Learning Arbitrary Quantities of Interest from Expensive Black-Box Functions through Bayesian Sequential Optimal Design

Estimating arbitrary quantities of interest (QoIs) that are non-linear operators of complex, expensive-to-evaluate, black-box functions is a challenging problem due to missing domain knowledge and finite budgets. Bayesian optimal design of experiments (BODE) is a family of methods that identify an optimal design of experiments (DOE) under different contexts, using only in a limited number of function evaluations. Under BODE methods, sequential design of experiments (SDOE) accomplishes this task by selecting an optimal sequence of experiments while using data-driven probabilistic surrogate models instead of the expensive black-box function. Probabilistic predictions from the surrogate model are used to define an information acquisition function (IAF) which quantifies the marginal value contributed or the expected information gained by a hypothetical experiment. The next experiment is selected by maximizing the IAF. A generally applicable IAF is the expected information gain (EIG) about a QoI as captured by the expectation of the Kullback-Leibler divergence between the predictive distribution of the QoI after doing a hypothetical experiment and the current predictive distribution about the same QoI. We model the underlying information source as a fully-Bayesian, non-stationary Gaussian process (FBNSGP), and derive an approximation of the information gain of a hypothetical experiment about an arbitrary QoI conditional on the hyper-parameters The EIG about the same QoI is estimated by sample averages to integrate over the posterior of the hyper-parameters and the potential experimental outcomes. We demonstrate the performance of our method in four numerical examples and a practical engineering problem of steel wire manufacturing. The method is compared to two classic SDOE methods: random sampling and uncertainty sampling.

[1]  D. Ginsbourger,et al.  Kriging is well-suited to parallelize optimization , 2010 .

[2]  R. Baierlein Probability Theory: The Logic of Science , 2004 .

[3]  D. Mackay,et al.  Introduction to Gaussian processes , 1998 .

[4]  Kenji Miki,et al.  Bayesian optimal experimental design for the Shock-tube experiment , 2013 .

[5]  A. Doostan,et al.  Least squares polynomial chaos expansion: A review of sampling strategies , 2017, 1706.07564.

[6]  Warren B. Powell,et al.  A Knowledge-Gradient Policy for Sequential Information Collection , 2008, SIAM J. Control. Optim..

[7]  Hans-Jürgen Reinhardt Analysis of Approximation Methods for Differential and Integral Equations , 1985 .

[8]  A. OHagan,et al.  Bayesian analysis of computer code outputs: A tutorial , 2006, Reliab. Eng. Syst. Saf..

[9]  Philipp Hennig,et al.  Entropy Search for Information-Efficient Global Optimization , 2011, J. Mach. Learn. Res..

[10]  Victor Picheny,et al.  Adaptive Designs of Experiments for Accurate Approximation of a Target Region , 2010 .

[11]  Richard J. Beckman,et al.  A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output From a Computer Code , 2000, Technometrics.

[12]  Thomas J. Santner,et al.  The Design and Analysis of Computer Experiments , 2003, Springer Series in Statistics.

[13]  Holger Dette,et al.  Generalized Latin Hypercube Design for Computer Experiments , 2010, Technometrics.

[14]  Christopher J Paciorek,et al.  Spatial modelling using a new class of nonstationary covariance functions , 2006, Environmetrics.

[15]  J. Sherman,et al.  Adjustment of an Inverse Matrix Corresponding to a Change in One Element of a Given Matrix , 1950 .

[16]  Robert B. Gramacy,et al.  Ja n 20 08 Bayesian Treed Gaussian Process Models with an Application to Computer Modeling , 2009 .

[17]  Christopher J. Paciorek,et al.  Nonstationary Gaussian Processes for Regression and Spatial Modelling , 2003 .

[18]  Nando de Freitas,et al.  A Tutorial on Bayesian Optimization of Expensive Cost Functions, with Application to Active User Modeling and Hierarchical Reinforcement Learning , 2010, ArXiv.

[19]  Nancy Flournoy,et al.  A Clinical Experiment in Bone Marrow Transplantation: Estimating a Percentage Point of a Quantal Response Curve , 1993 .

[20]  Vianney Perchet,et al.  Gaussian Process Optimization with Mutual Information , 2013, ICML.

[21]  G. Schwarz Estimating the Dimension of a Model , 1978 .

[22]  Aki Vehtari,et al.  Bayesian Optimization of Unimodal Functions , 2017 .

[23]  Wolfram Burgard,et al.  Nonstationary Gaussian Process Regression Using Point Estimates of Local Smoothness , 2008, ECML/PKDD.

[24]  X. Huan,et al.  Sequential Bayesian optimal experimental design via approximate dynamic programming , 2016, 1604.08320.

[25]  Radford M. Neal 5 MCMC Using Hamiltonian Dynamics , 2011 .

[26]  Layne T. Watson,et al.  Efficient global optimization algorithm assisted by multiple surrogate techniques , 2012, Journal of Global Optimization.

[27]  David J. C. MacKay,et al.  Information-Based Objective Functions for Active Data Selection , 1992, Neural Computation.

[28]  David B. Dunson,et al.  Bayesian Data Analysis , 2010 .

[29]  Agus Sudjianto,et al.  Relative Entropy Based Method for Probabilistic Sensitivity Analysis in Engineering Design , 2006 .

[30]  Adam D. Bull,et al.  Convergence Rates of Efficient Global Optimization Algorithms , 2011, J. Mach. Learn. Res..

[31]  Ying Ma,et al.  An Adaptive Bayesian Sequential Sampling Approach for Global Metamodeling , 2016 .

[32]  Carl E. Rasmussen,et al.  Infinite Mixtures of Gaussian Process Experts , 2001, NIPS.

[33]  Svetha Venkatesh,et al.  Bayesian functional optimisation with shape prior , 2018, AAAI.

[34]  Peter Challenor,et al.  Emulating computer models with step-discontinuous outputs using Gaussian processes , 2019, 1903.02071.

[35]  D. Lizotte Practical bayesian optimization , 2008 .

[36]  N. Zheng,et al.  Global Optimization of Stochastic Black-Box Systems via Sequential Kriging Meta-Models , 2006, J. Glob. Optim..

[37]  R. Grandhi,et al.  Polynomial Chaos Expansion with Latin Hypercube Sampling for Estimating Response Variability , 2003 .

[38]  Donald R. Jones,et al.  Efficient Global Optimization of Expensive Black-Box Functions , 1998, J. Glob. Optim..

[39]  A. Kennedy,et al.  Hybrid Monte Carlo , 1988 .

[40]  Julien Bect,et al.  User Preferences in Bayesian Multi-objective Optimization: The Expected Weighted Hypervolume Improvement Criterion , 2018, LOD.

[41]  Sankaran Mahadevan,et al.  Sensor placement for calibration of spatially varying model parameters , 2017, J. Comput. Phys..

[42]  J. Oakley Estimating percentiles of uncertain computer code outputs , 2004 .

[43]  Jitesh H. Panchal,et al.  Bayesian Optimal Design of Experiments for Inferring the Statistical Expectation of Expensive Black-Box Functions , 2019, Journal of Mechanical Design.

[44]  Marco Locatelli,et al.  Bayesian Algorithms for One-Dimensional Global Optimization , 1997, J. Glob. Optim..

[45]  Robert B. Gramacy,et al.  Practical Heteroscedastic Gaussian Process Modeling for Large Simulation Experiments , 2016, Journal of Computational and Graphical Statistics.

[46]  Alexis Boukouvalas,et al.  GPflow: A Gaussian Process Library using TensorFlow , 2016, J. Mach. Learn. Res..

[47]  Andrew Gordon Wilson,et al.  Practical Multi-fidelity Bayesian Optimization for Hyperparameter Tuning , 2019, UAI.

[48]  Pol D. Spanos,et al.  Stochastic Finite Element Method: Response Statistics , 1991 .

[49]  S. Kullback,et al.  Information Theory and Statistics , 1959 .

[50]  Yee Whye Teh,et al.  Variational Estimators for Bayesian Optimal Experimental Design , 2019, ArXiv.

[51]  Wolfram Burgard,et al.  Most likely heteroscedastic Gaussian process regression , 2007, ICML '07.

[52]  Ilias Bilionis,et al.  Bayesian Uncertainty Propagation Using Gaussian Processes , 2015 .

[53]  I. Papaioannou,et al.  Numerical methods for the discretization of random fields by means of the Karhunen–Loève expansion , 2014 .

[54]  Nicholas John Gaul,et al.  Modified Bayesian Kriging for noisy response problems and Bayesian confidence-based reliability-based design optimization , 2014 .

[55]  Juho Rousu,et al.  Non-Stationary Gaussian Process Regression with Hamiltonian Monte Carlo , 2015, AISTATS.

[56]  Jin Xu,et al.  Gaussian Processes and Polynomial Chaos Expansion for Regression Problem: Linkage via the RKHS and Comparison via the KL Divergence , 2018, Entropy.

[57]  Thierry Roncalli,et al.  Financial Applications of Gaussian Processes and Bayesian Optimization , 2019, SSRN Electronic Journal.

[58]  Andy J. Keane,et al.  On the Design of Optimization Strategies Based on Global Response Surface Approximation Models , 2005, J. Glob. Optim..

[59]  X. Huan,et al.  GRADIENT-BASED STOCHASTIC OPTIMIZATION METHODS IN BAYESIAN EXPERIMENTAL DESIGN , 2012, 1212.2228.

[60]  Mark J. Schervish,et al.  Nonstationary Covariance Functions for Gaussian Process Regression , 2003, NIPS.

[61]  E. Vázquez,et al.  Convergence properties of the expected improvement algorithm with fixed mean and covariance functions , 2007, 0712.3744.

[62]  Christopher K. I. Williams,et al.  Gaussian Processes for Machine Learning (Adaptive Computation and Machine Learning) , 2005 .

[63]  Xun Huan,et al.  Simulation-based optimal Bayesian experimental design for nonlinear systems , 2011, J. Comput. Phys..

[64]  Thomas J. Santner,et al.  Design and analysis of computer experiments , 1998 .

[65]  Jasper Snoek,et al.  Practical Bayesian Optimization of Machine Learning Algorithms , 2012, NIPS.

[66]  Yee Whye Teh,et al.  Variational Bayesian Optimal Experimental Design , 2019, NeurIPS.

[67]  Aki Vehtari,et al.  Expectation propagation for nonstationary heteroscedastic Gaussian process regression , 2014, 2014 IEEE International Workshop on Machine Learning for Signal Processing (MLSP).

[68]  Charu C. Aggarwal,et al.  On the Surprising Behavior of Distance Metrics in High Dimensional Spaces , 2001, ICDT.

[69]  M. D. McKay,et al.  A comparison of three methods for selecting values of input variables in the analysis of output from a computer code , 2000 .

[70]  Andreas Krause,et al.  Near-optimal sensor placements in Gaussian processes , 2005, ICML.

[71]  Joshua D. Knowles,et al.  ParEGO: a hybrid algorithm with on-line landscape approximation for expensive multiobjective optimization problems , 2006, IEEE Transactions on Evolutionary Computation.

[72]  Xiao Lin,et al.  Approximate computational approaches for Bayesian sensor placement in high dimensions , 2017, Inf. Fusion.