Abstract: |
In this paper, we investigate the relative performance of several well-known Multi-Objective GA’s (MOGA’s) on a quantitative and objective basis. Two aspects of MOGA performance are studied and compared: 1) convergence rate to the Pareto frontier; and 2) diversity of solutions. Two quantitative measures (i.e. quality indices) are selected accordingly, each addressing one aspect of performance. To account for the stochastic nature of GA’s, multiple runs are performed for each MOGA to create samples of solution sets for statistical comparison. Based on the results of this study, we will observe several characteristics of the tested MOGA’s, which can be helpful in selecting an appropriate MOGA for a given multi-objective optimization task. |
