A multi-SVM classification system

Abstract

It has been shown by several researchers that multi-classifier systems can result in effective solutions to dificult tasks. In this work, we propose a multi-classifier system based on both supervised and unsupervised learning. According to the principle of \divide-and- conquer", the input space is partitioned into overlapping subspaces and Support Vector Machines (SVMs) are subsequently used to solve the respective classification subtasks. Finally, the decisions of the individual SVMs are appropriately combined to obtain the final classification decision. We used the Fuzzy c-means (FCM) method for input space partitioning and we considered a scheme for combining the decisions of the SVMs based on a probabilistic interpretation. Compared to single SVMs, the multi-SVM classification system exhibits promising accuracy performance on well-known data sets.