Statistics Seminar (S3RI) - Infinite Mixtures of Infinite Factor Analysers (IMIFA), Claire Gormley (University College Dublin) Seminar
- Time:
- 14:00 - 15:00
- Date:
- 25 October 2018
- Venue:
- Room 7035, Lecture Theatre 7B, Building 54, Mathematical Sciences, University of Southampton, Highfield Campus, SO17 1BJ
For more information regarding this seminar, please email Dr Helen Ogden at H.E.Ogden@southampton.ac.uk .
Event details
Factor-analytic Gaussian mixture models are often employed as a model-based approach to clustering high-dimensional data. Typically, the numbers of clusters and latent factors must be specified in advance of model fitting, and the optimal pair selected using a model choice criterion. For computational reasons, models in which the number of latent factors is common across clusters are generally considered.
Here the infinite mixture of infinite factor analysers (IMIFA) model is introduced. IMIFA employs a Poisson-Dirichlet process prior to facilitate automatic inference on the number of clusters. Further, IMIFA employs shrinkage priors to allow cluster specific numbers of factors, automatically inferred via an adaptive Gibbs sampler. IMIFA is presented as the flagship of a family of factor-analytic mixture models, providing flexible approaches to clustering high-dimensional data.
Applications to benchmark and real data sets illustrate the IMIFA model and its advantageous features: IMIFA obviates the need for model selection criteria, reduces model search and associated computational burden, improves clustering performance by allowing cluster-specific numbers of factors, and quantifies uncertainty in the numbers of clusters and cluster-specific factors.
The IMIFA R package, available on CRAN, facilitates implementation of our method.
This is joint work with Keefe Murphy (University College Dublin) and Cinzia Viroli (Universita di Bologna)
Speaker information
Dr Claire Gormley , University College Dublin. Claire is an Associate Professor in the School of Mathematics and Statistics in University College Dublin where she conducts statistical research and teach statistics to undergraduate and graduate students. Claire is a Funded Investigator in the Insight Centre for Data Analytics, in UCD. Her research involves the development of novel statistical methods: she conceives, develops, implements and disseminates novel statistical methods to solve applied problems. Her statistical research has led to collaborations with a range of other disciplines, including epigenomics, metabolomics, genomics, social science, sports science and political science. She collaborates with statisticians and scientists, nationally and internationally. In 2016 Claire was awarded Chartered Statistician status (CStat) by the Royal Statistical Society and was elected to Council of the Royal Statistical Society in October 2016. Claire is an Associate Editor for the Annals of Applied Statistics.