A fast support vector machine classification algorithm based on Karush-Kuhn-Tucker conditions

Abstract

Although SVM have shown potential and promising performance in classification, they have been limited by speed particularly when the training data set is large. In this paper, we propose an algorithm called the fast SVM classification algorithm based on Karush-Kuhn-Tucker (KKT) conditions. In this algorithm, we remove points that are independent of support vectors firstly in the training process, and then decompose the remaining points into blocks to accelerate the next training. From the theoretical analysis, this algorithm can remarkably reduce the computation complexity and accelerate SVM training. And experiments on both artificial and real datasets demonstrate the efficiency of this algorithm. © 2009 Springer-Verlag Berlin Heidelberg.