Quantifying Variable Interactions in Continuous Optimization Problems

被引:41
作者
Sun, Yuan [1 ]
Kirley, Michael [2 ]
Halgamuge, Saman K. [3 ]
机构
[1] Univ Melbourne, Dept Mech Engn, Parkville, Vic 3010, Australia
[2] Univ Melbourne, Dept Comp & Informat Syst, Parkville, Vic 3010, Australia
[3] Australian Natl Univ, Res Sch Engn, Canberra, ACT 2601, Australia
关键词
Continuous optimization problem; exploratory landscape analysis (ELA); maximal information coefficient (MIC); variable interaction; FITNESS LANDSCAPES; SEARCH; ALGORITHM; EVOLUTION;
D O I
10.1109/TEVC.2016.2599164
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Interactions between decision variables typically make an optimization problem challenging for an evolutionary algorithm (EA) to solve. Exploratory landscape analysis (ELA) techniques can be used to quantify the level of variable interactions in an optimization problem. However, many studies using ELA techniques to investigate interactions have been limited to combinatorial problems, with very few studies focused on continuous variables. In this paper, we propose a novel ELA measure to quantify the level of variable interactions in continuous optimization problems. We evaluated the efficacy of this measure using a suite of benchmark problems, consisting of 24 multidimensional continuous optimization functions with differing levels of variable interactions. Significantly, the results reveal that our measure is robust and can accurately identify variable interactions. We show that the solution quality found by an EA is correlated with the level of variable interaction in a given problem. Finally, we present the results from simulation experiments illustrating that when our measure is embedded into an algorithm design framework, the enhanced algorithm achieves equal or better results on the benchmark functions.
引用
收藏
页码:249 / 264
页数:16
相关论文
共 57 条
[31]   Algorithm selection for black-box continuous optimization problems: A survey on methods and challenges [J].
Munoz, Mario A. ;
Sun, Yuan ;
Kirley, Michael ;
Halgamuge, Saman K. .
INFORMATION SCIENCES, 2015, 317 :224-245
[32]   Exploratory Landscape Analysis of Continuous Space Optimization Problems Using Information Content [J].
Munoz, Mario A. ;
Kirley, Michael ;
Halgamuge, Saman K. .
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2015, 19 (01) :74-87
[33]   A comparison of predictive measures of problem difficulty in evolutionary algorithms [J].
Naudts, B ;
Kallel, L .
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2000, 4 (01) :1-15
[34]   Designing benchmark problems for large-scale continuous optimization [J].
Omidvar, Mohammad Nabi ;
Li, Xiaodong ;
Tang, Ke .
INFORMATION SCIENCES, 2015, 316 :419-436
[35]   Cooperative Co-Evolution With Differential Grouping for Large Scale Optimization [J].
Omidvar, Mohammad Nabi ;
Li, Xiaodong ;
Mei, Yi ;
Yao, Xin .
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2014, 18 (03) :378-393
[36]   Empirical fitness landscapes reveal accessible evolutionary paths [J].
Poelwijk, Frank J. ;
Kiviet, Daniel J. ;
Weinreich, Daniel M. ;
Tans, Sander J. .
NATURE, 2007, 445 (7126) :383-386
[37]   A Cooperative Coevolutionary Algorithm with Correlation Based Adaptive Variable Partitioning [J].
Ray, Tapabrata ;
Yao, Xin .
2009 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1-5, 2009, :983-+
[38]  
Reeves C., 1995, Foundations of Genetic Algorithms 3, P7
[39]   Detecting Novel Associations in Large Data Sets [J].
Reshef, David N. ;
Reshef, Yakir A. ;
Finucane, Hilary K. ;
Grossman, Sharon R. ;
McVean, Gilean ;
Turnbaugh, Peter J. ;
Lander, Eric S. ;
Mitzenmacher, Michael ;
Sabeti, Pardis C. .
SCIENCE, 2011, 334 (6062) :1518-1524
[40]  
Rochet S, 1998, LECT NOTES COMPUT SC, V1363, P275, DOI 10.1007/BFb0026607