The Application of SVDD in Gene Expression Data Clustering

Abstract

Support Vector Domain Description (SVDD) is a kind of classification method based on Support Vector Machine. This paper discussed its application in gene expression data clustering. The training samples are mapped into a high dimension feature space through kernel function, at the same time, the non- objective samples are introduced in training to increase the refusing ability. After obtaining the support vectors to form the initial boundary by setting the kernel's parameter, the boundary energy function is constructed, so the real classified boundary can be approximated through finding the minimum energy function. The experimental results in Yeast Cell gene expression data show it could obtain a tighter hyper sphere and better clustering.