2014/01/18 by Rania Ibrahim, Noha A. Yousri, Ibrahim, Rania +7
Biochemistry, Genetics and Molecular Biology · #Cancer-related molecular mechanisms research #Computational Engineering #FOS: Computer and information sciences #Finance #Machine Learning (cs.LG) #Machine Learning in Bioinformatics #MicroRNA in disease regulation #and Science (cs.CE)
paper · pdf · doi:10.48550/arxiv.1401.4589
openalex publication_date 2014/01/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
miRNA and gene expression profiles have been proved useful for classifying\ncancer samples. Efficient classifiers have been recently sought and developed.\nA number of attempts to classify cancer samples using miRNA/gene expression\nprofiles are known in literature. However, the use of semi-supervised learning\nmodels have been used recently in bioinformatics, to exploit the huge corpuses\nof publicly available sets. Using both labeled and unlabeled sets to train\nsample classifiers, have not been previously considered when gene and miRNA\nexpression sets are used. Moreover, there is a motivation to integrate both\nmiRNA and gene expression for a semi-supervised cancer classification as that\nprovides more information on the characteristics of cancer samples. In this\npaper, two semi-supervised machine learning approaches, namely self-learning\nand co-training, are adapted to enhance the quality of cancer sample\nclassification. These approaches exploit the huge public corpuses to enrich the\ntraining data. In self-learning, miRNA and gene based classifiers are enhanced\nindependently. While in co-training, both miRNA and gene expression profiles\nare used simultaneously to provide different views of cancer samples. To our\nknowledge, it is the first attempt to apply these learning approaches to cancer\nclassification. The approaches were evaluated using breast cancer,\nhepatocellular carcinoma (HCC) and lung cancer expression sets. Results show up\nto 20% improvement in F1-measure over Random Forests and SVM classifiers.\nCo-Training also outperforms Low Density Separation (LDS) approach by around\n25% improvement in F1-measure in breast cancer.\n