Abstract:In various industrial contexts, estimating the distribution of unobserved random vectors Xi from some noisy indirect observations H(Xi) + Ui is required. If the relation between Xi and the quantity H(Xi), measured with the error Ui, is implemented by a CPU-consuming computer model H, a major practical difficulty is to perform the statistical inference with a relatively small number of runs of H. Following Fu et al. (2014), a Bayesian statistical framework is considered to make use of possible prior knowledge on the parameters of the distribution of the Xi, which is assumed Gaussian. Moreover, a Markov Chain Monte Carlo (MCMC) algorithm is carried out to estimate their posterior distribution by replacing H by a kriging metamodel build from a limited number of simulated experiments. Two heuristics, involving two different criteria to be optimized, are proposed to sequentially design these computer experiments in the limits of a given computational budget. The first criterion is a Weighted Integrated Mean Square Error (WIMSE). The second one, called Expected Conditional Divergence (ECD), developed in the spirit of the Stepwise Uncertainty Reduction (SUR) criterion, is based on the discrepancy between two consecutive approximations of the target posterior distribution. Several numerical comparisons conducted over a toy example then a motivating real case-study show that such adaptive designs can significantly outperform the classical choice of a maximin Latin Hypercube Design (LHD) of experiments. Dealing with a major concern in hydraulic engineering, a particular emphasis is placed upon the prior elicitation of the case-study, highlighting the overall feasibility of the methodology. Faster convergences and manageability considerations lead to recommend the use of the ECD criterion in practical applications.

Efficient Prediction Designs for Random Fields

An adaptive kriging method for solving nonlinear inverse statistical problems

Adaptive Experiment Design for Probabilistic Integration

Reduced Basis Kriging for Big Spatial Fields

Kriging prediction with isotropic Matérn correlations: Robustness and experimental design

Efficient Kriging Using Interleaved Lattice-Based Designs with Low Fill and High Separation Distance Properties

Sample and Computationally Efficient Stochastic Kriging in High Dimensions

On Prediction Properties of Kriging: Uniform Error Bounds and Robustness

A New Reliability Method Combining Adaptive Kriging and Active Variance Reduction Using Multiple Importance Sampling

G-optimal grid designs for kriging models

Universal Convergence of Kriging

Prediction of spatial functional random processes: Comparing functional and spatio-temporal kriging approaches

An Adaptive Algorithm based on High-Dimensional Function Approximation to obtain Optimal Designs

Universal kernel-type estimation of random fields

Optimization of Expensive Black-Box Problems Via Gradient-enhanced Kriging

Adaptive Spatial Sampling Design for Environmental Field Prediction using Low-Cost Sensing Technologies

Resolution Adaptive Fixed Rank Kriging

Estimation of the Pareto Front in Stochastic Simulation Through Stochastic Kriging.

A Two-Stage Kriging Estimation Variance Reduction Method for Efficient Time-Variant Reliability-Based Design Optimization.

Why not a thin plate spline for spatial models? A comparative study using Bayesian inference

Efficient construction of Bayes optimal designs for stochastic process models