Tumor classification via a simultaneous structure sparse representation model

Abstract

Cancer diagnosis is an important clinical application of gene expression microarray technology. This paper proposes a new model for tumor classification. The key idea is to implement the similar joint decomposition approach in the context of sparse coding with subsequences of gene expression data (SGED). Based on this idea, we formulate a simultaneous structure sparse model for tumor classification. Finally, experimental results in five tumor gene expression datasets show that the proposed method outperforms sparse representation method.