Linguistic Rule Extraction from Support Vector Machine Classifiers

Abstract

Classification decisions from linguistic rules are more desirable compared to complex mathematical formulas from support vector machine (SVM) classifiers due to the explicit explanation capability of linguistic rules. Linguistic rule extraction has been attracting much attention in explaining knowledge hidden in data. In this chapter, we show that the decisions from an SVM classifier can be decoded into linguistic rules based on the information provided by support vectors and decision function. Given a support vector of a certain class, cross points between each line, which is extended from the support vector along each axis, and an SVM decision hyper-curve are searched first. A hyper-rectangular rule is derived from these cross points. The hyper-rectangle is tuned by a tuning phase in order to exclude those out-class data points. Finally, redundant rules are merged to produce a compact rule set. Simultaneously, important attributes could be highlighted in the extracted rules. Rule extraction results from our proposed method could follow SVM classifier decisions very well. We compare the rule extraction results from SVM with RBF kernel function and linear kernel function. Experiment results show that rules extracted from SVM with RBF nonlinear kernel function are with better accuracy than rules extracted from SVM with linear kernel function. Comparisons between our method and other rule extraction methods are also carried out on several benchmark data sets. Higher rule accuracy is obtained in our method with fewer number of premises in each rule.