https://scholars.lib.ntu.edu.tw/handle/123456789/635961
標題: | Relational Bayesian Optimization for Permutation | 作者: | Huang, Bo Wei Liao, Hsu Chen Fang, Wen Zhong TIAN-LI YU |
關鍵字: | binary relation | estimation of distribution algorithms | permutation | 公開日期: | 15-七月-2023 | 來源出版物: | GECCO 2023 Companion - Proceedings of the 2023 Genetic and Evolutionary Computation Conference Companion | 摘要: | Relational Bayesian optimization for permutation (RBOP) is a new permutation estimation of distribution algorithm proposed in this paper. RBOP uses binary relations to represent the common property in permutations. Inspired by the Bayesian optimization algorithm, RBOP first builds a Bayesian network using binary relations. Then, RBOP samples genes using the most certain edge in the Bayesian network. In the scenario of black-box optimization, RBOP aims to solve various permutation problems with a limited number of function evaluations. Experiments show that in terms of average relative percentage deviation, RBOP outperforms edge histogram-based sampling algorithm on quadratic assignment problems, permutation flow shop problems and linear ordering problems. Additionally, RBOP also outperforms both node histogram-based sampling algorithm and kernels of Mallows model using Cayley distance on traveling salesman problems, permutation flow shop problems, linear ordering problems and 6 out of 10 instances of quadratic assignment problems. |
URI: | https://scholars.lib.ntu.edu.tw/handle/123456789/635961 | ISBN: | 9798400701207 | DOI: | 10.1145/3583133.3590606 |
顯示於: | 電機工程學系 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。