Feature Selection Method Based on Chaotic Maps and Butterfly Optimization Algorithm

Abstract

Feature selection (FS) is a challenging problem that attracted the attention of many researchers. FS can be considered as an NP hard problem, If dataset contains N features then solutions are generated with each additional feature, the complexity doubles. To solve this problem, we reduce the dimensionality of the feature by extracting the most important features. In this paper we integrate the chaotic maps in the standard butterfly optimization algorithm to increase the diversity and avoid trapping in local minima in this algorithm. The proposed algorithm is called Chaotic Butterfly Optimization Algorithm (CBOA).The performance of the proposed CBOA is investigated by applying it on 16 benchmark datasets and comparing it against six meta-heuristics algorithms. The results show that invoking the chaotic maps in the standard BOA can improve its performance with accuracy more than.