【作者】 王若鹏； 卓泽强；

【Author】 WANG Ruo-peng ZHUO Ze-qiang Department of Mathematic & Physics, Beijing Institute of Petro-chemical Technology, Beijing, 102617, China.

【机构】 北京石油化工学院数理部；

【摘要】 提出了一种新的支撑向量回归机(SVR)的光滑化方法,克服了已有SVR算法收敛速度慢且计算结构复杂的缺点．首先利用KKT条件,将SVR转化为不可微优化问题,再用调节熵函数构造出SVR问题的光滑一致逼近,然后利用现有无约束光滑优化的程序化算法求解．该算法对初始点不敏感,且收敛速度快、数值稳定．理论分析和数值实验结果表明模型和算法的可行性和有效性．更多还原

【Abstract】 In present paper, a novel smoothing function method for support vector regression (SVR) is proposed and attempts some drawbacks of former method which are complex, subtle, and sometimes difficult to implement. First, used KKT complementarity condition in optimization theory, unconstrained nondifferential optimization model are built. Then approximate algorithm adjustable entropy function is given. Finally, the paper trains the data sets with standard unconstraint optimization method. This algorithm is fast and insensitive to initial point. Theory analysis and primary numerical results illustrate that smoothing function method for SVR is feasible and effective.更多还原