Advanced search
2 files | 933.35 KB Add to list

Multi-objective Bayesian optimization for engineering simulation

Joachim van der Herten (UGent) , Nicolas Knudde (UGent) , Ivo Couckuyt (UGent) and Tom Dhaene (UGent)
Author
Organization
Abstract
Rather than optimizing expensive objective functions such as complex engineering simulations directly, Bayesian optimization methodologies fit a surrogate model (typically Kriging or a Gaussian Process) on evaluations of the objective function(s). To determine the next evaluation, an acquisition function is optimized (also referred to as infill criterion or sampling policy) which incorporates the model prediction and uncertainty and balances exploration and exploitation. Therefore, Bayesian optimization methodologies replace a single optimization of the objective function by a sequence of optimization problems: this makes sense as the acquisition function is cheap-to-evaluate whereas the objective is not. Depending on the goal different acquisition functions are available: multi-objective acquisition functions are relatively new and this chapter gives a state-of-the-art overview and illustrates some approaches based on hypervolume improvement. It is shown that the quality of the model is crucial for the performance of Bayesian optimization and illustrate this by using the more flexible Student-t processes as surrogate models.

Downloads

  • (...).pdf
    • full text (Published version)
    • |
    • UGent only
    • |
    • PDF
    • |
    • 533.36 KB
  • (...).pdf
    • full text (Accepted manuscript)
    • |
    • UGent only (changes to open access on 2022-12-01)
    • |
    • PDF
    • |
    • 399.99 KB

Citation

Please use this url to cite or link to this publication:

MLA
van der Herten, Joachim, et al. “Multi-Objective Bayesian Optimization for Engineering Simulation.” High-Performance Simulation-Based Optimization, edited by T Bartz-Beielstein et al., vol. 833, Springer, 2020, pp. 47–68, doi:10.1007/978-3-030-18764-4_3.
APA
van der Herten, J., Knudde, N., Couckuyt, I., & Dhaene, T. (2020). Multi-objective Bayesian optimization for engineering simulation. In T. Bartz-Beielstein, B. Filipič, P. Korošec, & E. Talbi (Eds.), High-performance simulation-based optimization (Vol. 833, pp. 47–68). Cham: Springer. https://doi.org/10.1007/978-3-030-18764-4_3
Chicago author-date
Herten, Joachim van der, Nicolas Knudde, Ivo Couckuyt, and Tom Dhaene. 2020. “Multi-Objective Bayesian Optimization for Engineering Simulation.” In High-Performance Simulation-Based Optimization, edited by T Bartz-Beielstein, B Filipič, P Korošec, and E Talbi, 833:47–68. Cham: Springer. https://doi.org/10.1007/978-3-030-18764-4_3.
Chicago author-date (all authors)
van der Herten, Joachim, Nicolas Knudde, Ivo Couckuyt, and Tom Dhaene. 2020. “Multi-Objective Bayesian Optimization for Engineering Simulation.” In High-Performance Simulation-Based Optimization, ed by. T Bartz-Beielstein, B Filipič, P Korošec, and E Talbi, 833:47–68. Cham: Springer. doi:10.1007/978-3-030-18764-4_3.
Vancouver
1.
van der Herten J, Knudde N, Couckuyt I, Dhaene T. Multi-objective Bayesian optimization for engineering simulation. In: Bartz-Beielstein T, Filipič B, Korošec P, Talbi E, editors. High-performance simulation-based optimization. Cham: Springer; 2020. p. 47–68.
IEEE
[1]
J. van der Herten, N. Knudde, I. Couckuyt, and T. Dhaene, “Multi-objective Bayesian optimization for engineering simulation,” in High-performance simulation-based optimization, vol. 833, T. Bartz-Beielstein, B. Filipič, P. Korošec, and E. Talbi, Eds. Cham: Springer, 2020, pp. 47–68.
@incollection{8694486,
  abstract     = {{Rather than optimizing expensive objective functions such as complex engineering simulations directly, Bayesian optimization methodologies fit a surrogate model (typically Kriging or a Gaussian Process) on evaluations of the objective function(s). To determine the next evaluation, an acquisition function is optimized (also referred to as infill criterion or sampling policy) which incorporates the model prediction and uncertainty and balances exploration and exploitation. Therefore, Bayesian optimization methodologies replace a single optimization of the objective function by a sequence of optimization problems: this makes sense as the acquisition function is cheap-to-evaluate whereas the objective is not. Depending on the goal different acquisition functions are available: multi-objective acquisition functions are relatively new and this chapter gives a state-of-the-art overview and illustrates some approaches based on hypervolume improvement. It is shown that the quality of the model is crucial for the performance of Bayesian optimization and illustrate this by using the more flexible Student-t processes as surrogate models.}},
  author       = {{van der Herten, Joachim and Knudde, Nicolas and Couckuyt, Ivo and Dhaene, Tom}},
  booktitle    = {{High-performance simulation-based optimization}},
  editor       = {{Bartz-Beielstein, T and Filipič, B and Korošec, P and Talbi, E}},
  isbn         = {{9783030187637}},
  issn         = {{1860-949X}},
  language     = {{und}},
  pages        = {{47--68}},
  publisher    = {{Springer}},
  series       = {{Studies in Computational Intelligence}},
  title        = {{Multi-objective Bayesian optimization for engineering simulation}},
  url          = {{http://dx.doi.org/10.1007/978-3-030-18764-4_3}},
  volume       = {{833}},
  year         = {{2020}},
}

Altmetric
View in Altmetric