双支持向量顺序回归机

Twin Support Vector Ordinal Regression

针对顺序回归问题的深入研究,基于支持向量顺序回归机提出了双支持向量顺序回归机。由于双支持向量顺序回归机所对应的2个优化问题是对称的,则只需求解其中的一个问题,进而得出一个分划超平面。又因其对应的优化问题的规模只是支持向量顺序回归机规模的一半,故其运算速度会快于支持向量顺序回归机。数值实验的结果表明:双支持向量顺序回归机在一些数据分析中具有较高的正确率。

With further research at ordinal regression,TSVOR which is in the spirit of SVOR is proposed.TSVOR determines a plane by solving a SVM-type problem,since the two related SVM-type problems are identical.It is faster than SVOR because the problem corresponding to it is smaller than that of SVOR on several benchmark datasets.Some experimental results indicate that TSVOR also has a higher accuracy.