|
|
Royal Statistical Society |
| 1996 | |
|---|---|
|
5 p.m., 11 December |
XIAO-LI MENG (University of Chicago) DAVID van DYK (Harvard University) The EM algorithm - an old folk song sung to a fast new tune This paper investigates the construction of EM-type algorithms that are simple stable and fast. We introduce the idea of a "working parameter" to facilitate the search for efficient data-augmentation schemes, and formulate a general alternating expectation/conditional maximisation (AECM) algorithm that allows data-augmentation schemes to vary with iterations. Appears (with discussion) in Journal of the Royal Statistical Society, Series B, 59(4), 511-567 (1997). |
| 1997 | |
|
5 p.m., 15 January |
SYLVIA RICHARDSON (INSERM, France) PETER GREEN (University of Bristol) On Bayesian analysis of mixtures with an unknown number of components We revisit the classic problem of finite mixture estimation with number of components unknown, proposing fully Bayesian modelling and analysis. Key ingredients - hierarchical modelling using weakly informative priors, MCMC methodology handling a variable-dimension parameter, and extraction of information from a posterior sample of functions - have relevance well beyond mixture analysis. Appears (with discussion) in Journal of the Royal Statistical Society, Series B, 59(4), 731-792 (1997). |
|
2 p.m., 14 May |
Design and analysis of COMPLEX SAMPLE SURVEYS (half day meeting) For all details, see Web page. |