TY - GEN
T1 - Performance analysis of localization strategy for island model genetic algorithm
AU - Gozali, Alfian Akbar
AU - Fujimura, Shigeru
N1 - Funding Information:
This work was supported by Indonesia Endowment Fund for Education (LPDP), a scholarship from Ministry of Finance, Republic of Indonesia. This work was conducted while at Graduate School of Information, Production, and Systems, Waseda University, Japan.
Publisher Copyright:
© 2017 IEEE.
PY - 2017/8/29
Y1 - 2017/8/29
N2 - Genetic algorithm (GA) is one of the standard solutions to solve many optimization problems. One of a GA type used for solving a case is island model GA (IMGA). Localization strategy is a brand-new feature for IMGA to better preserves its diversity. In the previous research, localization strategy could carry out 3SAT problem almost perfectly. In this study, the proposed feature is aimed to solve real parameter single objective computationally expensive optimization problems. Differ with an issue in previous research which has a prior knowledge and binary, the computationally expensive optimization has not any prior knowledge and floating type problem. Therefore, the localization strategy and its GA cores must adapt. The primary goal of this research is to analyze further the localization strategy for IMGA's performance. The experiments show that the new feature is successfully modified to meet the new requirement. Localization strategy for IMGA can solve all computationally expensive functions consistently. Moreover, this new feature could make IMGA reaches leading ratio 0.47 among other current solvers.
AB - Genetic algorithm (GA) is one of the standard solutions to solve many optimization problems. One of a GA type used for solving a case is island model GA (IMGA). Localization strategy is a brand-new feature for IMGA to better preserves its diversity. In the previous research, localization strategy could carry out 3SAT problem almost perfectly. In this study, the proposed feature is aimed to solve real parameter single objective computationally expensive optimization problems. Differ with an issue in previous research which has a prior knowledge and binary, the computationally expensive optimization has not any prior knowledge and floating type problem. Therefore, the localization strategy and its GA cores must adapt. The primary goal of this research is to analyze further the localization strategy for IMGA's performance. The experiments show that the new feature is successfully modified to meet the new requirement. Localization strategy for IMGA can solve all computationally expensive functions consistently. Moreover, this new feature could make IMGA reaches leading ratio 0.47 among other current solvers.
KW - Computationally expensive optimization
KW - Genetic algorithms
KW - Island model genetic algorithm
KW - Localization strategy
UR - http://www.scopus.com/inward/record.url?scp=85030843347&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85030843347&partnerID=8YFLogxK
U2 - 10.1109/SNPD.2017.8022741
DO - 10.1109/SNPD.2017.8022741
M3 - Conference contribution
AN - SCOPUS:85030843347
T3 - Proceedings - 18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017
SP - 327
EP - 332
BT - Proceedings - 18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017
A2 - Hirata, Hiroaki
A2 - Hiroki, Nomiya
A2 - Hochin, Teruhisa
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 18th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing, SNPD 2017
Y2 - 26 June 2017 through 28 June 2017
ER -