A new Bayesian paradigm for predition is proposed. Unlike the standard Bayes approach, which is tied to the underlying likelihood function, $\ell(y|\theta)$, this new aproach updates our prior belifs according to user-defined statistical criterion or loss function. Interestingly, we demonstrate how this new approach can generate posteriors over predictive densities, rather than posteriors over parameters.
Slides can be added in a few ways:
slides parameter in the front matter of the talk file
static/ and link using
url_slides parameter in the front matter of the talk file
Further talk details can easily be added to this page using Markdown and $\rm \LaTeX$ math code.