Abstract
Calculation of the Bayes factor for comparison of computationally expensive models is not a straight forward process. To obtain the model’s evidence, numerical integration methods are often implemented. To apply these methods, many samples should be obtained from the model, requiring the model to be run thousands of times. In this paper, we propose to use a probability distribution function as a surrogate model for the evidence. Considering the second axiom of probability, a modified scale factor used for fitting the surrogate model to the integrand would be an estimate of the model’s evidence, and no further integration is needed after the surrogate model is fitted to the integrand. Two structural models are compared to explore the advantages and disadvantages of the proposed method. The Bayes factor is estimated for both models using Monte Carlo integration and the proposed method. Results show that fewer numbers of samples of the structural models are needed when the proposed method is applied. Therefore, this method reduces the computational cost required to compare models in a probabilistic sense.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsReferences
Vakilzadeh, M.K., Huang, Y., Beck, J.L., Abrahamsson, T.: Approximate Bayesian computation by subset simulation using hierarchical state-space models. Mech. Syst. Signal Process. 84, 2–20 (2017)
Ortiz, A.R., Madarshahian, R., Caicedo, J.M., Rizos, D.: Transfer length probabilistic model updating in high performance concrete. In: Dynamics of Civil Structures, vol. 2, pp. 325–330. Springer, Cham (2016)
Papadimitriou, C.: Bayesian uncertainty quantification and propagation (UQ+ P): state-of-the-art tools for linear and nonlinear structural dynamics models. In: Identification Methods for Structural Health Monitoring, pp. 137–170. Springer, Cham (2016)
Behmanesh, I., Moaveni, B.: Accounting for environmental variability, modeling errors, and parameter estimation uncertainties in structural identification. J. Sound Vib. 374, 92–110 (2016)
Miles, P., Hays, M., Smith, R., Oates, W.: Bayesian uncertainty analysis of finite deformation viscoelasticity. Mech. Mater. 91, 35–49 (2015)
Boateng, L.K., Madarshahian, R., Yoon, Y., Caicedo, J.M., Flora, J.R.: A probabilistic approach for estimating water permeability in pressure-driven membranes. J. Mol. Model. 22(8), 185 (2016)
Kennedy, G.J., Martins, J.R.: A parallel finite-element framework for large-scale gradient-based design optimization of high-performance structures. Finite Elem. Anal. Des. 87, 56–73 (2014)
Oliver, J., Huespe, A., Sanchez, P.: A comparative study on finite elements for capturing strong discontinuities: E-FEM vs X-FEM. Comput. Methods Appl. Mech. Eng. 195(37), 4732–4752 (2006)
Wasserman, L.: Bayesian model selection and model averaging. J. Math. Psychol. 44(1), 92–107 (2000)
Wang, G.G., Shan, S.: Review of metamodeling techniques in support of engineering design optimization. J. Mech. Des. 129(4), 370–380 (2007)
Matott, L.S., Rabideau, A.J.: Calibration of complex subsurface reaction models using a surrogate-model approach. Adv. Water Resour. 31(12), 1697–1707 (2008)
Govers, Y., Khodaparast, H.H., Link, M., Mottershead, J.: A comparison of two stochastic model updating methods using the {DLR} {AIRMOD} test structure. Mech. Syst. Signal Process. 52–53, 105–114 (2015)
Madarshahian, R., Caicedo, J.M.: Metamodeling of Model Evidence, pp. 307–313. Springer, Cham (2016)
Madarshahian, R., Caicedo, J.M.: Reducing MCMC computational cost with a two layered Bayesian approach. In: Model Validation and Uncertainty Quantification, vol. 3, pp. 291–297. Springer, Cham (2015)
Lepage, G.P.: A new algorithm for adaptive multidimensional integration. J. Comput. Phys. 27(2), 192–203 (1978)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 The Society for Experimental Mechanics, Inc.
About this paper
Cite this paper
Madarshahian, R., Caicedo, J.M. (2017). Surrogate-Based Approach to Calculate the Bayes Factor. In: Barthorpe, R., Platz, R., Lopez, I., Moaveni, B., Papadimitriou, C. (eds) Model Validation and Uncertainty Quantification, Volume 3. Conference Proceedings of the Society for Experimental Mechanics Series. Springer, Cham. https://doi.org/10.1007/978-3-319-54858-6_27
Download citation
DOI: https://doi.org/10.1007/978-3-319-54858-6_27
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-54857-9
Online ISBN: 978-3-319-54858-6
eBook Packages: EngineeringEngineering (R0)