A Determination Method for Separation Hyperplane

Abstract

Support vector machines(SVM) take very long time in training and support vector domain classifier(SVDC) has the disadvantage of low accuracy in binary classification problem.A separation hyperplane is constructed based on support vector domain description(SVDD),which attempts the combination of SVDD with SVM.Firstly,two upper bounds for two-class error ratio are made equal so as to determinate the two parameter values C1,C2.Then,through description of the training samples from each class respectively using SVDD,two-class hypersphere boundary vectors are obtained.Finally,the normal vector of the separation hyperplane is deduced by the difference of the two hypersphere centers based on the rule of the distance sum maximization of the two hypersphere centers to the separation hyperplane.The threshold is determined to construct a separation hyperplane by the information provided by the sample size and radii of the two hyperspheres.Real experimental data show that the error ratio of the proposed algorithm decreases greatly than that of SVDC and SVM,and the training time decreases largely.

Cite this paper

@article{xue2007determination,
  title = {{A Determination Method for Separation Hyperplane}},
  author = {Xue, Zhenxia},
  journal = {Journal of North University of China},
  year = {2007}
}

Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.

Related papers