Cancer classification from microRNA (miRNA) gene expression data is a difficult task in system biology and machine learning as conventional classification methods require a sufficiently large number of labeled samples to train the classifiers accurately, particularly when the labeled samples are very expensive and difficult to collect. Therefore, conventional classification methods usually do not provide the desired classification accuracy due to the scarcity of training samples. In this context, we present an extreme learning machine (ELM) technique for cancer classification from miRNA gene expression data that can improve the classification accuracy as it is extremely fast and accurate compared to other traditional methods. The presented method is evaluated using publicly available miRNA gene expression datasets of breast cancer, pancreatic cancer, colorectal cancer, prostate cancer and lung cancer in terms of classification accuracy, precision, recall, macro F1-measure, micro F1-measure and kappa in comparison to four other state-of-the-art methods. Experimental results justify the dominance of the ELM method over the other compared methods for cancer sample classification from miRNA Gene Expression data.
