Тип публикации: доклад, тезисы доклада, статья из сборника материалов конференций
Конференция: International Workshop “Hybrid methods of modeling and optimization in complex systems” (HMMOCS 2022); Krasnoyarsk; Krasnoyarsk
Год издания: 2022
Идентификатор DOI: 10.15405/epct.23021.13
Ключевые слова: differential evolution, population size, parameter adaptation
Аннотация: In this paper a novel algorithmic scheme for differential evolution is proposed with two populations and new mutation strategies. The populations contain latest and best individuals, and the sizes of these populations are controlled independently. The LL-NADE algorithm uses the parameter adaptation scheme proposed in L-SHADE, and tПоказать полностьюhe sizes of both populations are reduced at the same time. The experiments described in the paper are performed on the CEC 2022 benchmark suite. The total standard score is used for comparison of different algorithms on a set of benchmark problems. The experiments are performed with six different mutation strategies, which utilize individuals from both populations. The most efficient mutation strategy utilizes directed search from newest to top individuals. It is shown that LL-NADE is capable of demonstrating results comparable with state-of-the-art algorithms. Comparison on various test functions and dimensions have shown that LL-NADE performs in a way that is not similar to other differential evolution algorithms, achieving better results on some problems, but worse on other.
Журнал: HYBRID METHODS OF MODELING AND OPTIMIZATION IN COMPLEX SYSTEMS
Номера страниц: 102-110
Место издания: London, United Kingdom
Издатель: European Proceedings