Abstract
Motivated by the parameter identification problem of a reaction-diffusion transport model in a vapor phase infiltration processes, we propose a Bayesian optimization procedure for solving the inverse problem that aims to find an input setting that achieves a desired functional output. The proposed algorithm improves over the standard single-objective Bayesian optimization by (i) utilizing the generalized chi-square distribution as a more appropriate predictive distribution for the squared distance objective function in the inverse problems, and (ii) applying functional principal component analysis to reduce the dimensionality of the functional response data, which allows for efficient approximation of the predictive distribution and the subsequent computation of the expected improvement acquisition function.
Similar content being viewed by others
References
Alvarez MA, Rosasco L, Lawrence ND (2012) Kernels for vector-valued functions: a review. Found Trends® Mach Learn 4(3):195–266
Astudillo R, Frazier P (2019) Bayesian optimization of composite functions. In International Conference on Machine Learning, pages 354–363. PMLR
Azpitarte I, Knez M (2018) Vapor phase infiltration: from a bioinspired process to technologic application, a prospective review. MRS Commun 8(3):727–741. https://doi.org/10.1557/mrc.2018.126
Box GE, Hunter WG (1962) A useful method for model-building. Technometrics 4(3):301–318
Dancik GM, Dorman KS (2008) mlegp: statistical analysis for computer models of biological systems using R. Bioinformatics 24(17):1966–1967
Delaigle A, Hall P (2010) Defining probability density for a distribution of random functions. Annal Stat 38(2):1171–1193
Duchesne P, de Micheaux PL (2010) Computing the distribution of quadratic forms: further comparisons between the liu-tang-zhang approximation and exact methods. Comput Stat Data Anal 54:858–862
Frazier PI (2018) Bayesian optimization. INFORMS Tutorials, pages 255–278
Golub GH, Welsch JH (1969) Calculation of gauss quadrature rules. Math Comput 23(106):221–230
Gutmann MU, Corander J (2016) Bayesian optimization for likelihood-free inference of simulator-based statistical models. J Mach Learn Res
Hines AL, Maddox RN (1985) Mass transfer: fundamentals and applications, vol 434. Prentice-Hall Englewood Cliffs, NJ
Imhof J-P (1961) Computing the distribution of quadratic forms in normal variables. Biometrika 48(3/4):419–426
Ingram WF, Jur JS (2019) Properties and applications of vapor infiltration into polymeric substrates. Jom 71(1):238–245. https://doi.org/10.1007/s11837-018-3157-9
Järvenpää M, Gutmann MU, Vehtari A, Marttinen P et al (2018) Gaussian process modelling in approximate bayesian computation to estimate horizontal gene transfer in bacteria. Annal Appl Stat 12(4):2228–2251
Jones DR (2001) A taxonomy of global optimization methods based on response surfaces. J Glob Optim 21(4):345–383
Jones DR, Schonlau M, Welch WJ (1998) Efficient global optimization of expensive black-box functions. J Glob Optim 13(4):455–492
Joseph VR, Yan H (2015) Engineering-driven statistical adjustment and calibration. Technometrics 57(2):257–267
Joseph VR, Gul E, Ba S (2015) Maximum projection designs for computer experiments. Biometrika 102(2):371–380
Kennedy MC, O’Hagan A (2001) Bayesian calibration of computer models. J R Stat Soc Series B 63(3):425–464
Kirsch A (2011) An introduction to the mathematical theory of inverse problems, vol 120. Springer Science & Business Media, Berlin
Kushner HJ (1964) A new method of locating the maximum point of an arbitrary multipeak curve in the presence of noise. J Basic Eng 86:97–106
Leng CZ, Losego MD (2017) Vapor phase infiltration (vpi) for transforming polymers into organic-inorganic hybrid materials: a critical review of current progress and future challenges. Mater Horiz 4:747–771. https://doi.org/10.1039/C7MH00196G
Liu H, Tang Y, Zhang HH (2009) A new chi-square approximation to the distribution of non-negative definite quadratic forms in non-central normal variables. Comput Stat Data Anal 53(4):853–856
Loeppky JL, Sacks J, Welch WJ (2009) Choosing the sample size of a computer experiment: A practical guide. Technometrics 51(4):366–376
Losego MD, Peng Q (2019) Atomic Layer Deposition and Vapor Phase Infiltration, chapter 5, pages 135–159. John Wiley & Sons, Ltd. ISBN 9783527819249. https://doi.org/10.1002/9783527819249.ch5
Mak S, Joseph VR (2018) Support points. Annal Stat 46(6A):2562–2592
Marttinen P, Croucher NJ, Gutmann MU, Corander J, Hanage WP (2015) Recombination produces coherent bacterial species clusters in both core and accessory genomes. Microb Genomics. https://doi.org/10.1099/mgen.0.000038
Matsui K, Kusakawa S, Ando K, Kutsukake K, Ujihara T, Takeuchi I (2019) Bayesian active learning for structured output design. arXiv preprint arXiv:1911.03671
Mockus J (1994) Application of bayesian approach to numerical methods of global and stochastic optimization. J Glob Optim 4(4):347–365
Nelder JA, Mead R (1965) A simplex method for function minimization. Comput J 7(4):308–313. https://doi.org/10.1093/comjnl/7.4.308 (01, ISSN 0010-4620.)
Ramsay JO, Silverman BW (2004) Functional data analysis. Springer, New York
Ramsay JO, Graves S, Hooker D (2020) fda: Functional Data Analysis. URL https://CRAN.R-project.org/package=fda. R package version 5.1.9
Ren Y, McGuinness EK, Huang C, Joseph VR, Lively RP, Losego MD (2021) Reaction-diffusion transport model to predict precursor uptake and spatial distribution in vapor-phase infiltration processes. Chem Mater 33(13):5210–5222. https://doi.org/10.1021/acs.chemmater.1c01283
Rusmassen C, Williams C (2006) Gaussian process for machine learning. MIT Press, Cambridge
Santner TJ, Williams BJ, Notz WI (2018) The design and analysis of computer experiments. Springer, New York
Subramanian A, Tiwale N, Nam C-Y (2019) Review of recent advances in applications of vapor-phase material infiltration based on atomic layer deposition. JOM 71(1):185–196. https://doi.org/10.1007/s11837-018-3141-4
Tan MH (2018) Gaussian process modeling of a functional output with information from boundary and initial conditions and analytical approximations. Technometrics 60(2):209–221
Uhrenholt AK, Jensen BS (2019) Efficient bayesian optimization for target vector estimation. In The 22nd International Conference on Artificial Intelligence and Statistics, pages 2661–2670. PMLR
Waldman RZ, Mandia DJ, Yanguas-Gil A, Martinson ABF, Elam JW, Darling SB (2019) The chemical physics of sequential infiltration synthesis-a thermodynamic and kinetic perspective. J Chem Phys. https://doi.org/10.1063/1.5128108
Wood SN (2010) Statistical inference for noisy nonlinear ecological dynamic systems. Nature 466(7310):1102–1104
Xiong Y, Chen W, Apley D, Ding X (2007) A non-stationary covariance-based kriging method for metamodelling in engineering design. Int J Num Methods Eng 71(6):733–756
Xun X, Cao J, Mallick B, Maity A, Carroll RJ (2013) Parameter estimation of partial differential equation models. J Am Stat Assoc 108(503):1009–1020
Zhao Y, Huo X, Mei Y (2021) Identification of underlying dynamic system from noisy data with splines. arXiv preprint arXiv:2103.10231
Acknowledgements
This material is based upon work supported by the National Science Foundation (DMREF-1921873). EKM was also supported by the Department of Defense (DoD) through the National Defense Science & Engineering Graduate Fellowship (NDSEG) Program.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Huang, C., Ren, Y., McGuinness, E.K. et al. Bayesian optimization of functional output in inverse problems. Optim Eng 22, 2553–2574 (2021). https://doi.org/10.1007/s11081-021-09677-1
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11081-021-09677-1
Keywords
- Expected improvement
- Gaussian process
- Generalized chi-square distribution
- Model calibration
- Functional principal component analysis