An adaptive ensemble-based algorithm

Simulation-driven optimization problems are often used with the aid of metamodels, as the latter provide predicted objective values at a lower computational cost. To further improve the prediction accuracy, an ensemble of metamodel incorporates several metamodels concurrently and aggregates their prediction into a single one. An open question is which types of metamodels should be incorporated, and this is typically not known a-priori, while using a fixed topology may hamper the optimization process. To address this issue, this study proposes a new metamodel-assisted algorithm with dynamic topology adaptation, namely, it autonomously adapts the ensemble topology during the search, and selects the most suitable topology as the search progresses. An extensive performance analysis shows the effectiveness of the proposed algorithm, and highlights the merit of the proposed topology adaptation.