CMStatistics 2017: Start Registration
View Submission - CMStatistics
B0173
Title: Bayesian optimal design of experiments: Review, challenges and examples Authors:  David Woods - University of Southampton (United Kingdom) [presenting]
Abstract: The design of any experiment is implicitly Bayesian, with prior knowledge being used informally to aid decisions such as which factors to vary and the choice of plausible causal relationships between the factors and measured responses. Adoption of formal Bayesian methods allow uncertainty in these decisions to be incorporated into design selection through prior distributions that encapsulate information available from scientific knowledge or previous experimentation. Further, a design may be explicitly tailored to the aim of the experiment through a decision-theoretic approach with an appropriate loss function. However, finding decision-theoretic optimal designs is challenging, largely due to the typically high-dimensional and intractable integration required to evaluate the expected loss. We review some of the recent research in this area, expand on the some of the challenges in Bayesian design, and present some example solutions.