|
Theory of computing
The “one-fifth rule” with rollbacks for self-adjustment of the population size in the $(1 + (\lambda,\lambda))$ genetic algorithm
A. O. Bassin, M. V. Buzdalov, A. A. Shalyto ITMO University, 49 Kronverkskiy ave., Saint Petersburg 197101, Russia
Abstract:
Self-adjustment of parameters can significantly improve the performance of evolutionary algorithms. A notable example is the $(1 + (\lambda,\lambda))$ genetic algorithm, where adaptation of the population size helps to achieve the linear running time on the OneMax problem. However, on problems which interfere with the assumptions behind the self-adjustment procedure, its usage can lead to the performance degradation. In particular, this is the case with the “one-fifth rule” on problems with weak fitness-distance correlation.
We propose a modification of the “one-fifth rule” in order to have less negative impact on the performance in the cases where the original rule is destructive. Our modification, while still yielding a provable linear runtime on OneMax, shows better results on linear function with random weights, as well as on random satisfiable MAX-3SAT problems.
Keywords:
parameter adaptation, $(1 + (\lambda,\lambda))$ GA, linear functions, MAX-3SAT.
Received: 22.10.2020 Revised: 18.11.2020 Accepted: 16.12.2020
Citation:
A. O. Bassin, M. V. Buzdalov, A. A. Shalyto, “The “one-fifth rule” with rollbacks for self-adjustment of the population size in the $(1 + (\lambda,\lambda))$ genetic algorithm”, Model. Anal. Inform. Sist., 27:4 (2020), 488–508
Linking options:
https://www.mathnet.ru/eng/mais730 https://www.mathnet.ru/eng/mais/v27/i4/p488
|
Statistics & downloads: |
Abstract page: | 119 | Full-text PDF : | 35 | References: | 23 |
|