A Multi-class SVM Classifier Utilizing Binary Decision Tree
Gjorgji Madzarov, Dejan Gjorgjevikj and Ivan Chorbev
Abstract: In this paper a novel architecture of Support Vector Machine classifiers utilizing binary decision tree (SVM-BDT) for solving multiclass problems is presented. The hierarchy of binary decision subtasks using SVMs is designed with a clustering algorithm. For consistency between the clustering model and SVM, the clustering model utilizes distance measures at the kernel space, rather than at the input space. The proposed SVM based Binary Decision Tree architecture takes advantage of both the efficient computation of the decision tree architecture and the high classification accuracy of SVMs. The SVMBDT architecture was designed to provide superior multi-class classification performance. Its performance was measured on samples from MNIST, Pendigit, Optdigit and Statlog databases of handwritten digits and letters. The results of the experiments indicate that while maintaining comparable or offering better accuracy with other SVM based approaches, ensembles of trees (Bagging and Random Forest) and neural network, the training phase of SVM-BDT is faster. During recognition phase, due to its logarithmic complexity, SVM-BDT is much faster than the widely used multi-class SVM methods like “one-against-one” and “one-against-all”, for multiclass problems. Furthermore, the experiments showed that the proposed method becomes more favourable as the number of classes in the recognition problem increases.