Meetings on mixtures

While preparing a future meeting on mixtures, I (surprisingly) found a still active link to the very first meeting I organised. It took place in the CNRS vacation center of Aussois (French Alps) in 1995 and it already focussed on mixtures. It was a very exciting meeting with some great talks, different perspectives and schools, and lasting connections. I also took part later in organising with Mike Titterington a smaller scale meeting on mixtures held at the International Centre for Mathematical Sciences (ICMS) in March 2001, where the small size of the audience allowed for open (and sometimes heated!) discussions.

The future meeting in 2010 will be held at ICMS, in Edinburgh, in March 2010. It is organised jointly with Kerrie Mengersen and Mike Titterington, and it will be of the same type as the previous one, with a limited and restricted audience made of invitees, and a small number of long talks followed by discussions and exchanges. However, anyone who would desperately like to attend, in particular PhD students from the UK with a topic related to mixtures, should feel free to contact me. (The purpose of restricting attendance is of course to preserve the congenial and interactive feeling of small meetings, which are mostly those where “something” happens!, and to concentrate the talks on a limited number of topics.) The topics involved for the 2010 meeting are

• frequentist and Bayesian advances in mixture inference, especially concerning identifiability and connection with non-parametric and semi-parametric statistics—this includes the specific Bayesian issue of the label-switching phenomenon and proposals for its resolution or dismissal;
• use of mixtures in mis-specified problems, in particular for data that are not independent and identically distributed;
• new theory and methodology for computational issues, identifying advances and bottlenecks, covering associated issues of convergence, mixing and ways of reducing computational expense;
• new latent-structure models, including the use of covariates in connection with components;
• cross-fertilization concerning these and other aspects among those who contribute to the statistical, computer science and other literatures.

The attached picture of the Ring of Steall, facing Ben Nevis, was taken in 2003 during one of the most beautiful day “hikes” I ever did. (Sadly, another hiker fell to his death on the same ridge the very same day.)