The Application of Support Vector Machine in Classifying Large Namber of Catalogs
WANG Jian fen, CAO Yuan da (Dept. of Computer Science and Engineering, Beijing Institute of Technology, Beijing100081, China)
Support vector machine is a highly performance classification method. The basic support vector machine (SVM) is for pair class problem. The principle of SVM was introduced and a classifier based on SVM for a large number catalogs was proposed. The method was named SVM decision tree. The types of SVM decision tree selected in various cases were discussed in detail, and the features of SVM decision tree were also analysed. It was proved that the mean error of classifying N catalogs with SVM decision tree is smaller.