An MPEC Model for Selecting Optimal Parameter in Support Vector Machines

摘要：

In this paper, we present a new MPEC model for calculating the optimal value of cost parameter C for particular problems of linear non-separability of data. The objective function of the new model is an integer lower semi-continuous one. Smoothing technique is employed for solving this model, and the relationship between the MPEC model and its associated smoothing problem is given. It is proved that one of the global solution of the smoothing problem is also a solution of the MPEC problem. Numerical experiments show that this model is more efficient for choosing the parameter C.

关键词： Support vector machine cost parameter MPEC problem nonsmooth optimization.

作者: Yu-Lin Dong Zun-Quan Xia Ming-Zheng Wang

作者单位: College of Information Science and Technology, Shandong University of Science and Technology, Qingda Department of Applied Mathematics, Dalian University of Technology, Dalian 116024, China Department of Applied Mathematics, Dalian University of Technology, Dalian 116024, China Management

会议类型: 国际会议

会议名称: 首届伏化合系统生物学国际会议(The First International Symposium OSB07)

会议地点: 北京

会议语种:英文

页码: 351-357

在线出版日期: 2007-08-08（万方平台首次上网日期，不代表论文的发表时间）