Uncertainty quantification and propagation in surrogate-based Bayesian inference

被引:0
|
作者
Reiser, Philipp [1 ]
Aguilar, Javier Enrique [1 ,2 ]
Guthke, Anneli [1 ]
Buerkner, Paul-Christian [1 ,2 ]
机构
[1] Univ Stuttgart, Cluster Excellence SimTech, Stuttgart, Germany
[2] TU Dortmund Univ, Dept Stat, Dortmund, Germany
关键词
Surrogate modeling; Uncertainty quantification; Uncertainty propagation; Bayesian inference; MAXIMUM-LIKELIHOOD; R PACKAGE; MODELS; CALIBRATION; SELECTION; NETWORKS;
D O I
10.1007/s11222-025-10597-8
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Surrogate models are statistical or conceptual approximations for more complex simulation models. In this context, it is crucial to propagate the uncertainty induced by limited simulation budget and surrogate approximation error to predictions, inference, and subsequent decision-relevant quantities. However, quantifying and then propagating the uncertainty of surrogates is usually limited to special analytic cases or is otherwise computationally very expensive. In this paper, we propose a framework enabling a scalable, Bayesian approach to surrogate modeling with thorough uncertainty quantification, propagation, and validation. Specifically, we present three methods for Bayesian inference with surrogate models given measurement data. This is a task where the propagation of surrogate uncertainty is especially relevant, because failing to account for it may lead to biased and/or overconfident estimates of the parameters of interest. We showcase our approach in three detailed case studies for linear and nonlinear real-world modeling scenarios. Uncertainty propagation in surrogate models enables more reliable and safe approximation of expensive simulators and will therefore be useful in various fields of applications.
引用
收藏
页数:28
相关论文
共 50 条
  • [31] Uncertainty quantification in mechanical properties for Cu-based SMA wires and strands based on Bayesian inference
    Medina, Christian D.
    Ruiz, Rafael O.
    Herrera, Ricardo A.
    Beltran, Juan F.
    ENGINEERING STRUCTURES, 2024, 318
  • [32] SPUX framework: a scalable package for bayesian uncertainty quantification and propagation
    Šukys, Jonas
    Bacci, Marco
    arXiv, 2021,
  • [33] Bayesian uncertainty quantification and propagation for prediction of milling stability lobe
    Li, Kai
    He, Songping
    Liu, Hongqi
    Mao, Xinyong
    Li, Bin
    Luo, Bo
    MECHANICAL SYSTEMS AND SIGNAL PROCESSING, 2020, 138 (138)
  • [34] Bayesian Inference and Uncertainty Quantification for Medical Image Reconstruction with Poisson Data
    Zhou, Qingping
    Yu, Tengchao
    Zhang, Xiaoqun
    Li, Jinglai
    SIAM JOURNAL ON IMAGING SCIENCES, 2020, 13 (01): : 29 - 52
  • [35] Surrogate-Based Bayesian Inverse Modeling of the Hydrological System: An Adaptive Approach Considering Surrogate Approximation Error
    Zhang, Jiangjiang
    Zheng, Qiang
    Chen, Dingjiang
    Wu, Laosheng
    Zeng, Lingzao
    WATER RESOURCES RESEARCH, 2020, 56 (01)
  • [36] Uncertainty quantification in the mathematical modelling of a suspension strut using BAYESIAN inference
    Mallapur, Shashidhar
    Platz, Roland
    MECHANICAL SYSTEMS AND SIGNAL PROCESSING, 2019, 118 : 158 - 170
  • [37] Scalable Inverse Uncertainty Quantification by Hierarchical Bayesian Modeling and Variational Inference
    Wang, Chen
    Wu, Xu
    Xie, Ziyu
    Kozlowski, Tomasz
    ENERGIES, 2023, 16 (22)
  • [38] Uncertainty quantification for polymer micromilling force models using Bayesian inference
    Shekhar, Shivang
    Ozutemiz, Kadri Bugra
    Onler, Recep
    Nahata, Sudhanshu
    Ozdoganlar, O. Burak
    48TH SME NORTH AMERICAN MANUFACTURING RESEARCH CONFERENCE, NAMRC 48, 2020, 48 : 611 - 618
  • [39] Hybrid analytical surrogate-based process optimization via Bayesian symbolic regression
    Jog, Sachin
    Vazquez, Daniel
    Santos, Lucas F.
    Caballero, Jose A.
    Guillen-Gosalbez, Gonzalo
    COMPUTERS & CHEMICAL ENGINEERING, 2024, 182
  • [40] Assessment of uncertainty propagation within compaction-based exhumation studies using Bayesian inference
    Makuluni, Patrick
    Hauser, Juerg
    Clark, Stuart
    TECTONOPHYSICS, 2025, 900