Population Diversity in Permutation-Based Genetic Algorithm
Abstract
This paper presents an empirical study of population diversity measures and adaptive control of diversity in the context of a permutation-based algorithm for Traveling Salesman Problems and Vehicle Routing Problems. We provide detailed graphical observations and discussion of the relationship among the four diversity measures and suggest a moderate correlation between diversity and search performance under simple conditions. We also study the effects of adapting key genetic control parameters such as crossover and mutation rates on the population diversity. We are able to show that adaptive control of the genetic operations based on population diversity effectively outperforms fixed parameter genetic algorithms.
Cite
Text
Zhu and Liu. "Population Diversity in Permutation-Based Genetic Algorithm." European Conference on Machine Learning, 2004. doi:10.1007/978-3-540-30115-8_49Markdown
[Zhu and Liu. "Population Diversity in Permutation-Based Genetic Algorithm." European Conference on Machine Learning, 2004.](https://mlanthology.org/ecmlpkdd/2004/zhu2004ecml-population/) doi:10.1007/978-3-540-30115-8_49BibTeX
@inproceedings{zhu2004ecml-population,
title = {{Population Diversity in Permutation-Based Genetic Algorithm}},
author = {Zhu, Kenny Qili and Liu, Ziwei},
booktitle = {European Conference on Machine Learning},
year = {2004},
pages = {537-547},
doi = {10.1007/978-3-540-30115-8_49},
url = {https://mlanthology.org/ecmlpkdd/2004/zhu2004ecml-population/}
}