method based on the conjugate gradient method and evaluate its effectiveness by using benchmark data sets on both pattern classification and regression problems.