Control of JADE Population in Limited Number of Searches for Realistic Situations
Tomoya Matsuki, Akira Notsu, Katsuhiro Honda, Takuya Kato, Masakazu Shibahara · 2024
JADE is an optimization algorithm that can be applied to black-box optimization problems. However, real op-timization problems, such as our optimization target AI line heating, allow only a small number of searches. Here, diversity is not required to find a solution at the end of the search, and therefore, it is very easy to improve performance by deleting search points appropriately. In this study, we propose a method for JADE that adapts the method for deleting individuals in the final stages of the search. Numerical experiments confirm that the proposed method improves the solution for all benchmark problems.