A Novel decision tree induction heuristic

Abstract

One technique for obtaining categorization information from a set of feature-based examples is decision tree induction. The lowest entropy is the most commonly utilized heuristic information in decision tree construction. This heuristic knowledge has a significant disadvantage: it is incapable of generalization. Support vector machine (SVM) is a machine learning classification approach based on statistical learning theory. It has a high degree of generality. Given the link between the classification margin of a support vector machine (SVM) and its generalization capabilities, a big margin of SVM may be utilized as decision tree heuristic information.