Individuals redistribution based on differential evolution for covariance matrix adaptation evolution strategy.

Area of Science:

• Computational intelligence
• Optimization algorithms
• Evolutionary computation

Background:

• Differential Evolution (DE) and Covariance Matrix Adaptation Evolution Strategy (CMA-ES) are leading population-based metaheuristics for real parameter single objective optimization.
• CMA-ES often exhibits premature stagnation compared to DE, limiting its effectiveness in complex optimization landscapes.

Purpose of the Study:

• To propose an enhanced optimization algorithm, IR-CMA-ES, that addresses the early stagnation issue in CMA-ES.
• To improve the performance and robustness of CMA-ES in real parameter single objective optimization.

Main Methods:

• A novel approach, IR-CMA-ES, is developed by integrating individuals redistribution strategy from DE into CMA-ES.
• The proposed IR-CMA-ES algorithm was rigorously tested against nine peer algorithms using two benchmark test suites.

Main Results:

• Experimental results demonstrate that IR-CMA-ES effectively mitigates the stagnation problem inherent in standard CMA-ES.
• The proposed algorithm achieves competitive performance compared to existing state-of-the-art methods in real parameter single objective optimization.

Conclusions:

• IR-CMA-ES presents a viable and competitive alternative for real parameter single objective optimization.
• The integration of DE's redistribution mechanism enhances CMA-ES's ability to escape local optima and maintain population diversity.