Jack Fewtrell
Flexible Recalibration of Approximate Bayesian Models
“The main aim of this research is to develop more flexible transformations for use with BSC (Bayesian Score Calibration). We will consider parametric and non-parametric functions, for example polynomials and Gaussian processes, respectively. Our flexible transforms should result in more accurate posterior approximations for the model of interest, still only relying on a small number of complex model simulations. Furthermore, our approach should be more effective at recalibrating approximate posteriors over a wider region of the parameter space. There are many motivating applications which use a surrogate model to form an approximate posterior.”
Jack Fewtrell
Queensland University of Technology
Jack Fewtrell is a final-year Bachelor of Mathematics and Information Technology student majoring in Statistics and Computer Science at Queensland University of Technology. He plans to pursue a career in Mathematics and Statistics by continuing into research and academia. His current interests lie in Bayesian statistics and Statistical Inference. Jack enjoys the problem-solving challenges from a Mathematics degree and is passionate about applying Statistics in real world systems.
You may be interested in
Jamie Hill
Frequency-domain model selection for large-scale time series via Bayesian variable selection
