Title | Theoretical Analyses of Multi-Objective Evolutionary Algorithms on Multi-Modal Objectives |
Author | |
Corresponding Author | Doerr, Benjamin; Zheng, Weijie |
Publication Years | 2021
|
Conference Name | 35th AAAI Conference on Artificial Intelligence, AAAI 2021
|
ISBN | 9781713835974
|
Source Title | |
Volume | 14A
|
Pages | 12293-12301
|
Conference Date | February 2, 2021 - February 9, 2021
|
Conference Place | Virtual, Online
|
Publisher | |
Abstract | Previous theory work on multi-objective evolutionary algorithms considers mostly easy problems that are composed of unimodal objectives. This paper takes a first step towards a deeper understanding of how evolutionary algorithms solve multi-modal multi-objective problems. We propose the ONEJUMPZEROJUMP problem, a bi-objective problem whose single objectives are isomorphic to the classic jump functions benchmark. We prove that the simple evolutionary multi-objective optimizer (SEMO) cannot compute the full Pareto front. In contrast, for all problem sizes n and all jump sizes k ∈ [4..n/2 - 1], the global SEMO (GSEMO) covers the Pareto front in T((n - 2k)nk) iterations in expectation. To improve the performance, we combine the GSEMO with two approaches, a heavy-tailed mutation operator and a stagnation detection strategy, that showed advantages in singleobjective multi-modal problems. Runtime improvements of asymptotic order at least kω(k) are shown for both strategies. Our experiments verify the substantial runtime gains already for moderate problem sizes. Overall, these results show that the ideas recently developed for single-objective evolutionary algorithms can be effectively employed also in multiobjective optimization. © 2021, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. |
SUSTech Authorship | Corresponding
|
Language | English
|
Indexed By | |
Funding Project | This work was supported by a public grant as part of the Investissement d’avenir project, reference ANR-11-LABX-0056-LMH, LabEx LMH.This work was also supported by Guangdong Basic and Applied Basic Research Foundation (Grant No. 2019A1515110177); Guangdong Provincial Key Laboratory (Grant No. 2020B121201001), the Program for Guangdong Introducing Innovative and Enterpreneurial Teams (Grant No. 2017ZT07X386); Shenzhen Peacock Plan (Grant No. KQTD2016112514355531); and the Program for University Key Laboratory of Guangdong Province (Grant No. 2017KSYS008).
|
EI Accession Number | 20222012122223
|
EI Keywords | Artificial intelligence
; Genetic algorithms
|
ESI Classification Code | Artificial Intelligence:723.4
; Optimization Techniques:921.5
|
Data Source | EV Compendex
|
Document Type | Conference paper |
Identifier | http://kc.sustech.edu.cn/handle/2SGJ60CL/411703 |
Department | Department of Computer Science and Engineering |
Affiliation | 1.Laboratoire d'Informatique (LIX), Ecole Polytechnique, CNRS, Institut Polytechnique de Paris, Palaiseau, France 2.Guangdong Provincial Key Laboratory of Brain-inspired Intelligent Computation, Department of Computer Science and Engineering, Southern University of Science and Technology, Shenzhen, China |
Corresponding Author Affilication | Department of Computer Science and Engineering |
Recommended Citation GB/T 7714 |
Doerr, Benjamin,Zheng, Weijie. Theoretical Analyses of Multi-Objective Evolutionary Algorithms on Multi-Modal Objectives[C]:Association for the Advancement of Artificial Intelligence,2021:12293-12301.
|
Files in This Item: | There are no files associated with this item. |
|
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment