Support vector machine classification for large datasets using decision tree and Fisher linear discriminant

• The decision trees and Fisher linear discriminate are combined.
• Classification accuracy of SVM is improved.
• Low entropy regions are used for decision boundaries.
• Several benchmark problems are applied.

Training a support vector machine (SVM) with data number nn has time complexity between O(n2)O(n2) and O(n3)O(n3). Most training algorithms for SVM are not suitable for large datasets. Decision trees can simplify SVM training, however classification accuracy becomes lower when there are inseparable points. This paper introduces a novel method for SVM classification. A decision tree is used to detect the low entropy regions in input space, and Fisher’s linear discriminant is applied to detect the data near to support vectors. The experimental results demonstrate that our approach has good classification accuracy and low standard deviation, the training is significantly faster than the other existing methods.