Feature selection applied to classification of cancerous tissue
Joel Pitt
Department of Computer Science
University of Canterbury
Abstract
This report examines the use of Support Vector Machines and Genetic Programming Classifers (GPCs) to distinguish between classes of cancer based on gene expression data. The effect of feature selection on classifer accuracy and on the convergence time of GPCs is experimentally investigated, with the goal of making classifcation problems on gene expression data tractable to GPCs.