Usual time: Thursdays 14:00-15:00
Location: Room 102, Department of Statistical Science, 1-19 Torrington Place (1st floor). Some seminars are held at different locations and at different times. Please click on the abstract for further details.
- 09 May 2019 1400-1500: Prof. Adrian Bowman (University of Glasgow)
Statistics with a human face
Three-dimensional surface imaging, through laser-scanning or stereo-photogrammetry, provides high-resolution data defining the surface shape of objects. Human faces are of particular interest and there are many biological and anatomical applications, including assessing the success of facial surgery and investigating the possible developmental origins of some adult conditions. An initial challenge is to structure the raw images by identifying features of the face. Ridge and valley curves provide a very good intermediate level at which to approach this, as these provide a good compromise between informative representations of shape and simplicity of structure. Some of the issues involved in analysing data of this type will be discussed and illustrated. Modelling issues include simple comparison of groups, the measurement of asymmetry and longitudinal patterns of shape change. This last topic is relevant at short scale in facial animation, medium scale in individual growth patterns, and very long scale in phylogenetic studies.
- 09 May 2019 1530-1630: Dr. Dootika Vats (University of Warwick)
Revisiting the Gelman-Rubin diagnostic
Gelman and Rubin's (1992) convergence diagnostic is one of the most popular methods for terminating a Markov chain Monte Carlo (MCMC) sampler. Since the seminal paper, researchers have developed sophisticated methods of variance estimation for Monte Carlo averages. We show that this class of estimators find immediate use in the Gelman-Rubin statistic, a connection not established in the literature before. We incorporate these estimators to upgrade both the univariate and multivariate Gelman-Rubin statistics, leading to increased stability in MCMC termination time. An immediate advantage is that our new Gelman-Rubin statistic can be calculated for a single chain. In addition, we establish a relationship between the Gelman-Rubin statistic and effective sample size. Leveraging this relationship, we develop a principled cutoff criterion for the Gelman-Rubin statistic. Finally, we demonstrate the utility of our improved diagnostic via an example. This work is joint with Christina Knudson, University of St. Thomas, Minnesota.