基于支持向量机的散乱数据拟合  

Large Scattered Data Set Fitting by Support Vector Machine

作　　者：徐洪波[1] 程辉[1] 柳健[1] 田金文[1] 

机构地区：[1]华中科技大学图像所,武汉430079

出　　处：《计算机工程与应用》2005年第28期84-87,共4页Computer Engineering and Applications

摘　　要：文章提出了一种有效的大规模散乱点拟合方法,它是采用最小均方支持向量机局部拟合对Shepard插值方法进行改进。支持向量机基于结构风险最小化准则,在数据拟合方面具有较好的泛化能力,而改进的Shepard法能有效拟合大规模样本点。实验结果表明该算法对大规模散乱数据点具有较好的拟合性能。An efficient method for fitting large scattered data set is presented.h represents a improvement of the Shepard＇s method based on the local use of least square support vector machine.Support vector machine is designed to minimize the structural risk,and the generalization performance suits for function approximation.Modified Shepard＇s method can efficiently interpolate large data set.The proposed algorithm has been implemented and some results confirm its efficiency.

关 键 词：散乱数据集 支持向量机 改进Shepard法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]