Bipolar Mating Tendency: Harmony Between the Best and the Worst Individuals

MC Gençal, M Oral - Arabian Journal for Science and Engineering, 2022 - Springer
Arabian Journal for Science and Engineering, 2022Springer
Evolutionary algorithms (EAs) are successfully employed to solve complex optimization
problems such as network design problems, path finding problems, scheduling problems
and social and economic planning. By using intelligent strategies, EAs iteratively improve
the initial solution to produce new solutions. Particularly in genetic algorithms (GAs), one of
the most known EAs, selection mechanism, has crucial importance due to its effect on the
diversity of a population and convergence ability to optimal solution. As most of the selection …
Abstract
Evolutionary algorithms (EAs) are successfully employed to solve complex optimization problems such as network design problems, path finding problems, scheduling problems and social and economic planning. By using intelligent strategies, EAs iteratively improve the initial solution to produce new solutions. Particularly in genetic algorithms (GAs), one of the most known EAs, selection mechanism, has crucial importance due to its effect on the diversity of a population and convergence ability to optimal solution. As most of the selection methods of GAs favor the good solutions for reproductions, average or poor solutions are overlooked. Recently, we proposed a new selection method, called Bipolar Mating Tendency (BMT), that is based on Standard Tournament (ST) selection. Unlike the state-of-the-art selection methods of GAs, BMT diversifies the mating process with the inclusion of less favorable solutions. In this paper, the performance of BMT was compared with the performance of ST and prevalent selection methods that are also based on ST: restricted tournament selection, unbiased tournament selection, fine-grained tournament selection and cooperative selection. For the comparison, twenty-one well-known benchmark optimization functions and sixteen IEEE CEC 2018 test functions were utilized. Also, nonparametric statistical tests were performed to demonstrate the significance of the results. In order to test the effectiveness of GA-BMT (GA that uses BMT as a selection method), three real-world engineering problems were used. In addition to showing predominant results comparing to the other methods in test functions, BMT has also provided successful results in the problems.
Springer
以上显示的是最相近的搜索结果。 查看全部搜索结果