摘要
支持向量顺序回归机是标准支持向量分类机的一个推广,它是一个凸的二次规划问题。本文根据l1范数与l2范数等价关系和优化问题的对偶原理,把凸的二次规划转化成线性规划。由此提了支持向量顺序回归机的线性规划算法,进一步用数值实验验证了此算法的可行性和有效性。并与支持向量顺序回归机相比,它的运行时间缩短了,而且误差i不超过支持向量顺序回归机。
Support vector ordinal regression machine is the extension of standard support vector machine, which is a convex quadratic programming. Using l1 norm and l2 norm equivalence theorem and dual theorem of optimize problem, a convex quadratic programming can be changed to a linear programming. Therefore, the linear programming algorithm of sup- port vector ordinal regression is present. Compared of support vector ordinal regression, the numerical experiment verify that the error evaluation is not increased and cputime is decreased for linear programming form of support vector ordinal regression.
出处
《兵团教育学院学报》
2010年第2期36-38,共3页
Journal of Bingtuan Education Institute
关键词
顺序回归问题
支持向量顺序回归机
二次规划
线性规划
Ordinal regression problem
Support vector ordinal regression machine
quadratic programming
linear programming