Multiobjective Optimization Using Age-MOEA-II with Improved Environmental Selection

This paper aims to develop a multiobjective evolutionary algorithm (MOEA) to tackle multiobjective optimization problems with various Pareto front geometries. We propose an improved version of AGE-MOEA-II by incorporating a hyper-dominance-based filtering mechanism to promote convergence and an adaptive distance metric that balances geodesic and parallel distances to enhance diversity. Extensive experiments on the MaF test suite demonstrate that I-AGE-MOEA-II achieves superior performance compared to several state-of-The-Art MOEAs. Additional component-wise analysis confirms the effectiveness of the proposed improvements. These results suggest that the proposed I-AGE-MOEA-II algorithm offers a robust and scalable solution for handling complex MOPs.