Cooperative co-evolutionary algorithms have a huge potential in optimizing large-scale problems. In a such divide-and-conquer strategy, the decomposition step plays a crucial part in the performance of the algorithm. Automatic decomposition strategies that can uncover the interaction structure between decision variables have been introduced in recent years. However, such strategies for large-scale constrained problems are quite limited in number so far and yet, they are interesting for at least two reasons. On the one hand, they help to find a feasible region faster. On the other hand, they also improve the convergence rate for the optimization itself. In this paper, we propose a novel cooperative co-evolutionary algorithm, DGD-EA for Differential Grouping Evolutionary Algorithm, that performs an automatic decomposition of decision variables and allows to optimize large-scale constrained problems. Its performance is evaluated on a set of 10 benchmark functions specially created for this study.
|Pages (de - à)||697-704|
|Nombre de pages||8|
|journal||Proceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO '17)|
|état||Publié - 15 juil. 2017|
Contient cette citation
Blanchard, J., Beauthier, C., & Carletti, T. (2017). A Cooperative Co-evolutionary Algorithm for solving Large-Scale Constrained Problems with Interaction Detection. Proceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO '17), 697-704. https://doi.org/10.1145/3071178.3071187