Artificial bee colony algorithm with adaptive covariance matrix for hearing loss detection

Abstract Artificial bee colony algorithm (ABC) is an efficient and popular evolutionary algorithm (EAs), which has been attracted wide attention by researchers, and improved ABC with various characteristics (ABCs) have been proposed. It is widely acknowledged that the search operator is the core element in the performance of ABC. However, the generally designed search operators of ABCs are rotation-variable processes and are dependent mainly on the natural coordinates and, as a result, the performance of those ABC is limited. In this paper, the mathematical characteristic of the search operator is deeply analyzed, and on this basis, ABC with adaptive covariance matrix (ACoM-ABC) is proposed, the adaptive covariance matrix (ACoM) is used to establish a proper coordinates by making use of the population distribution information, which can relieve the dependence of ABC on the coordinates to a certain extent and improve the exploitation capability. To balance the exploitation and exploration abilities of ABC, the search operators of ABC are implemented on eigen coordinates and natural coordinates Then, to estimate the performance of ACoM-ABC, which compares with six ABCs and other six EAs, and tests on CEC2014. The excellent experimental result shows that ACoM-ABC is an efficient and outstanding algorithm. Moreover, the proposed algorithm is applied to hearing loss detection, and the experiment result shows that the overall accuracy is 96.67%, which higher than the other five state-of-the-art approaches about 1%. Therefore, the ACoM-ABC has the practicability for realistic problems.