基于支持向量机的空间数据拟合  

Space Data Fitting Based on the Support Vector Machine-SVM

作　　者：冯杨民[1] 梁晶[1] 

机构地区：[1]长安大学地质工程与测绘学院,陕西西安710054

出　　处：《海洋测绘》2009年第3期31-33,37,共4页Hydrographic Surveying and Charting

摘　　要：空间数据拟合是以搜集空间信息、建立合理的数据解算模型,用以表达空间数据之间的关系。支持向量机应用于空间数据拟合时,建立在结构风险最小的原则上,使得其模型有较强的泛化性能。借鉴泰勒公式的思想,在拟合空间起伏较大数据时,提出了对支持向量残差再拟合,以提高数据拟合的精度。最后,通过实例验证了残差再拟合算法的有效性。Space data fitting is used to illustrate the relationship of them by the collection of spatial information and the construction of rational data processing model. The usage of support vector machine in space data fitting is based on the minimal risk. Taking the advantage of Taylor formula, the second fitting of the SVM residual error is proposed in order to improve the fitting accuracy when fitting the undulating space data. The validity of the thought of the paper is demonstrated by an exampe.

关 键 词：空间数据库 支持向量机 优化 残差再拟合 

分 类 号：P282[天文地球—地图制图学与地理信息工程]