BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian Clustering with the Dirichlet-Process Prior - Dr. Audrey 
 Q Fu\, Department of Physiology\, Development and Neuroscience
DTSTART:20110601T153000Z
DTEND:20110601T163000Z
UID:TALK31080@talks.cam.ac.uk
CONTACT:Richard Samworth
DESCRIPTION:We consider the problem of clustering measurements collected f
 rom\nmultiple replicates at multiple time points\, with an unknown number 
 of clusters.\nWe propose a mixture random-effects model coupled with a Dir
 ichlet-process prior.\nThe mixture model formulation allows for probabilis
 tic cluster assignments. The\nrandom-effects formulation enables decomposi
 tion of total variability in the data\ninto variabilities that are consist
 ent with the experimental design. The\nDirichlet-process prior induces a p
 rior distribution on partitions and helps to\nestimate the number of clust
 ers (or mixture components) from the data. We also\ntackle two challenges 
 associated with Dirichlet-process prior-based methods. One\nis efficient s
 ampling\, for which we develop a novel Metropolis-Hastings Markov\nChain M
 onte Carlo (MCMC) procedure. The other is efficient use of the MCMC\nsampl
 es in forming clusters\, for which we propose a two-step procedure for\npo
 sterior inference\, which involves resampling and relabeling to estimate t
 he\nposterior allocation probability matrix. The effectiveness of this mod
 el and\nsampling procedure is demonstrated on simulated data. We use this 
 method to\nanalyze time-course gene expression data from Drosophila cells 
 to characterize\nthe genome-wide temporal responses to Notch activation.\n
 \nFraley\, C and Rafery\, AE (2002) Model-based clustering\, discriminant 
 analysis\,\nand density estimation. JASA 97\, 611-631.\n\nHeard\, N et al.
  (2006) A quantitative\nstudy of gene regulation involved in the immune re
 sponse of Anopheline\nmosquitoes: an application of Bayesian hierarchical 
 clustering of curves. JASA\n101\, 18-29.\n\nNeal\, RM (2000) Markov chain 
 sampling methods for Dirichlet process\nmixture models. J. Comput. Graph. 
 Stat. 9\, 249-265.
LOCATION:MR11\, CMS
END:VEVENT
END:VCALENDAR