SVM classification for imbalanced data using conformal kernel transformation
Yong Zhang, Panpan Fu, Wenzhe Liu, Li Zou · 2014
The problem of classifying imbalanced datasets has drawn a significant amount of interest from academia and industry. In this paper, we propose a modified support vector machine (SVM) approach using conformal kernel transformation to address the class imbalance problem. The proposed method first uses standard SVM algorithm to obtain an approximate hyperplane. And then, we give a kernel function and compute its parameters using the chi-square test. Finally, an experimental analysis is carried out with a wide range of highly imbalanced datasets over the proposal and several other methods. The results show that our proposal outperforms previously proposed methods.