Performance Evaluation of Kernels in Multiclass Support Vector Machines
R. Sangeetha1, B. Kalpana2
1Sangeetha R., Department of Computer Science, Avinashilingam Institute for Home Science and Higher Education for Women, Coimbatore, India.
2Dr. B. Kalpan, Department of Computer Science, Avinashilingam Institute for Home Science and Higher Education for Women, Coimbatore, India.
Manuscript received on October 04, 2011. | Revised Manuscript received on October 21, 2011. | Manuscript published on November 05, 2011. | PP: 138-145 | Volume-1 Issue-5, November 2011. | Retrieval Number: E0171091511/2011©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In recent years, Kernel based learning algorithm has been receiving increasing attention in the research domain. Kernel based learning algorithms are related internally with the kernel functions as a key factor. Support Vector Machines are gaining popularity because of their promising performance in classification and prediction. The success of SVM lies in suitable kernel design and selection of its parameters. SVM is theoretically well-defined and exhibits good generalization result for many real world problems. SVM is extended from binary classification to multiclass classification since many real-life datasets involve multiclass data. In this paper, we propose an optimal kernel for one-versus-one (OAO) and one-versus-all (OAA) multiclass support vector machines. The performance of the OAO and OAA are evaluated using the metrics like accuracy, support vectors, support vector percentage, classification error, and speed. The empirical results demonstrate the ability to use more generalized kernel functions and it goes to prove that the polynomial kernel’s performance is consistently better than other kernels in SVM for these datasets.
Keywords: Support Vector Machine, Multiclass Classification, Kernel function, One versus One, One versus All.