In several articles I have read, I found the authors using log-linear
models with priors, usually a Gaussian prior, in order to perform
smoothing or to eliminate the possibility of infinitely large weights.
My question is thus: has anyone using priors with these models done
analyses to show that the resulting posteriors remain log-linear?
UC Santa Cruz
This archive was generated by hypermail 2b29 : Fri Feb 28 2003 - 03:23:11 MET