Using Wolfe's Method in Support Vector Machines Learning Stage

Authors:
Juan Frausto-Solís;Miguel González-Mendoza;Roberto López-Díaz
Affiliations:
Instituto Tecnológico de Estudios Superiores de Monterrey,;Instituto Tecnológico de Estudios Superiores de Monterrey,;Instituto Tecnológico de Estudios Superiores de Monterrey,
Venue:
MICAI '09 Proceedings of the 8th Mexican International Conference on Artificial Intelligence
Year:
2009

Citing 6
Cited 0

Practical methods of optimization; (2nd ed.)

Practical methods of optimization; (2nd ed.)
Improving support vector machine classifiers by modifying kernal functions

Neural Networks
An introduction to support Vector Machines: and other kernel-based learning methods

An introduction to support Vector Machines: and other kernel-based learning methods
A Tutorial on Support Vector Machines for Pattern Recognition

Data Mining and Knowledge Discovery
SMO algorithm for least-squares SVM formulations

Neural Computation
Training linear SVMs in linear time

Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining

Quantified Score

Hi-index	0.01

Visualization

Abstract

In this paper, the application of Wolfe's method in Support Vector Machines learning stage is presented. This stage is usually performed by solving a quadratic programming problem and a common approach for solving it, is breaking down that problem in smaller subproblems easier to solve and manage. In this manner, instead of dividing the problem, the application of Wolfe's method is proposed. The method transforms a quadratic programming problem into an Equivalent Linear Model and uses a variation of simplex method employed in linear programming. The proposed approach is compared against QuadProg Matlab function used to solve quadratic programming problems. Experimental results show that the proposed approach has better quality of classification compared with that function.