Kernel Based Regularized Multiple Criteria Linear Programming Model

Authors:
Yuehua Zhang;Peng Zhang;Yong Shi
Affiliations:
CAS Research Center on Fictitious Economy and Data Sciences, Beijing, China 100080;CAS Research Center on Fictitious Economy and Data Sciences, Beijing, China 100080;CAS Research Center on Fictitious Economy and Data Sciences, Beijing, China 100080 and College of Information Science & Technology, University of Nebraska at Omaha, Omaha, USA NE 68182
Venue:
ICCS 2009 Proceedings of the 9th International Conference on Computational Science
Year:
2009

Citing 3
Cited 0

Knowledge discovery in databases: an overview

AI Magazine
A Regularized Multiple Criteria Linear Program for Classification

ICDMW '07 Proceedings of the Seventh IEEE International Conference on Data Mining Workshops
Optimization Techniques for Semi-Supervised Support Vector Machines

The Journal of Machine Learning Research

Quantified Score

Hi-index	0.00

Visualization

Abstract

Although Regularized Multiple Criteria Linear Programming (RMCLP) model has shown its effectiveness in classification problems, its inherent drawback of linear formulation limits itself into only solving linear classification problems. To extend RMCLP into solving non-linear problems, in this paper, we propose a kernel based RMCLP model by using a form $ w = \sum\limits^{N}_{i=1}\beta_{i}\phi(x_{i})$ to replace the original weight w in RMCLP model. Empirical studies on synthetic and real-life datasets demonstrate that our new model is capable to classify non-linear datasets. Moreover, comparisons to SVM and MCQP also exhibit the fact that our new model is superior to other non-linear models in classification problems.