Classification of Microarray Data using Functional Link Neural Network

Abstract

Microarray dataset often contains a huge number of insignificant and irrelevant features that might lead to loss of useful information. The classes with both high relevance and having high significance feature sets are generally preferred for selecting the features, which determines the sample classification into their respective classes. This property has gained a lot of significance among the researchers and practitioners in DNA micro array classification. In this paper, Functional link neural network (FLNN) with four different basis functions named as Power series polynomial, Trigonometric, Chebyshev polynomial and Legendre polynomial functions have been considered to classify microarray data sets using t-test as a feature selection method. Further, this paper presents a comparative analysis on the obtained classification accuracy by coupling FLNN with different basis function and other existing models available in the literature. Performance parameters available in literature such as precision, recall, specificity, F-Measure, ROC curve and accuracy are applied in this comparative analysis to analyze the behavior of the classifiers. From the proposed approach, it is apparent that FLNN using Legendre polynomial is the suitable classification model among FLNN using different basis functions and other classifiers.