Collaborative and Adaptive Bayesian Optimization for bounding variances and probabilities under hybrid uncertainties

Publikation: Beitrag in FachzeitschriftArtikelForschungPeer-Review

Autoren

  • Fangqi Hong
  • Pengfei Wei
  • Jingwen Song
  • Marcos A. Valdebenito
  • Matthias G.R. Faes
  • Michael Beer

Externe Organisationen

  • Northwestern Polytechnical University
  • Technische Universität Dortmund
  • The University of Liverpool
  • International Joint Research Center for Engineering Reliability and Stochastic Mechanics
  • Tongji University
Forschungs-netzwerk anzeigen

Details

OriginalspracheEnglisch
Aufsatznummer116410
FachzeitschriftComputer Methods in Applied Mechanics and Engineering
Jahrgang417
Frühes Online-Datum6 Sept. 2023
PublikationsstatusVeröffentlicht - 1 Dez. 2023

Abstract

Uncertainty quantification (UQ) has been widely recognized as of vital importance for reliability-oriented analysis and design of engineering structures, and three groups of mathematical models, i.e., the probability models, the imprecise probability models and the non-probabilistic models, have been developed for characterizing uncertainties of different forms. The propagation of these three groups of models through expensive-to-evaluate simulators to quantify the uncertainties of outputs is then one of the core, yet highly challenging task in reliability engineering, as it involves a demanding double-loop numerical dilemma. For addressing this issue, the Collaborative and Adaptive Bayesian Optimization (CABO) has been developed in our previous work, but it only applies to imprecise probability models and is only capable of bounding the output expectation. We present a substantial improvement of CABO to incorporate all three categories of uncertainty models and to bound arbitrary probabilistic measures such as output variance and failure probability. The algorithm is based on a collaborative active learning mechanism, that is, jointly performing Bayesian optimization in the epistemic uncertainty subspace and Bayesian cubature in the aleatory uncertainty subspace, thus allowing to adaptively produce training samples in the joint uncertainty space. An efficient conditional Gaussian process simulation algorithm is embedded in CABO for acquiring training points and Bayesian inference in both uncertain subspaces. Benchmark studies show that CABO exhibits a remarkable performance in terms of numerical efficiency, accuracy, and global convergence.

ASJC Scopus Sachgebiete

Zitieren

Collaborative and Adaptive Bayesian Optimization for bounding variances and probabilities under hybrid uncertainties. / Hong, Fangqi; Wei, Pengfei; Song, Jingwen et al.
in: Computer Methods in Applied Mechanics and Engineering, Jahrgang 417, 116410, 01.12.2023.

Publikation: Beitrag in FachzeitschriftArtikelForschungPeer-Review

Hong F, Wei P, Song J, Valdebenito MA, Faes MGR, Beer M. Collaborative and Adaptive Bayesian Optimization for bounding variances and probabilities under hybrid uncertainties. Computer Methods in Applied Mechanics and Engineering. 2023 Dez 1;417:116410. Epub 2023 Sep 6. doi: 10.1016/j.cma.2023.116410
Download
@article{b45ba88c03b3469fb04c4a1621fdb53c,
title = "Collaborative and Adaptive Bayesian Optimization for bounding variances and probabilities under hybrid uncertainties",
abstract = "Uncertainty quantification (UQ) has been widely recognized as of vital importance for reliability-oriented analysis and design of engineering structures, and three groups of mathematical models, i.e., the probability models, the imprecise probability models and the non-probabilistic models, have been developed for characterizing uncertainties of different forms. The propagation of these three groups of models through expensive-to-evaluate simulators to quantify the uncertainties of outputs is then one of the core, yet highly challenging task in reliability engineering, as it involves a demanding double-loop numerical dilemma. For addressing this issue, the Collaborative and Adaptive Bayesian Optimization (CABO) has been developed in our previous work, but it only applies to imprecise probability models and is only capable of bounding the output expectation. We present a substantial improvement of CABO to incorporate all three categories of uncertainty models and to bound arbitrary probabilistic measures such as output variance and failure probability. The algorithm is based on a collaborative active learning mechanism, that is, jointly performing Bayesian optimization in the epistemic uncertainty subspace and Bayesian cubature in the aleatory uncertainty subspace, thus allowing to adaptively produce training samples in the joint uncertainty space. An efficient conditional Gaussian process simulation algorithm is embedded in CABO for acquiring training points and Bayesian inference in both uncertain subspaces. Benchmark studies show that CABO exhibits a remarkable performance in terms of numerical efficiency, accuracy, and global convergence.",
keywords = "Bayesian optimization, Imprecise probabilities, Interval analysis, Machine learning, Non-probabilistic model, Uncertainty quantification",
author = "Fangqi Hong and Pengfei Wei and Jingwen Song and Valdebenito, {Marcos A.} and Faes, {Matthias G.R.} and Michael Beer",
note = "Funding Information: This work is supported by the National Natural Science Foundation of China under grant number 72171194 and 12202358 , as well as the Sino-German Mobility Programme under grant number M-0175 (2021–2023) .",
year = "2023",
month = dec,
day = "1",
doi = "10.1016/j.cma.2023.116410",
language = "English",
volume = "417",
journal = "Computer Methods in Applied Mechanics and Engineering",
issn = "0045-7825",
publisher = "Elsevier",

}

Download

TY - JOUR

T1 - Collaborative and Adaptive Bayesian Optimization for bounding variances and probabilities under hybrid uncertainties

AU - Hong, Fangqi

AU - Wei, Pengfei

AU - Song, Jingwen

AU - Valdebenito, Marcos A.

AU - Faes, Matthias G.R.

AU - Beer, Michael

N1 - Funding Information: This work is supported by the National Natural Science Foundation of China under grant number 72171194 and 12202358 , as well as the Sino-German Mobility Programme under grant number M-0175 (2021–2023) .

PY - 2023/12/1

Y1 - 2023/12/1

N2 - Uncertainty quantification (UQ) has been widely recognized as of vital importance for reliability-oriented analysis and design of engineering structures, and three groups of mathematical models, i.e., the probability models, the imprecise probability models and the non-probabilistic models, have been developed for characterizing uncertainties of different forms. The propagation of these three groups of models through expensive-to-evaluate simulators to quantify the uncertainties of outputs is then one of the core, yet highly challenging task in reliability engineering, as it involves a demanding double-loop numerical dilemma. For addressing this issue, the Collaborative and Adaptive Bayesian Optimization (CABO) has been developed in our previous work, but it only applies to imprecise probability models and is only capable of bounding the output expectation. We present a substantial improvement of CABO to incorporate all three categories of uncertainty models and to bound arbitrary probabilistic measures such as output variance and failure probability. The algorithm is based on a collaborative active learning mechanism, that is, jointly performing Bayesian optimization in the epistemic uncertainty subspace and Bayesian cubature in the aleatory uncertainty subspace, thus allowing to adaptively produce training samples in the joint uncertainty space. An efficient conditional Gaussian process simulation algorithm is embedded in CABO for acquiring training points and Bayesian inference in both uncertain subspaces. Benchmark studies show that CABO exhibits a remarkable performance in terms of numerical efficiency, accuracy, and global convergence.

AB - Uncertainty quantification (UQ) has been widely recognized as of vital importance for reliability-oriented analysis and design of engineering structures, and three groups of mathematical models, i.e., the probability models, the imprecise probability models and the non-probabilistic models, have been developed for characterizing uncertainties of different forms. The propagation of these three groups of models through expensive-to-evaluate simulators to quantify the uncertainties of outputs is then one of the core, yet highly challenging task in reliability engineering, as it involves a demanding double-loop numerical dilemma. For addressing this issue, the Collaborative and Adaptive Bayesian Optimization (CABO) has been developed in our previous work, but it only applies to imprecise probability models and is only capable of bounding the output expectation. We present a substantial improvement of CABO to incorporate all three categories of uncertainty models and to bound arbitrary probabilistic measures such as output variance and failure probability. The algorithm is based on a collaborative active learning mechanism, that is, jointly performing Bayesian optimization in the epistemic uncertainty subspace and Bayesian cubature in the aleatory uncertainty subspace, thus allowing to adaptively produce training samples in the joint uncertainty space. An efficient conditional Gaussian process simulation algorithm is embedded in CABO for acquiring training points and Bayesian inference in both uncertain subspaces. Benchmark studies show that CABO exhibits a remarkable performance in terms of numerical efficiency, accuracy, and global convergence.

KW - Bayesian optimization

KW - Imprecise probabilities

KW - Interval analysis

KW - Machine learning

KW - Non-probabilistic model

KW - Uncertainty quantification

UR - http://www.scopus.com/inward/record.url?scp=85170435509&partnerID=8YFLogxK

U2 - 10.1016/j.cma.2023.116410

DO - 10.1016/j.cma.2023.116410

M3 - Article

AN - SCOPUS:85170435509

VL - 417

JO - Computer Methods in Applied Mechanics and Engineering

JF - Computer Methods in Applied Mechanics and Engineering

SN - 0045-7825

M1 - 116410

ER -

Von denselben Autoren