Abstract
Support vector regression (SVR) is a common surrogate model for computationally expensive simulation. It is able to balance the model complexity and the error tolerance. Whether SVR interpolates the training samples is dependent on its parameters. For the nonlinear function approximation without noise, when SVR is not an interpolator, it is advisable to model the errors and use them to compensate the prediction response. In this paper, the errors of SVR are modeled by using Gaussian process, and the final model response is obtained by the combination of SVR and the Gaussian process of the errors. The numerical experiments show the proposed method is able to further improve the prediction accuracy of SVR.
Original language | English |
---|---|
Title of host publication | Advances in Structural and Multidisciplinary Optimization |
Subtitle of host publication | Proceedings of the 12th World Congress of Structural and Multidisciplinary Optimization (WCSMO12) |
Editors | A Schumacher, T Vietor, S Fiebig, K-U Bletzinger, K Maute |
Place of Publication | Cham, Switzerland |
Publisher | Springer |
Pages | 281-286 |
ISBN (Electronic) | 978-3-319-67988-4 |
ISBN (Print) | 978-3-319-67987-7 |
DOIs | |
Publication status | Published - 2017 |
Event | WCSMO 2017: 12th World Congress of Structural and Multidisciplinary Optimisation - Braunschweig, Germany Duration: 5 Jun 2017 → 9 Jun 2017 |
Conference
Conference | WCSMO 2017: 12th World Congress of Structural and Multidisciplinary Optimisation |
---|---|
Country/Territory | Germany |
City | Braunschweig |
Period | 5/06/17 → 9/06/17 |
Keywords
- Support vector regression
- Gaussian kernel
- Error modeling
- Gaussian process