2007 | OriginalPaper | Chapter
Adaptive Evaluation Strategy Based on Surrogate Model
Authors : Yi-nan Guo, Dun-wei Gong, Hui Wang
Published in: Human-Computer Interaction. Interaction Design and Usability
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Human fatigue is a key problem existing in interactive genetic algorithms which limits population size and generations. Aiming at this problem, evaluation strategies based on surrogate models are presented, in which some individuals are evaluated by models instead of human. Most of strategies adopt fixed substitution proportion, which can not alleviate human fatigue farthest. A novel evaluation strategy with variable substitution proportion is proposed. Substitution proportion lies on models’ precision and human fatigue. Different proportion cause three evaluation phases, which are evaluated by human only, mixed evaluated by human and the model, evaluated by the model only. In third phase, population size is enlarged. Taking fashion evolutionary design system as an example, the validity of the strategy is proved. Simulation results indicate the strategy can effectively alleviate human fatigue and improve the speed of convergence.