Embedded Files
Professor Michael Goldstein, University of Oxford
The basic questions that we should ask of any real world uncertainty analysis are as follows: what do we mean by the uncertainty statements that we make, what have we done to ensure that our uncertainty statements do have the stated meaning and for what purposes are these uncertainty statements useful. We will discuss these questions from a subjectivist Bayes viewpoint, looking at the strengths and weaknesses of this framework for carrying out such an uncertainty analysis and suggesting ways in which the usual theory may be augmented to sharpen the value and meaning of the statements made. While our discussion will be quite general, it will be largely motivated in the framework of problems of uncertainty quantification for complex physical systems modelled by computer simulators.
Speaker biosketch
Michael Goldstein is a statistician at the University of Durham, who has worked for many years on the foundations, methodology and applications of Bayesian statistics. In particular, he has developed the general approach termed Bayes linear statistics which is similar in spirit to conventional Bayes but takes expectation, rather than probability, as the fundamental primitive for the theory. For the last thirty years, his main area of application for this theory has been to problems of uncertainty quantification and decision making for complex physical systems modelled by computer simulators.
Page updated
Google Sites
Report abuse