Graph Learning Assisted Multi-Objective Integer Programming

被引:0
|
作者
Wu, Yaoxin [1 ]
Song, Wen [2 ]
Cao, Zhiguang [3 ]
Zhang, Jie [1 ]
Gupta, Abhishek [3 ]
Lin, Mingyan Simon [3 ]
机构
[1] Nanyang Technol Univ, Singapore, Singapore
[2] Shandong Univ, Inst Marine Sci & Technol, Jinan, Peoples R China
[3] Singapore Inst Mfg Technol A STAR, Singapore, Singapore
基金
中国国家自然科学基金;
关键词
OPTIMIZATION METHODS; SEARCH REGION; ALGORITHM;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Objective-space decomposition algorithms (ODAs) are widely studied for solving multi-objective integer programs. However, they often encounter difficulties in handling scalarized problems, which could cause infeasibility or repetitive non-dominated points and thus induce redundant runtime. To mitigate the issue, we present a graph neural network (GNN) based method to learn the reduction rule in the ODA. We formulate the algorithmic procedure of generic ODAs as a Markov decision process, and parameterize the policy (reduction rule) with a novel two-stage GNN to fuse information from variables, constraints and especially objectives for better state representation. We train our model with imitation learning and deploy it on a state-of-the-art ODA. Results show that our method significantly improves the solving efficiency of the ODA. The learned policy generalizes fairly well to larger problems or more objectives, and the proposed GNN outperforms existing ones for integer programming in terms of test and generalization accuracy.
引用
收藏
页数:14
相关论文
共 50 条
  • [21] A cutting plane approach for multi-objective integer indefinite quadratic programming problem
    Arora R.
    Arora S.R.
    OPSEARCH, 2015, 52 (2) : 367 - 381
  • [22] Evolutionary multi-objective integer programming for the design of adaptive cruise control systems
    Laumanns, N
    Laumanns, M
    Kitterer, H
    DEVELOPMENTS IN APPLIED ARTIFICAIL INTELLIGENCE, PROCEEDINGS, 2002, 2358 : 200 - 210
  • [23] Multi-Objective Optimization Model of Industrial Lubricants Based on Integer Nonlinear Programming
    Yuan, Min
    Li, Yu
    Xu, Wenqiang
    Cui, Wei
    APPLIED SCIENCES-BASEL, 2021, 11 (18):
  • [24] A multi-objective mixed integer linear programming model for thesis defence scheduling
    Almeida, Joao
    Santos, Daniel
    Figueira, Jose Rui
    Francisco, Alexandre P.
    EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 2024, 312 (01) : 92 - 116
  • [25] A simple augmented ε-constraint method for multi-objective mathematical integer programming problems
    Zhang, Weihua
    Reimann, Marc
    EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 2014, 234 (01) : 15 - 24
  • [26] A new cutting plane method for lexicographic multi-objective integer linear programming
    Cococcioni, Marco
    Cudazzo, Alessandro
    Fiaschi, Lorenzo
    Pappalardo, Massimo
    Sergeyev, Yaroslav D.
    COMMUNICATIONS IN NONLINEAR SCIENCE AND NUMERICAL SIMULATION, 2024, 129
  • [27] A multi-objective mixed-integer programming model for a multi-floor facility layout
    Hathhorn, Jonathan
    Sisikoglu, Esra
    Sir, Mustafa Y.
    INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH, 2013, 51 (14) : 4223 - 4239
  • [28] Machine Learning Assisted Evolutionary Multi-Objective Optimization
    Zhang, Xingyi
    Cheng, Ran
    Feng, Liang
    Jin, Yaochu
    IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE, 2023, 18 (02) : 16 - 17
  • [29] Human Assisted Learning by Evolutionary Multi-Objective Optimization
    Liu, Dan-Xuan
    Mu, Xin
    Qian, Chao
    THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 10, 2023, : 12453 - 12461
  • [30] MULTI-OBJECTIVE INTERACTIVE PROGRAMMING
    WHITE, DJ
    JOURNAL OF THE OPERATIONAL RESEARCH SOCIETY, 1980, 31 (06) : 517 - 523