A Novel Hybrid JFA-SVM Algorithm for Feature Selection

Abstract

In this study, we have proposed a hybrid model with a combination of both jumper firefly algorithm (JFA) and multi-filter approach. We have considered three variants of filter methods such as IG, RefiefF, and Chi-square. Then, best N subset of gene from the four filter approaches are inputted to a novel jumper firefly algorithm (JFA) to identify the optimal feature subsets. Later, meta-search model is used with a well-known classifier SVM to calculate the accuracy of the biomarker feature subset. JFA was used in this study with SVM to achieve optimal convergence time and a stochastic approach with the improvement of convergence time. The performance of the proposed one is compared with FA-SVM, FA-NB, FA-DT, JFA-SVM, JFA-NB, and JFA-DT. From the result analysis, it is clear that accuracy achieved by the proposed model is better as compared to its counter parts and results high-quality solutions.