A Many-objective Evolutionary Algorithm using Determinantal Point Process in Potential Region
Mengzhen Wang, Fangzhen Ge, Debao Chen, Huaiyu Liu · 2022
Most current many-objective optimization algorithms mainly attempt to construct various strategies to achieve convergence and maintain diversity. To simplify the complexity of algorithm design, we propose a many-objective optimization algorithm which introduces a ratio-based infinite norm indicator to find the optimal solutions in the current population and uses them to determine potential region where optimal solutions exist; then samples solutions with better convergence and diversity within the potential region using a determinantal point process. The results of comparing the algorithm with four algorithms on the WFG, MaF and DTLZ test sets show that our algorithm is competitive.