Kriging Hyperparameter Tuning Strategies
- 1 May 2008
- journal article
- Published by American Institute of Aeronautics and Astronautics (AIAA) in AIAA Journal
- Vol. 46 (5) , 1240-1252
Abstract
Response surfaces have been extensively used as a method of building effective surrogate models of high-fidelity computational simulations. Of the numerous types of response surface models, kriging is perhaps one of the most effective due to its ability to model complicated responses through interpolation or regression of known data, whilst also providing an estimate of the error in its prediction. There is however, little information indicating the extent to which the hyperparameters of a kriging model need to be tuned in order for the resulting surrogate model to be effective. The following paper addresses this issue by investigating how often and how well it is necessary to tune the hyperparameters of a kriging model as it is updated during an optimization process. To this end, an optimization benchmarking procedure is introduced and used to assess the performance of five different tuning strategies over a range of problem sizes. The results of this benchmark demonstrate the performance gains that can be associated with reducing the complexity of the hyperparameter tuning process for complicated design problems. The strategy of tuning hyperparameters only once after the initial design of experiments is shown to perform poorly.Keywords
This publication has 18 references indexed in Scilit:
- Design Optimization of a Two-Dimensional Subsonic Engine Air IntakeAIAA Journal, 2006
- Design and Analysis of "Noisy" Computer ExperimentsAIAA Journal, 2006
- Sequential kriging optimization using multiple-fidelity evaluationsStructural and Multidisciplinary Optimization, 2006
- Statistical Improvement Criteria for Use in Multiobjective Design OptimizationAIAA Journal, 2006
- Optimization using surrogate models and partially converged computational fluid dynamics simulationsProceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2006
- Multi-Objective Evolutionary Optimization of Subsonic Airfoils by Kriging Approximation and Evolution Control2005
- Structural optimization using Kriging approximationComputer Methods in Applied Mechanics and Engineering, 2003
- Comparative studies of metamodelling techniques under multiple modelling criteriaStructural and Multidisciplinary Optimization, 2001
- Predicting the output from a complex computer code when fast approximations are availableBiometrika, 2000
- Design and Analysis of Computer ExperimentsStatistical Science, 1989