An Expensive Multi-Objective Evolutionary Algorithm with RBF-IDW Surrogate Model

In surrogate-assisted multi-objective evolutionary optimization, surrogate models provide approximations for each solution on each objective, with varying quality. In the proposed algorithm, the RBF-IDW surrogate model replaces the Gaussian process model used in the original algorithm, constructing...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2024 International Conference on New Trends in Computational Intelligence (NTCI) S. 86 - 90
Hauptverfasser: Liu, Yuhao, Fang, Hongyi, Li, Fei
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 18.10.2024
Schlagworte:
Adaptation models
Approximation algorithms
Computational modeling
Gaussian processes
infill criterion
Linear programming
multi-/many objective optimization
Optimization
performance indicator
Prediction algorithms
Predictive models
Surrogate model
Uncertainty
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In surrogate-assisted multi-objective evolutionary optimization, surrogate models provide approximations for each solution on each objective, with varying quality. In the proposed algorithm, the RBF-IDW surrogate model replaces the Gaussian process model used in the original algorithm, constructing a model for each objective function and identifying candidate solutions during the optimization phase. A Gaussian process model is then used to approximate performance indicators measuring convergence and diversity. Finally, the solution with the maximum expected improvement is selected and evaluated using the expensive objective function. The algorithm's feasibility is demonstrated through tests on the DTLZ and WFG suites.
DOI:10.1109/NTCI64025.2024.10776091