An Optimized Multi-class Classification Algorithm Based on SVM Decision Tree

Abstract

An optimized multi-class classification algorithm based on SVM decision tree (SVMDT) is proposed. But by SVMDT, the generalization ability depends on the tree structure. In this paper, the relativity separability measure between classes is defined based on the distribution of the training samples to improve the generalization ability of SVMDT. SVM is extended to non-linear SVM by using kernel functions and the classification experiments prove the algorithm is more effective and feasible for classification accuracy.