Wrapper-based gene selection with Markov blanket

Gene selection seeks to find a small subset of discriminant genes from the gene expression profiles. Current gene selection methods such as wrapper-based models mainly address the issue of obtaining high-quality gene subsets. However, they are considerably time consuming, due to the existence of irrelevant and redundant genes. In this study, we present an improved wrapper-based gene selection method by introducing the Markov blanket technique to reduce the required wrapper evaluation time. In addition, our method can identify targeting genes while eliminating redundant ones in an efficient way. We use ten publicly available microarray datasets to evaluate the proposed method. The results show that our method can handle gene selection effectively. Our experimental results also show that wrapper-based method combined with the Markov blanket outperforms other competing methods in terms of classification accuracy and time/space complexity.