 Article highlights the identification of gene features in the microarray dataset is a challenging task. The research tries to propose a multi-stage algorithm for biomedical deep feature selection. Two steps were involved in classification, combination of three feature selection techniques and unsupervised neural network for creating feature representation. On various datasets, the accuracy found to be either similar or improved. This article was authored by Ismail Keshta, Pallavi Sagar Deshpande, Muhammad Shabazz, and others.