In order to improve classification capability, an advanced support vector machine (SVM) algorithm (L2-SVM) was studied in this chapter. Theoretical analysis of L1-SVM and L2-SVM illuminated the solving process of classification problem by using these two algorithms. The analysis also indicated that middle-scale datasets can be correctly classified by SVM algorithm. Both algorithms can be realized with the help of MATLAB. Simulation was conducted by using two spirals datasets. The results showed that both L1-SVM and L2-SVM had very good classification capability. The identification rates of these two algorithms were better than those of neural network algorithm and nearest neighbor algorithm. In some conditions, L1-SVM and L2-SVM were consistent. Compared with L1-SVM, L2-SVM had larger model parameter adaptable space. L2-SVM was proved to be an excellent algorithm for optimal parameter searching and classification both theoretically and experimentally, which will play an important role in recognition field. © 2014 Springer Science+Business Media New York.
CITATION STYLE
Dai, W., Li, H., & Liu, Q. (2014). Application research of support vector machine classification Algorithm. In Lecture Notes in Electrical Engineering (Vol. 238 LNEE, pp. 2103–2110). Springer Verlag. https://doi.org/10.1007/978-1-4614-4981-2_230
Mendeley helps you to discover research relevant for your work.