DEA2H2: differential evolution architecture based adaptive hyper-heuristic algorithm for continuous optimization

This paper proposes a novel differential evolution (DE) architecture based hyper-heuristic algorithm (DEA 2 H 2 ) for solving continuous optimization tasks. A representative hyper-heuristic algorithm consists of two main components: low-level and high-level components. In the low-level component, DE...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Cluster computing Jg. 27; H. 9; S. 12239 - 12266
Hauptverfasser: Zhong, Rui, Yu, Jun
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York Springer US 01.12.2024
Springer Nature B.V
Schlagworte:
Ablation
Adaptation
Adaptive algorithms
Computer Communication Networks
Computer Science
Deep learning
Distance learning
Effectiveness
Evolutionary algorithms
Evolutionary computation
Genetic algorithms
Heuristic
Heuristic methods
Mutation
Operating Systems
Operators (mathematics)
Optimization
Optimization techniques
Performance evaluation
Processor Architectures
Searching
Source code
Statistical analysis
Success
Hyper-heuristic (HH)
Success history knowledge
Continuous optimization
Differential evolution (DE)
ISSN:1386-7857, 1573-7543
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This paper proposes a novel differential evolution (DE) architecture based hyper-heuristic algorithm (DEA 2 H 2 ) for solving continuous optimization tasks. A representative hyper-heuristic algorithm consists of two main components: low-level and high-level components. In the low-level component, DEA 2 H 2 leverages ten DE-derived search operators as low-level heuristics (LLHs). In the high-level component, we incorporate a success-history-based mechanism inspired by the success-history-based parameter adaptation in success-history adaptive DE (SHADE). Specifically, if a parent individual successfully evolves an offspring individual using a specific search operator, that corresponding operator is preserved for subsequent iterations. On the contrary, if the evolution is unsuccessful, the search operator is replaced by random initialization. To validate the effectiveness of DEA 2 H 2 , we conduct comprehensive numerical experiments on both CEC2020 and CEC2022 benchmark functions, as well as eight engineering problems. We compare the performance of DEA 2 H 2 against fifteen well-known metaheuristic algorithms (MA). Additionally, ablation experiments are performed to investigate the effectiveness of the success-history-based high-level component independently. The experimental results and statistical analyses affirm the superiority and robustness of DEA 2 H 2 across diverse optimization tasks, highlighting its potential as an effective tool for continuous optimization problems. The source code of this research can be downloaded from https://github.com/RuiZhong961230/DEA2H2 .
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1386-7857
1573-7543
DOI:10.1007/s10586-024-04587-0