A New Formulation for Classification by Ellipsoids
We propose a new formulation for the optimal separation problems. This robust formulation is based on finding the minimum volume ellipsoid covering the points belong to the class. Idea is to separate by ellipsoids in the input space without mapping data to a high dimensional feature space unlike Support Vector Machines. Thus the distance order in the input space is preserved. Hopfield Neural Network is described for solving the optimization problem. The benchmark Iris data is given to evaluate the formulation.
KeywordsSupport Vector Machine Feature Space Input Space Polynomial Kernel Radial Basis Function Kernel
Unable to display preview. Download preview PDF.
- 3.Zhang, B.: Is the Maximal Margin Hyperplane Special in a Feature Space? Hewlett-Packard Research Laboratories Palo Alto (2001)Google Scholar
- 5.Vandenberghe, L., Boyd, S.: Applications of Semidefinite Programming. Technical Report of California University (1998)Google Scholar
- 6.Glineur, F.: Pattern Separation Via Ellipsoids and Conic Programming, Mémoire de D.E.A., Faculté Polytechnique de Mons, Mons, Belgium (September 1998)Google Scholar
- 8.Doğan, H.: Gradient Networks Design for Clustering in Novel Optimization Frameworks. Dokuz Eylül University, PhD. Thesis (December 2004)Google Scholar
- 9.Kruss, M.: Nonlinear Multivariate Analysis with Geodesic Kernels. Berlin Technical University, Thesis (February 2002)Google Scholar
- 10.Zhang, Z.: Learning Metrics Via Discriminant Kernels and Multidimensional Scaling: To-ward Expected Euclidean Representation. In: ICML 2003, pp. 872–879 (2003)Google Scholar
- 13.Vincent, W.: SVM and Co-Training. Honc Konc Bapist University, Technical Report (2002)Google Scholar