Competitive and cooperative coevolution for pareto multiobjective optimization

Tan, Tse Guan (2008) Competitive and cooperative coevolution for pareto multiobjective optimization. Masters thesis, Universiti Malaysia Sabah.


Download (2MB) | Preview


In the real world, it is rare for any problem or task to concern only a single value or objective. Generally, most of the real-life search and optimization problems naturally involve multiple objectives or parameters. In recent years, a number of Multiobjective Optimization Evolutionary Algorithms (MOEAs) have been proposed to solve multiobjective optimization problems (MOPs). MOEAs can find a set of optimal solutions that trade-off quality in a space shared by several conflicting functions. In this study, the enhancement of a standard MOEA with the coevolutionary approach is investigated. The objective is to augment a well-known MOEA, the Strength Pareto Evolutionary Algorithm 2 (SPEA2), with two types of the coevolutionary approach respectively, Competitive Coevolution (CE) and Cooperative Coevolution (CC). Currently, SPEA2 has been noticed to provide less favorable maintenance of diverse solutions in certain multi-objective optimization problems and to a lesser extent, its convergence to the true Pareto front. Two different CE approaches were studied, namely Hall of Fame and K-Random Opponents resulting in two new SPEA2 algorithms using CE, which are the SPEA2-CE-HOF and SPEA2-CE-KR algorithms respectively. Using the CC approach, one new SPEA2 algorithm was proposed, which is the SPEA2-CC algorithm. The proposed algorithms were benchmarked against the original SPEA2 using seven scalable benchmark test problems having 3 to 5 objectives based on the generational distance, spacing and coverage metrics. Overall, the experimental results indicate that the three proposed algorithms performed better than the original SPEA2 in terms of the average distance of the nondominated solutions to the true Pareto front, spacing index of the obtained solutions and also the coverage level. Additionally, the CC approach was found to be superior to the CE approach in terms of enhancing the performance of the original SPEA2 algorithm.

Item Type: Thesis (Masters)
Uncontrolled Keywords: multiobjective optimization, Strenght Pareto Evolutionary Algorithm (SPEA), competitive, cooperative
Subjects: Q Science > QA Mathematics > QA76 Computer software
Divisions: SCHOOL > School of Engineering and Information Technology
Date Deposited: 28 Jun 2013 08:31
Last Modified: 11 Oct 2017 04:38

Actions (login required)

View Item View Item