TY - JOUR
T1 - Island artificial bee colony for global optimization
AU - Awadallah, Mohammed A.
AU - Al-Betar, Mohammed Azmi
AU - Bolaji, Asaju La’aro
AU - Doush, Iyad Abu
AU - Hammouri, Abdelaziz I.
AU - Mafarja, Majdi
N1 - Awadallah, M. A., Al-Betar, M. A., Bolaji, A. L., Doush, I. A., Hammouri, A. I., & Mafarja, M. (2020). Island artificial bee colony for global optimization. Soft Computing, 24(17), 13461–13487. https://doi.org/10.1007/s00500-020-04760-8
PY - 2020/9/1
Y1 - 2020/9/1
N2 - This paper proposes an efficient version of artificial bee colony (ABC) algorithm based on the island model concepts. The new version is called the island artificial bee colony (iABC) algorithm. It uses the structured population concept by applying the island model to improve the diversification capabilities of ABC. In the island model, the population is divided into a set of sub-populations called islands, each of which is manipulated separately by an independent variant of the ABC. After a predefined number of iterations, the islands exchange their solutions by migration. This process can help ABC in controlling the diversity of the population during the search process and thus improve the performance. The proposed iABC is evaluated using global optimization functions established by the IEEE-CEC 2015 which include 15 test functions with various dimensions and complexities (i.e., 10, 30, and 50). In order to evaluate the performance of iABC, various parameter settings are utilized to test the effectiveness of their convergence properties. Furthermore, the performance of iABC is compared against 19 comparative methods that used the same IEEE-CEC 2015 test functions. The results show that iABC produced better results when compared with ABC in all IEEE-CEC 2015 test functions, while the results of iABC better than those of the other island-based algorithm on almost all test functions. Furthermore, iABC is able to obtain three new results for three test functions better than all the comparative methods. Using Friedman test and Holm’s procedure, iABC is ranked third, seventh, and ninth out of 19 comparative methods for the test functions with 10, 30, 50 dimensionality, respectively.
AB - This paper proposes an efficient version of artificial bee colony (ABC) algorithm based on the island model concepts. The new version is called the island artificial bee colony (iABC) algorithm. It uses the structured population concept by applying the island model to improve the diversification capabilities of ABC. In the island model, the population is divided into a set of sub-populations called islands, each of which is manipulated separately by an independent variant of the ABC. After a predefined number of iterations, the islands exchange their solutions by migration. This process can help ABC in controlling the diversity of the population during the search process and thus improve the performance. The proposed iABC is evaluated using global optimization functions established by the IEEE-CEC 2015 which include 15 test functions with various dimensions and complexities (i.e., 10, 30, and 50). In order to evaluate the performance of iABC, various parameter settings are utilized to test the effectiveness of their convergence properties. Furthermore, the performance of iABC is compared against 19 comparative methods that used the same IEEE-CEC 2015 test functions. The results show that iABC produced better results when compared with ABC in all IEEE-CEC 2015 test functions, while the results of iABC better than those of the other island-based algorithm on almost all test functions. Furthermore, iABC is able to obtain three new results for three test functions better than all the comparative methods. Using Friedman test and Holm’s procedure, iABC is ranked third, seventh, and ninth out of 19 comparative methods for the test functions with 10, 30, 50 dimensionality, respectively.
KW - Artificial bee colony
KW - Island-based model
KW - Optimization
KW - Population diversity
KW - Structured population
UR - http://www.scopus.com/inward/record.url?scp=85079412215&partnerID=8YFLogxK
U2 - 10.1007/s00500-020-04760-8
DO - 10.1007/s00500-020-04760-8
M3 - Article
SN - 1432-7643
VL - 24
SP - 13461
EP - 13487
JO - Soft Computing
JF - Soft Computing
IS - 17
ER -