Imbalanced Support Vector Machine Classification Based on Hyper-Sphere

Abstract:

In classification, when the distribution of the training data between classes is uneven, the learning algorithm is generally dominated by the feature of the majority classes. Features in the minority classes are normally difficult to be fully recognized. Hyper-sphere support vector machine is an important method for unbalanced classification which is an important issue, but this algorithm has a defect. In order to significantly improve the classification performance of imbalanced datasets, we propose a new method based on Generalized Hyper-sphere Support Vector Machine to enhance the classification accuracy for the minority classes. Support vector machine (SVM) is then used as the base classifier to train the reprocessed dataset. Our experimental results demonstrate that the proposed selection technique improves the classification rate of the rare events, and it also improves the overall accuracy of SVM without data pre-processing.

• Periodical: Applied Mechanics and Materials (Volume 339)
• Pages: 384-388
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.339.384
• Online since: July 2013
• Authors: Cun He Li, Rui Xue Chen, Yi Zhao Ouyang
• Keywords: Generalized Hyper-Sphere SVM, Imbalanced Classification, Margin, Support Vector Machine (SVM)

References

[1] M. Kubat, R.C. Holte and S. Matwin 30:195–215 (1998) Machine learning for the detection of oil spills in satellite radar images. Mach Learn

DOI: 10.1023/a:1007452223027

[2] P. Chan, S.J. Stolfo (1998) Toward scalable learning with non-uniform class and cost distributions: a case study in credit card fraud detection. In Proceedings of the fourth international conference on knowledge discovery and data mining AAAI Press:164–168

[3] G.M. Weiss, H. Hirsh (1998) Learning to predict rare events in event sequences. In Proceedings of the fourth international conference on knowledge discovery and data mining AAAI Press:359–363

[4] V.N. Vapnik (1998) Statistical learning theory. Wiley New York

[5] L. Polkowski (2003) Rough Mereology: A Rough Set Paradigm for Unifying Rough Set Theory and Fuzzy Set Theory. Fundamenta Informaticae 54: 67–88

DOI: 10.1007/3-540-39205-x_9

[6] C.T. Su, Y.H. Hsiao, (2007) An Evaluation of the Robustness of MTS for Imbalanced Data. IEEE Trans Knowledge and Data Engineering 19:1321–1332

DOI: 10.1109/tkde.2007.190623

[7] X. Hong, S. Chen and C.J. Harris (2007) A Kernel-Based Two-Class Classifier for Imbalanced Data Set. IEEE Trans Neural Networks 18:28–41

DOI: 10.1109/tnn.2006.882812

[8] X.F. Zhang, Y.W. Liu (2008) Study of Generalized Hyper-Sphere Support Vector Machine. Journal of Computer Research and Development 45( 11):1807-1816

[9] Z.M. Miao, G.Y. Hu, L. Ding, L.W. Zhao and Z.S. Pan (2008) Support Vector Date Description Implemented in Class-Imbalance Learning 26(1):79-84