Computational Methods for Bayesian Model Choice (2)

I have updated the slides for the series of four (advanced graduate) lectures about computational methods for Bayesian model choice, they are here:

and I will keep updating them till next Thursday when the course finishes. When giving my lecture this morning, I realised one thing about bridge sampling, namely that, while the principle of estimating the Bayes factor by

\widehat B_{12} = \frac{1}{n} \sum_{i=1}^n \frac{\tilde\pi_1(\theta_i|x)}{\tilde\pi_2(\theta_i|x)}, \quad  \theta_i\sim\pi_2(\theta),

applies as soon as both parameter spaces are the same, the estimator is potentially disastrous unless the parameterisations of both models are compatible. I mean, if\theta_1in model 1 does not correspond to the same intrinsic quantity as\theta_2in model 2 (for instance the first moments of x), there is no reason for both posteriors to overlap…

In case you happen to be in Edinburgh tomorrow (with nothing better to do), I am also giving a seminar along those same slides at the James Clark Maxwell Building (LT C) of the Kings Buildings Campus of the University of Edinburgh at 4pm.

2 Responses to “Computational Methods for Bayesian Model Choice (2)”

  1. […] of flights require boarding by a bus that may take up to 20 minutes to reach the plane. (I remember taking one plane bound for Edinburgh that was parked so far that both the bus driver and the Air France […]

  2. […] in Montpellier, so no need to re-post the slides, which will be a condensed version of my recent course as […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.