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

DOI10.1007/S11222-017-9734-XzbMATH Open1384.62277DBLPjournals/sac/OverstallMD18arXiv1608.05815OpenAlexW2586750255WikidataQ62899331 ScholiaQ62899331MaRDI QIDQ1702013FDOQ1702013

Authors: Antony M. Overstall, J. M. McGree, C. C. Drovandi 

Publication date: 27 February 2018

Published in: Statistics and Computing (Search for Journal in Brave)

Abstract: The generation of decision-theoretic Bayesian optimal designs is complicated by the significant computational challenge of minimising an analytically intractable expected loss function over a, potentially, high-dimensional design space. A new general approach for approximately finding Bayesian optimal designs is proposed which uses computationally efficient normal-based approximations to posterior summaries to aid in approximating the expected loss. This new approach is demonstrated on illustrative, yet challenging, examples including hierarchical models for blocked experiments, and experimental aims of parameter estimation and model discrimination. Where possible, the results of the proposed methodology are compared, both in terms of performance and computing time, to results from using computationally more expensive, but potentially more accurate, Monte Carlo approximations. Moreover the methodology is also applied to problems where the use of Monte Carlo approximations is computationally infeasible.

Full work available at URL: https://arxiv.org/abs/1608.05815

Recommendations

• A predictive approach to the Bayesian design problem with application to normal regression models
• An induced natural selection heuristic for finding optimal Bayesian experimental designs
• Optimal Bayesian experimental design for linear models
• Towards Bayesian experimental design for nonlinear models that require a large number of sampling times
• scientific article; zbMATH DE number 418918

zbMATH Keywords

parameter estimationhierarchical modelBayesian optimal designloss functionmodel discrimination

Mathematics Subject Classification ID

Bayesian inference (62F15) Optimal statistical designs (62K05)

Cites Work

• The no-U-turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo
• Bayesian data analysis.
• The Bayesian Choice
• The design and analysis of computer experiments.
• Bayesian experimental design: A review
• Finding near-optimal Bayesian experimental designs via genetic algorithms
• Optimization
• Bayes and empirical-Bayes multiplicity adjustment in the variable-selection problem
• Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors' reply
• Title not available (Why is that?)
• Advanced Calculus with Applications in Statistics
• Bayes linear analysis for Bayesian optimal experimental design
• Bayes Linear Statistics
• Title not available (Why is that?)
• Optimal Bayesian Design by Inhomogeneous Markov Chain Simulation
• On a Measure of the Information Provided by an Experiment
• Bayesian-Optimal Design via Interacting Particle Systems
• The Coordinate-Exchange Algorithm for Constructing Exact Optimal Experimental Designs
• Optimal Design via Curve Fitting of Monte Carlo Experiments
• Computational enhancements to Bayesian design of experiments using Gaussian processes
• Title not available (Why is that?)
• Fast estimation of expected information gains for Bayesian experimental designs based on Laplace approximations
• Developments of the total entropy utility function for the dual purpose of model discrimination and parameter estimation in Bayesian design
• An approach for finding fully Bayesian optimal designs using normal-based approximations to loss functions
• A Review of Modern Computational Algorithms for Bayesian Optimal Design

Cited In (14)

• A predictive approach to the Bayesian design problem with application to normal regression models
• Bayesian optimal design for ordinary differential equation models with application in biological science
• Optimal Bayesian design for discriminating between models with intractable likelihoods in epidemiology
• Optimal experimental design: formulations and computations
• Bayesian design of experiments for intractable likelihood models using coupled auxiliary models and multivariate emulation
• Bayesian design for minimizing prediction uncertainty in bivariate spatial responses with applications to air quality monitoring
• Bayesian adaptive \(N\)-of-1 trials for estimating population and individual treatment effects
• An approach for finding fully Bayesian optimal designs using normal-based approximations to loss functions
• Efficient construction of Bayes optimal designs for stochastic process models
• Model-robust Bayesian design through generalised additive models for monitoring submerged shoals
• Approximate Laplace importance sampling for the estimation of expected Shannon information gain in high-dimensional Bayesian design for nonlinear models
• Bayesian sequential design for copula models
• A Laplace-based algorithm for Bayesian adaptive design
• Optimal Bayesian design for model discrimination via classification

Uses Software

• R
• BayesDA
• acebayes
• NUTS

This page was built for publication: An approach for finding fully Bayesian optimal designs using normal-based approximations to loss functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1702013)