Genetic Programming With Knowledge Transfer and Guided Search for Uncertain Capacitated Arc Routing Problem

被引:14
作者
Ardeh, Mazhar Ansari [1 ]
Mei, Yi [1 ]
Zhang, Mengjie [1 ]
机构
[1] Victoria Univ Wellington, Sch Engn & Comp Sci, Evolutionary Computat Res Grp, Wellington 6140, New Zealand
关键词
Routing; Task analysis; Genetic programming; Costs; Search problems; Statistics; Sociology; Arc routing; genetic programming (GP); guided search; hyperheuristics; transfer optimization; ROBUST OPTIMIZATION; SYMBOLIC REGRESSION; BUILDING-BLOCKS; SERVICE;
D O I
10.1109/TEVC.2021.3129278
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The uncertain capacitated arc routing problem has many real-world applications in logistics domains. Genetic programming (GP) is a promising approach to training routing policies to make real-time decisions and handle uncertain events effectively. In the real world, there are various problem domains and no single routing policy can work effectively in all of them. Instead of training in isolation, we can leverage the relatedness between the problems and transfer knowledge from previously solved source problems to solve the target problem. The existing transfer methods are not effective enough due to the loss of diversity during the knowledge transfer. To increase the diversity of the transferred knowledge, in this article, we propose a novel GP method that removes phenotypic duplicates from the source individuals to initialize the target individuals. Furthermore, assuming that the transferred knowledge used in initialization already includes all the important knowledge explored for the source problem, it is more effective to explore new regions that have not been explored for the source problem. Therefore, we propose novel genetic operators that prohibit the search from revisiting the source individuals when solving the target problem. To speed up the revisit check, we propose to adapt a powerful hashing method for routing policies that greatly improves the efficiency of the genetic operators. Our experimental results show that the proposed method significantly outperforms the existing GP approaches with knowledge transfer in terms of both initial and final solution quality.
引用
收藏
页码:765 / 779
页数:15
相关论文
共 79 条
[1]   Multitree Genetic Programming With New Operators for Transfer Learning in Symbolic Regression With Incomplete Data [J].
Al-Helali, Baligh ;
Chen, Qi ;
Xue, Bing ;
Zhang, Mengjie .
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2021, 25 (06) :1049-1063
[2]   The investigation of a class of capacitated arc routing problems: the collection of garbage in developing countries [J].
Amponsah, SK ;
Salhi, S .
WASTE MANAGEMENT, 2004, 24 (07) :711-721
[3]  
[Anonymous], 1963, DISTRIBUTION EREE MU
[4]  
[Anonymous], 2006, Nearest-Neighbor Methods in Learning and Vision: Theory and Practice (Neural Information Processing)
[5]   An efficiency-based path-scanning heuristic for the capacitated arc routing problem [J].
Arakaki, Rafael Kendy ;
Usberti, Fabio Luiz .
COMPUTERS & OPERATIONS RESEARCH, 2019, 103 :288-295
[6]  
Ardeh M. A., 2020, IEEE C EVOL COMPUTAT, P1, DOI [DOI 10.1109/cec48606.2020.9185714, 10.1109/CEC48606.2020.9185714]
[7]  
Ardeh MA, 2020, 2020 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE (SSCI), P2407, DOI [10.1109/ssci47803.2020.9308501, 10.1109/SSCI47803.2020.9308501]
[8]   Genetic Programming Hyper-Heuristic with Knowledge Transfer for Uncertain Capacitated Arc Routing Problem [J].
Ardeh, Mazhar Ansari ;
Mei, Yi ;
Zhang, Mengjie .
PROCEEDINGS OF THE 2019 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE COMPANION (GECCCO'19 COMPANION), 2019, :334-335
[9]   A Novel Genetic Programming Algorithm with Knowledge Transfer for Uncertain Capacitated Arc Routing Problem [J].
Ardeh, Mazhar Ansari ;
Mei, Yi ;
Zhang, Mengjie .
PRICAI 2019: TRENDS IN ARTIFICIAL INTELLIGENCE, PT I, 2019, 11670 :196-200
[10]  
Ardeh MA, 2019, IEEE C EVOL COMPUTAT, P49, DOI [10.1109/CEC.2019.8789920, 10.1109/cec.2019.8789920]