Modifying the Decision Function in One-Against-All Algorithm for Multi-Classification

Support vector machines (SVMs) are originally designed for binary classifications. As for multi-classifications, they are usually converted into binary ones, up to now, several methods have been proposed to decompose and reconstruct multi-class classification problems. In order to enhance the performance of one-against-all algorithm for multi-classification, in this paper, we modify the decision function of one-against-all approach. In order to examine the generalization performance of the proposed method, one-against-all and proposed approaches are applied to four UCI data sets. The results show that the training and testing accuracies of proposed method is higher than those of one-against-all. One-against-all performs just as well as one-against-one approaches