Improving SVM for Learning Multi-Class Domains with Pareto Multi-Objective Optimization

摘要：

Multi-class classification is very useful in data mining and related applications. This paper proposes a multi-class classification method that divides a multi-class training data into several two-class training data sets. SVM, improved with tabu search is used to train two-class data sets. Training with these two-class data sets and optimization with the training results are critical. A Pareto multi-objective optimization strategy is applied to optimize these training results. This method uses tabu search in the search process. Moreover, AUC (Area Under the ROC Curve) is used as the evaluation criterion. The experimental results show that the proposed method has better learning performance.

作者: Xiaolong Zhang ZeweiQiu Xiaofang Zhang

作者单位: School of Computer Science and Technology, Wuhan University of Science and Technology, 430081, China

会议类型: 国际会议

会议名称: Third International Symposium on Intelligence Computation and Applications(ISICA 2008)(第三届智能自动化、计算与制造国际研讨会)

会议地点: 武汉

会议语种:英文

页码: 238-241

在线出版日期: 2008-12-19（万方平台首次上网日期，不代表论文的发表时间）