LS Bound based gene selection for DNA microarray data

• DOI: 10.1093/bioinformatics/bti216
• Publication Date: 2004-12-15T02:58:40Z
• Authors: X. Zhou; K. Z. Mao
• Keywords: Microarray databases; Gene selection

Motivation: One problem with discriminant analysis of DNA microarray data is that each sample represented by quite a large number genes, and many them are irrelevant, insignificant or redundant to the at hand. Methods for selecting important genes are, therefore, much significance in analysis. In present study, new criterion, called LS Bound measure, proposed address gene selection problem. The measure derived from leave-one-out procedure LS-SVMs (least squares support vector machines), as upper bound classification results it reflects some extent generalization performance subsets.