SVM-based tumor classification with gene expression data

Gene expression data that are gathered from tissue samples are expected to significantly help the development of efficient tumor diagnosis and classification platforms. Since DNA microarray experiments provide us with huge amount of gene expression data and only a few of genes are related to tumor, gene selection algorithms should be emphatically explored to extract those informative genes related tumor from gene expression data. So we propose a novel feature selection approach to further improve the SVM-based classification performance of gene expression data, which projects high dimensional data onto lower dimensional feature space. We examine a set of gene expression data that include sets of tumor and normal clinical samples by means of SVMs classifier. Experiments show that SVM has a superior performance in classification of gene expression data as long as the selected features can represent the principal components of all gene expression samples.