Theoretical Analyses of Multiobjective Evolutionary Algorithms on Multimodal Objectives

被引:5
作者
Zheng, Weijie [1 ]
Doerr, Benjamin [2 ]
机构
[1] Harbin Inst Technol, Int Res Inst Artificial Intelligence, Sch Comp Sci & Technol, Shenzhen, Peoples R China
[2] Inst Polytech Paris, Ecole Polytech, CNRS, Lab Informat LIX, Palaiseau, France
关键词
Multiobjective evolutionary algorithms; multimodal problems; theory of computing; runtime analysis; EXPECTED RUNTIMES; SELECTION; CROSSOVER;
D O I
10.1162/evco_a_00328
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Multiobjective evolutionary algorithms are successfully applied in many real-world multiobjective optimization problems. As for many other AI methods, the theoretical understanding of these algorithms is lagging far behind their success in practice. In particular, previous theory work considers mostly easy problems that are composed of unimodal objectives.As a first step towards a deeper understanding of how evolutionary algorithms solve multimodal multiobjective problems, we propose the OneJumpZeroJump problem, a bi-objective problem composed of two objectives isomorphic to the classic jump function benchmark. We prove that the simple evolutionary multiobjective optimizer (SEMO) with probability one does not compute the full Pareto front, regardless of the runtime. In contrast, for all problem sizes n and all jump sizes k is an element of[4..n/2-1], the global SEMO (GSEMO) covers the Pareto front in an expected number of Theta((n-2k)n(k)) iterations. For k=o(n), we also show the tighter bound 3/2en(k+1)+/- o(n(k+1)), which might be the first runtime bound for an MOEA that is tight apart from lower-order terms. We also combine the GSEMO with two approaches that showed advantages in single-objective multimodal problems. When using the GSEMO with a heavy-tailed mutation operator, the expected runtime improves by a factor of at least k(Omega(k)). When adapting the recent stagnation-detection strategy of Rajabi and Witt (2022) to the GSEMO, the expected runtime also improves by a factor of at least k(Omega(k)) and surpasses the heavy-tailed GSEMO by a small polynomial factor in k. Via an experimental analysis, we show that these asymptotic differences are visible already for small problem sizes: A factor-5 speed-up from heavy-tailed mutation and a factor-10 speed-up from stagnation detection can be observed already for jump size 4 and problem sizes between 10 and 50. Overall, our results show that the ideas recently developed to aid single-objective evolutionary algorithms to cope with local optima can be effectively employed also in multiobjective optimization.
引用
收藏
页码:337 / 373
页数:37
相关论文
共 74 条
[1]  
[Anonymous], 2008, P GECCO 2008
[2]   Fast Mutation in Crossover-Based Algorithms [J].
Antipov, Denis ;
Buzdalov, Maxim ;
Doerr, Benjamin .
ALGORITHMICA, 2022, 84 (06) :1724-1761
[3]  
Antipov D, 2022, ALGORITHMICA, V84, P1573, DOI 10.1007/s00453-021-00907-7
[4]   Lazy Parameter Tuning and Control: Choosing All Parameters Randomly From a Power-Law Distribution [J].
Antipov, Denis ;
Buzdalov, Maxim ;
Do Err, Benjamin .
PROCEEDINGS OF THE 2021 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE (GECCO'21), 2021, :1115-1123
[5]  
Antipov D, 2020, TH CO SC GE ISS, V12270, P545, DOI 10.1007/978-3-030-58115-2_38
[6]  
BACK T, 1993, PROCEEDINGS OF THE FIFTH INTERNATIONAL CONFERENCE ON GENETIC ALGORITHMS, P2
[7]   A Rigorous Runtime Analysis of the 2-MMASib on Jump Functions: Ant Colony Optimizers Can Cope Well With Local Optima [J].
Benbaki, Riade ;
Benomar, Ziyad ;
Doerr, Benjamin .
PROCEEDINGS OF THE 2021 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE (GECCO'21), 2021, :4-13
[8]   Better Running Time of the Non-dominated Sorting Genetic Algorithm II (NSGA-II) by Using Stochastic Tournament Selection [J].
Bian, Chao ;
Qian, Chao .
PARALLEL PROBLEM SOLVING FROM NATURE - PPSN XVII, PPSN 2022, PT II, 2022, 13399 :428-441
[9]  
Bian C, 2018, PROCEEDINGS OF THE TWENTY-SEVENTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, P1405
[10]  
Brockhoff D, 2007, GECCO 2007: GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE, VOL 1 AND 2, P765