Study of Double SMO Algorithm Based on Attributes Reduction

Authors:
Chen Chen;Liu Hong;Haigang Song;Xueguang Chen;Tiemin Hou
Affiliations:
Institute of System Engineering, Huazhong University of Science and Technology, Wuhan, P.R. China 430074;Institute of System Engineering, Huazhong University of Science and Technology, Wuhan, P.R. China 430074;Basic Research Service of the Ministry of Science and Technology of the P. R. China, Beijing, P.R. China 100862;Institute of System Engineering, Huazhong University of Science and Technology, Wuhan, P.R. China 430074;Key Lab. for Image Processing and Intelligent control, Huazhong University of Science and Technology, Wuhan, P.R. China 430074
Venue:
ISNN 2009 Proceedings of the 6th International Symposium on Neural Networks: Advances in Neural Networks - Part II
Year:
2009

Citing 5
Cited 0

The nature of statistical learning theory

The nature of statistical learning theory
Induction of Decision Trees

Machine Learning
Training Support Vector Machines: an Application to Face Detection

CVPR '97 Proceedings of the 1997 Conference on Computer Vision and Pattern Recognition (CVPR '97)
Successive overrelaxation for support vector machines

IEEE Transactions on Neural Networks
Back-propagation learning in expert networks

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.01

Visualization

Abstract

To solve the classification problem in data mining, this paper proposes double SMO algorithm based on attributes reduction. Firstly attributes reduction deletes irrelevant attributes (or dimensions) to reduce data amount, consequently the total calculation is reduced, the training speed is fastened and Classification mode is easy to understand. Secondly applying SMO algorithm on the sampling dataset to get the approximate separating hyperplane, and then we obtain all the support vectors of original dataset. Finally again use SMO algorithm on the support vectors to get the final separating hyperplane. It is shown in the experiments that the algorithm reduces the memory space, effectively avoids the noise points' effect on the final separating hyperplane and the precision of the algorithm is better than Decision Tree, Bayesian and Neural Network.