An enhanced reference vectors-based multi-objective evolutionary algorithm with neighborhood-based adaptive adjustment

The decomposition-based evolutionary algorithms have shown great potential in multi-objective optimization and many-objective optimization. However, their performance strongly depends on the Pareto front shapes. This may result from the fixed reference vectors, which will waste computing resources w...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Neural computing & applications Jg. 32; H. 15; S. 11767 - 11789
Hauptverfasser: Fan, Rui, Wei, Lixin, Sun, Hao, Hu, Ziyu
Format: Journal Article
Sprache:Englisch
Veröffentlicht: London Springer London 01.08.2020
Springer Nature B.V
Schlagworte:
Adaptive algorithms
Artificial Intelligence
Computational Biology/Bioinformatics
Computational Science and Engineering
Computer Science
Computer simulation
Convergence
Data Mining and Knowledge Discovery
Evolutionary algorithms
Genetic algorithms
Image Processing and Computer Vision
Multiple criterion
Multiple objective analysis
Neighborhoods
Original Article
Pareto optimization
Probability and Statistics in Computer Science
Shape optimization
Solution space
Multi-objective optimization
Many-objective optimization
Reference vector
Adaptive adjustment
Evolutionary computation
ISSN:0941-0643, 1433-3058
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The decomposition-based evolutionary algorithms have shown great potential in multi-objective optimization and many-objective optimization. However, their performance strongly depends on the Pareto front shapes. This may result from the fixed reference vectors, which will waste computing resources when handling irregular Pareto fronts. Inspired by this issue, an enhanced reference vectors-based multi-objective evolutionary algorithm with neighborhood-based adaptive adjustment (MOEA-NAA) is proposed. Firstly, a few individuals of the population are used to search the solution space to accelerate the convergence speed until enough non-dominated solutions are found. Then, a multi-criteria environment selection mechanism is implemented to achieve the balance between convergence and diversity, which makes a fusion between dominance-based method and reference vector-based method. Finally, according to the neighborhood information, a small-scale reference vectors adaptive fine-tuning strategy is introduced to enhance the adaptability of different Pareto fronts. To validate the efficiency of MOEA-NAA, experiments are conducted to compare it with four state-of-the-art evolutionary algorithms. The simulation results have shown that the proposed algorithm outperforms the compared algorithms for overall performance.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0941-0643
1433-3058
DOI:10.1007/s00521-019-04660-5