Authors:
Kittipong Boonlong
1
;
Nachol Chaiyaratana
2
and
Kuntinee Maneeratana
3
Affiliations:
1
Burapha University, Thailand
;
2
King Mongkut´s University of Technology North Bangok, Thailand
;
3
Chulalongkorn University, Thailand
Keyword(s):
Genetic algorithm, Multi-objective optimization, Objective compression.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Computational Intelligence
;
Evolutionary Computing
;
Evolutionary Multiobjective Optimization
;
Genetic Algorithms
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Soft Computing
Abstract:
This paper presents an improved version of compressed objective genetic algorithm to solve problems with a large number of objectives. The improved compressed objective genetic algorithm (COGA-II) employs a rank assignment for the screening of non-dominated solutions that best approximate the Pareto front from vast numbers of available non-dominated solutions. Since the winning non-dominated solutions are heuristically determined from the survival competition, the procedure is referred to as a winning-score based ranking mechanism. In COGA-II, an m-objective vector is transformed to only one criterion, the winning score of which assignment is improved from that of the previous version, COGA. COGA-II is subsequently benchmarked against a non-dominated sorting genetic algorithm II (NSGA-II) and an improved strength Pareto genetic algorithm (SPEA-II), in seven scalable DTLZ benchmark problems. The results reveal that for the closeness to the true Pareto front COGA-II is much better than
NSGA-II, and SPEA-II. For diversity of solutions, the diversity of the solutions by COGA-II is comparable to that of SPEA-II, while NSGA-II has poor diversity. COGA-II can also prevent solutions diverging from true Pareto solutions that occur on NSGA-II and SPEA-II for problems with more than 4 objectives. Thus, it can be concluded that COGA-II is suitable for solving an optimization problem with a large number of objectives.
(More)