基于支持向量机的算子逼近方法  

OPERATOR APPROXIMATION METHOD BASED ON SUPPORT VECTOR MACHINE

作　　者：马儒宁[1] 陈天平[1] 

机构地区：[1]复旦大学数学研究所,上海200433

出　　处：《系统科学与数学》2005年第5期634-640,共7页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金(69982003;60074005)资助课题.

摘　　要：Rn中连续算子的逼近问题的数值方法,一直是计算科学中研究的热点。本文引进了新兴的智能机器一支持向量机,以解决Rn中连续算子的逼近问题。在给出支持向量机用于算子逼近问题的详细数学表示之后,我们提出了分块逼近的算法,并通过具体的实例说明支持向量机在算子逼近问题中的有效性与优越性。The numerical methods for continuous operator approximation in R^n have attracted the attention of a great many researchers from the fields of theory and application. In this paper, we study a new intelligent machine- support vector machine for solving the problem of continuous operator approximation in R^n. After giving the mathematic formulation of support vector machine used in operator approximation, we introduce the dispartment algorithms. The concrete numerical test shows that our method does well in this problem.

关 键 词：连续算子 支持向量机 二次规划问题 逼近方法 回归估计 

分 类 号：O212.1[理学—概率论与数理统计]