Evaluating Ensemble Learning Impact on Gene Selection for Automated Cancer Diagnosis

Abstract

Modern artificial intelligence (AI) research shows that cancers are detectable and diagnosable by classification of DNA micro-arrays in molecular level. DNA micro-arrays data has the special property of high-dimension with redundancy, which may include thousands of features. In this study, a novel hybrid feature selection framework is proposed based on ensemble learning techniques to select the most important genes. Experimental results show that the proposed method effectively improves the classification accuracy compared to conventional methods.