Properties of Fisher information gain for Bayesian design of experiments

DOI10.1016/J.JSPI.2021.10.006MaRDI QIDQ2059463zbMATH OpenOpenAlexFDO

Authors Antony M. Overstall

Publication date 14 December 2021

Published in Journal of Statistical Planning and Inference (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2003.07315

zbMATH Keywords

Bayesian design of experimentsnon-identifiabilityparameter redundancyunder-supported designs

Mathematics Subject Classification ID

Statistical aspects of information-theoretic topics (62B10) Bayesian problems; characterization of Bayes procedures (62C10)

Abstract: The Bayesian decision-theoretic approach to design of experiments involves specifying a design (values of all controllable variables) to maximise the expected utility function (expectation with respect to the distribution of responses and parameters). For most common utility functions, the expected utility is rarely available in closed form and requires a computationally expensive approximation which then needs to be maximised over the space of all possible designs. This hinders practical use of the Bayesian approach to find experimental designs. However, recently, a new utility called Fisher information gain has been proposed. The resulting expected Fisher information gain reduces to the prior expectation of the trace of the Fisher information matrix. Since the Fisher information is often available in closed form, this significantly simplifies approximation and subsequent identification of optimal designs. In this paper, it is shown that for exponential family models, maximising the expected Fisher information gain is equivalent to maximising an alternative objective function over a reduced-dimension space, simplifying even further the identification of optimal designs. However, if this function does not have enough global maxima, then designs that maximise the expected Fisher information gain lead to non-identifiablility.

Recommendations

• Fast estimation of expected information gains for Bayesian experimental designs based on Laplace approximations
• Bayesian optimal experimental designs for binary responses in an adaptive framework
• scientific article; zbMATH DE number 418918
• Bayesian information in an experiment and the Fisher information distance
• Multimodal information gain in Bayesian design of experiments

Cites work

• A Review of Modern Computational Algorithms for Bayesian Optimal Design
• Bayesian experimental design: A review
• Bayesian information in an experiment and the Fisher information distance
• Design of experiments. An introduction based on linear models.
• Detecting parameter redundancy
• scientific article; zbMATH DE number 3854294 (Why is no real title available?)
• Identifiability, Improper Priors, and Gibbs Sampling for Generalized Linear Models
• Mean and median bias reduction in generalized linear models
• On a Measure of the Information Provided by an Experiment
• Optimum experimental designs, with SAS
• The Bayesian Choice
• Theory of statistics

Cited in

(5)

• Multimodal information gain in Bayesian design of experiments
• Bayesian information in an experiment and the Fisher information distance
• Bayesian experimental design without posterior calculations: an adversarial approach
• Modern Bayesian experimental design
• Optimal experimental design: formulations and computations

This page was built for publication: Properties of Fisher information gain for Bayesian design of experiments

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2059463)