Overview
Topic status: We're looking for students to study this topic.
Classical statistical design involves the choice of treatment level combinations in experiments. More recent applications include determining the times to observe the progress of a plant disease epidemic; times and at which drugs should be administered; and doses to give to subjects so that toxic levels can be determined. Bayesian design incorporates prior knowledge of model unknowns and objective functions which reflect experimenters. This project will develop Bayesian design by using recent computational approaches such as sequential Monte Carlo (SMC). The SMC approach has the advantage that it avoids some of the problems involving Markov chain Monte Carlo, that is slow burn-in and convergence, and incomplete exploration of the design space. The Bayesian algorithms will also be adapted from static to sequential design problems.
- Study level
- PhD
- Supervisors
- QUT
- Organisational unit
Science and Engineering Faculty
- Research area
- Contact
-
Please contact the supervisor.
Professor Tony Pettitt
