Abstract:We address the computational efficiency in solving the A-optimal Bayesian design of experiments problems for which the observational map is based on partial differential equations and, consequently, is computationally expensive to evaluate. A-optimality is a widely used and easy-to-interpret criterion for Bayesian experimental design. This criterion seeks the optimal experimental design by minimizing the expected conditional variance, which is also known as the expected posterior variance. This study presents a novel likelihood-free approach to the A-optimal experimental design that does not require sampling or integrating the Bayesian posterior distribution. The expected conditional variance is obtained via the variance of the conditional expectation using the law of total variance, and we take advantage of the orthogonal projection property to approximate the conditional expectation. We derive an asymptotic error estimation for the proposed estimator of the expected conditional variance and show that the intractability of the posterior distribution does not affect the performance of our approach. We use an artificial neural network (ANN) to approximate the nonlinear conditional expectation in the implementation of our method. We then extend our approach for dealing with the case that the domain of experimental design parameters is continuous by integrating the training process of the ANN into minimizing the expected conditional variance. Through numerical experiments, we demonstrate that our method greatly reduces the number of observation model evaluations compared with widely used importance sampling-based approaches. This reduction is crucial, considering the high computational cost of the observational models. Code is available at <a class="link-external link-https" href="https://github.com/vinh-tr-hoang/DOEviaPACE" rel="external noopener nofollow">this https URL</a>.

An approach for finding fully Bayesian optimal designs using normal-based approximations to loss functions

Scalable method for Bayesian experimental design without integrating over posterior distribution

Optimal Bayesian design for model discrimination via classification

Laplace-based strategies for Bayesian optimal experimental design with nuisance uncertainty

Bayesian design of experiments for generalised linear models and dimensional analysis with industrial and scientific application

Efficient computation of Bayesian optimal discriminating designs

An Induced Natural Selection Heuristic for Finding Optimal Bayesian Experimental Designs

Bayesian optimal blocking of factorial designs

Bayesian and minimax estimators of loss

Optimal Bayesian stepped-wedge cluster randomised trial designs for binary outcome data

Computing optimal experimental designs with respect to a compound Bayes risk criterion

Efficient construction of Bayes optimal designs for stochastic process models

A Bayesian Risk Approach to Data-driven Stochastic Optimization: Formulations and Asymptotics

Optimal designs for frequentist model averaging

A New Bayesian Approach to Multi-Response Surface Optimization Integrating Loss Function with Posterior Probability

Likelihood-free Simulation-based Optimal Design

A reinforced learning approach to optimal design under model uncertainty

Bayesian Optimal Experimental Design for Inferring Causal Structure

Efficient Real-world Testing of Causal Decision Making via Bayesian Experimental Design for Contextual Optimisation

Optimal design of large-scale Bayesian linear inverse problems under reducible model uncertainty: good to know what you don't know

Metrics for Bayesian Optimal Experiment Design under Model Misspecification