Semi-supervised Ensemble Learning for Efficient Cancer Sample Classification from miRNA Gene Expression Data

Abstract

Traditional classifiers often fail to produce desired classification accuracy because of inadequate training samples present in microRNA (miRNA) gene expression cancer datasets. In this context, we propose a novel semi-supervised ensemble learning (SSEL) strategy combining the (advantages of) semi-supervised learning and ensemble learning which is able to produce better results than the individual constituent classifiers. The proposed method is validated using eight publicly available miRNA gene expression datasets of pancreatic and colorectal cancers with respect to classification accuracy, precision, recall, macro $$F_{1}$$ -measure and kappa in comparison to six other state-of-the-art methods. The experimental results reveal that the proposed SSEL method significantly dominates other compared methods for cancer sample classification. The results of the statistical significance tests, receiver operating characteristic curve and area under curve justify the relevance of the better results in favor of the proposed method.