A Supervised Surrogate-Assisted Evolutionary Algorithm for Complex Optimization Problems

Surrogate-assisted evolutionary algorithms (SAEAs), which use surrogate models to evaluate the individuals' fitness, show high efficiency in solving complex optimization problems. In an SAEA, the solution quality and algorithm efficiency are the two most concerned performance measures. It is ne...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on instrumentation and measurement Ročník 72; s. 1
Hlavní autoři: Zhao, Xin, Jia, Xue, Zhang, Tao, Liu, Tianwei, Cao, Yahui
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York IEEE 01.01.2023
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
algorithm efficiency
Computational modeling
Computing time
Efficiency
Evolutionary algorithms
Finite element method
Fitness
Genetic algorithms
Mathematical models
new individual generation strategy
Optimization
Sociology
solution quality
Statistics
Supervised surrogate-assisted evolutionary algorithm
surrogate model management strategy
Training
ISSN:0018-9456, 1557-9662
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Surrogate-assisted evolutionary algorithms (SAEAs), which use surrogate models to evaluate the individuals' fitness, show high efficiency in solving complex optimization problems. In an SAEA, the solution quality and algorithm efficiency are the two most concerned performance measures. It is necessary to bring novel strategies to SAEAs to improve their solution quality and algorithm efficiency. In this paper, we propose a supervised surrogate-assisted evolutionary algorithm (SSAEA). The SSAEA takes the fitness evaluation accuracy (FEA) as a supervisor. Under the supervisor, the SSAEA brings two novel strategies, including the FEA-based surrogate model management strategy and the FEA-based new individual generation strategy. In our experiments, we compare the proposed SSAEA with several state-of-art SAEAs. The experimental results show that our proposed algorithm can obtain higher-quality solutions in a shorter computational time.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9456
1557-9662
DOI:10.1109/TIM.2023.3261905