Saliency Analysis of Support Vector Machines for Gene Selection in Tissue Classification

Abstract

This paper deals with the application of saliency analysis to Support Vector Machines (SVMs) for gene selection in tissue classification. The importance of genes is ranked by evaluating the sensitivity of the output to the inputs in terms of the partial derivative. A systematic learning algorithm called the Recursive Saliency Analysis (RSA) algorithm is developed to remove irrelevant genes. One simulated data and two gene expression data sets for tissue classification are evaluated in the experiment. The simulation results demonstrate that RSA is effective in SVMs for identifying important genes.