METAMODEL-ASSISTED MULTI-OBJECTIVE EVOLUTIONARY OPTIMIZATION
Marios K. Karakasis, Kyriakos C. Giannakoglou · DSpace - NTUA (National Technical University of Athens) · 2005
The use of surrogate evaluation models or metamodels in multi-objective Evolutionary Algorithms with computationally expensive evaluations for the reduction of computational cost, through controlled approximate evaluations of generation members, is presented. The metamodels assist the Evolutionary Algorithm by filtering the poorly performing individuals within each generation and subsequently by allowing only the most promising among them to be exactly evaluated.