期刊
NEURAL COMPUTING & APPLICATIONS
卷 21, 期 5, 页码 945-955出版社
SPRINGER
DOI: 10.1007/s00521-011-0673-8
关键词
Least squares support vector machine; Kernel function; Fibonacci; Parameter; Multiclass classification
资金
- China Postdoctoral Science Foundation [2005038515]
- Science Foundation of Hebei University of Science and Technology [XL2006081]
The parameter values of kernel function affect classification results to a certain extent. In the paper, a multiclass classification model based on improved least squares support vector machine (LSSVM) is presented. In the model, the non-sensitive loss function is replaced by quadratic loss function, and the inequality constraints are replaced by equality constraints. Consequently, quadratic programming problem is simplified as the problem of solving linear equation groups, and the SVM algorithm is realized by least squares method. When the LSSVM is used in multiclass classification, it is presented to choose parameter of kernel function on dynamic, which enhances preciseness rate of classification. The Fibonacci symmetry searching algorithm is simplified and improved. The changing rule of kernel function searching region and best shortening step is studied. The best multiclass classification results are obtained by means of synthesizing kernel function searching region and best shortening step. The simulation results show the validity of the model.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据