MULTI-OBJECTIVE EVOLUTIONARY ALGORITHM WITH ENSEMBLE OF EXTERNAL ARCHIVES

作者：Zhao Shi Zheng^*; Suganthan Ponnuthurai Nagaratnam

来源：International Journal of Innovative Computing Information and Control, 2010, 6(4): 1713-1726.

• Ensemble of external archives
• Multi-objective evolutionary algorithm
• Multi-objective particle swarm optimization
• Epsilon non-dominance sorting
• Optimal asset allocation
• Portfolio optimization

This paper proposes an ensemble of e parameter values and an ensemble of external archives in a multi-objective optimization algorithm namely the multi-objective particle swarm optimization algorithm (MOPSO) so that the difficulty of tuning the numerical values of the E parameters for every objective in every problem can be overcome. From the literature, we observe that different objectives and different optimization algorithms require different c values to perform the e non-dominance sorting in order to maintain the diversity of the population. We also observe that there is only a trial-and-error procedure for determining a suitable E value for each objective in a multi-objective optimization problem for a given optimization algorithm. Our experimental results show that an. ensemble of external archives with different e values yields good spread and convergence to the true Pareto-optimal front for difficult problems with different characteristics. Furthermore, we apply the proposed approach to solve the optimal asset allocation problem in portfolio optimization. The experimental results also show the superior performance of the ensemble of external archives with different E values over implementations with only one archive. Although we form an ensemble using different E values, it is also possible to use different diversity preservation methods in different archives in an ensemble of external archives.

• 出版日期2010-4
• 单位南阳理工学院