A Chaotic Levy Flights Bat Algorithm for Diagnosing Diabetes Mellitus

Abstract

algorithm is a meta-heuristic algorithm that is based on the echolocation behavior of bats. The searching behavior of the algorithm depends on generating uniformly distributed random walks in the search space. Hence, it may suffer from being tapped in local optima. In this paper, a classification using Bat inspired algorithm with chaotic levy flight variable is proposed. The chaotic variable has set of characteristics that enable it to enrich the searching behavior and prevent the Bat algorithm from being trapped into local optimum. The chaotic sequence and a chaotic Levy flight are incorporated with Bat algorithm for many purposes including, efficiently generating new solutions via randomization, increase the diversity of the solutions, avoid trapping in a local optimum and increase the chances of finding global optimum solution. The proposed algorithm aims to help physicians in early diagnosis and treatment of Diabetes Mellitus (DM). DM is a major health problem in both industrial and developing countries and its incidence is rising. The proposed algorithm is applied on Pima Indians Diabetes data set from UCI repository of machine learning data bases. The experimental results prove the superiority of the proposed algorithm over the traditional Bat algorithm as well as different classifiers which were implemented on the same data set and within the same environment. KeywordsInspired Algorithm (BIA), Levy Flight, Chaotic