Local Capability Analysis and Comparative Study of Kernel Functions in Support Vector Machine

Local Capability Analysis and Comparative Study of Kernel Functions in Support Vector Machine

Hailun Wang Daxing Xu

Logistics Engineering College, Shanghai Maritime University, Shanghai 200000, China

College of Electrical and Information Engineering, Quzhou University, Quzhou 324000, China

Corresponding Author Email: 
xiaohong1920@126.com, daxingxu@163.com
18 April 2017
18 May 2017
30 June 2017
| Citation



Kernel function, the centrepiece of Support Vector Machine (SVM), is classified into local kernel function and global kernel function. The features of the local and global kernel functions can be demonstrated all at once in a combined kernel function. This paper analyses the local capability of SVM kernel function through comparative analysis. Specifically, the local capability of combined kernel function was defined and analysed for the first time; the local capability features of typical kernel functions and combined kernel function were detailed and compared with each other. Finally, the correctness and rationality of the analysis was verified through simulation.


Local capability, Combined kernel function, Local kernel function, Global kernel function, Support Vector Machine (SVM)

1. Introduction
2. Problem Description
3. Definition of Local Capability and Feature Analysis of Kernel Function
4. Comparative Analysis of Local Capabilities of Kernel Functions
5. Experimental Simulation
6. Conclusion

[1] K. Sopyła, P. Drozda, Stochastic gradient descent with Barzilai–Borwein update step for SVM, 2015, Information Sciences, vol. 31, no. 6, pp. 218-233.

[2] C. Cortesc, V.N. Vapnik, Support vector networks, 1995, Machine Learning, vol. 20, no. 1, pp. 273 -297.

[3] G. Baudat, F. Anouar, Generalized discriminant analysis using a kernel approach, 2010, Neural computation, vol. 12, no. 10, pp. 2385-2404.

[4] M. Szummer, T.S. Jaakkola, Kernel expansions with unlabeled examples, 2011, Advances in Neural Information Processing Systems, pp. 626-632.

[5] P. Rujan, M. Marchand, Computing the Bayes kernel classifier, 1999.

[6] C.W. Hsu, C.C. Chang, C.J. Lin, A practical guide to support vector classification, 2013, Department of information engineering, National Taiwan University, pp. 1-16.

[7] E.B. Huerta, B. Duval, J.K. Hao, A hybrid GA/SVM approach for gene selection and classification of microarray data, 2006, Applications of Evolutionary Computing, vol. 3907, no. 1, pp. 34-44.

[8] J. Zhou, T. Bai, C. Suo, The SVM optimized by culture genetic algorithm and its application in forecasting share price, 2008, IEEE International Conference on Granular Computing, Art. No. 4664698, pp. 838–843

[9] J. Zhou, T. Bai, A. Zhang, J. Tian, The integrated methodology of wavelet transform and GA based-SVM for forecasting share price, 2016, IEEE International Conference on Granular Computing, Art. No. 4608094, pp. 729–733

[10] F. Kuang, S. Zhang, Z. Jin, W. Xu, A novel SVM by combining kernel principal component analysis and improved chaotic particle swarm optimization for intrusion detection, 2015, Soft Computing, vol. 19, no. 5, pp. 1-13.

[11] Z. Xu, Z.Y. Dong, W.Q. Liu, Short-term electricity price forecasting using wavelet and SVM techniques, 2013, Dynamics of Continuous Discrete & Impulsive Systems, vol. 16, no. 1, pp. 372-377.

[12] S.K. Aggarwal, L.M. Saini, A. Kumar, Electricity price forecasting in deregulated markets: a review and evaluation, International Journal of Electrical Power & Energy Systems, vol. 31, no. 1, pp. 13–22

[13] T. Mu, A.K. Nandi, Automatic tuning of L2-SVM parameters employing the extended Kalman filter, 2009, Expert Systems, vol. 26, no. 2, pp. 160- 175.

[14] H.L. Wang, D.X. Xu, Analysis of Local Capability Characteristics of Kernel Function, 2017, Journal of Residuals Science and Technology, vol. 13, no. 7, pp. 110-116.

[15] X. Huang, The study on Kernel in Support Vector Machine, 2008, Soochow University, China.

[16] H. Yi, X. Song, B. Jiang, Y. Liu, Z. Zhou, Flexible support vector regression and its application to fault detection, 2013, Acta Automatica Sinica, vol. 39, no. 3, pp. 272-284

[17] O. Chapelle, V. Vapnik, O. Bousquet, S. Mukherjee, Choosing multiple parameters for support vector machines, 2002, Machine Learning, vol. 46, no. 1, pp. 131-159.

[18] T.H. Loutas, D. Roulias, G. Georgoulas, Remaining useful life estimation in rolling bearings utilizing data-driven probabilistic e-support vectors regression, 2013, IEEE Transactions on Reliability, vol. 62, no. 4, pp. 821-832.

[19] J.P. Jacobs, Bayesian support vector regression with automatic relevance determination kernel for modeling of antenna input characteristics, 2012, IEEE Transactions on Antennas and Propagation, vol. 60, no. 4, pp. 2114-2118.

[20] N. Wang, H.Z. Ai, Combining local and global information for hair shape modeling, 2014, Acta Automatica Sinica, vol. 40, no. 4, pp. 615-623.