An algorithm of Model Selection for Support Vector Regression

摘要：

　　To solve the problem of SVR (support vector regression) model selection,this paper proposed a SVM (support vector machine) model parameter optimization algorithm based on gradient descent algorithm.The algorithm obtained the local optimal model parameter by minimizing the model evaluation criteria over the parameter set.Then on the basis of Riemannian geometry,a conformal transformation suitable for SVR was proposed which corrected kernel function in a data-based way.This algorithm can further enhance the generalization ability of SVR.The simulated results are illustrated to show the feasibility and effectiveness of the algorithm.

关键词： support vector regression (SVR) model selection gradient descent Riemannian geometry

作者: Li Xuesi Yang Hongqiao Sun Jing Bi Yangang Wu Yuanli

作者单位: The 309th Hospital of PLA Tsinghua University

会议类型: 国际会议

会议名称: 2012 2nd International Conference on Computer Application and System Modeling(2012第二届计算机应用与系统建模国际会议)(ICCASM-2012)

会议地点: 沈阳

会议语种:英文

页码: 96-99

在线出版日期: 2012-07-27（万方平台首次上网日期，不代表论文的发表时间）