K-RSVCR: A rough margin-based multi-class support vector machine

By introducing Rough set theory into K-SVCR, a rough margin-based K-SVCR (K-RSVCR) is proposed to deal with the over-fitting problem due to noise data and outlier. By maximizing rough margin in K-RSVCR but not margin in K-SVCR, more points are adaptively considered rather than the few extreme value points used in K-SVCR. In addition, different support vectors in different position are proposed to have different effect on separating hyperplane, namely, support vectors in the lower margin have more effects than those in the boundary of the rough margin. Therefore, K-RSVCR is not sensitive to noises or outliers, moreover, it will produce greater generalization performance. Numerical experiment results show the feasibility and validity of the proposed algorithm. ICIC International © 2010 ISSN 1881-803X.