Analysis of classification margin for classification accuracy with applications

Classification margin is commonly used for describing the classification capability of a committee of classifiers. In this paper, we study the relation between classification margin and misclassification error, focusing on exploring useful information about misclassification error from the known classification margin. We propose a max–min type bound concerning the minimal misclassification rate, and present the classification margins, and devise an algorithm for improving average classification accuracy based on the proposed bound. Experimental results show the effectiveness of the proposed algorithm and also validate our analytic results. & 2008 Elsevier B.V. All rights reserved.