PARAMETER OPTIMIZATION FOR SVM USING SEQUENTIAL NUMBER THEORETIC FOR OPTIMIZATION

摘要：

In this paper, we propose a support vector machine (SVM) meta-parameter optimization method which uses Sequential Number Theoretic Optimization (SNTO) and gradient information for better optimization performance. SNTO is a new global optimization approach whose foundation is numeric and statistic theory This method has less computation time than genetic algorithm (GA) based and grid search based methods and better performance on finding global optimal value than gradient based methods. Simulations demonstrate that it is robust and works effectively and efficiently on a variety of problems.

关键词： Support Vector Machines (SVM) meta-parameter selection SNTO gradient descent method

作者: HUI-ZHI YANG XIAO-NAN JIAO LI-QUN ZHANG FA-CHAO LI

作者单位: College of Economics & Management, Hebei University of Science and Technology, Shijiazhuang 050018, China

会议类型: 国际会议

会议名称: 2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)

会议地点: 大连

会议语种:英文

页码: 3461-3464

在线出版日期: 2006-08-13（万方平台首次上网日期，不代表论文的发表时间）