4 - 6 April 2016
Registration for this event is now closed
This master class on Nonparametric Bayesian Statistics will consist of 4 lectures each from two speakers who are leaders of the field, David Dunson (Duke University) and Harry van Zanten (University of Amsterdam). This will be an excellent opportunity for students, academics, and industry professionals to network and to learn about the exciting field of Nonparametric Bayesian modelling. Scholarship opportunities for students as well as a poster session for students and postdocs will be available.
|Students and post-doctoral researchers||£30|
|Lecturers, professors, and other academics||£65|
A limited number of scholarships is available. The scholarships will cover accommodation for three nights and the registration fees for early career researchers (PhD student or postdoctoral researcher with PhD obtained within last 5 years).
Included in the registration fee are coffee breaks and lunch for each day, and wine and refreshments during the poster session. There will be an additional administrative cost of £ 7.50. There will also be a gala dinner available for attendees for an additional £ 36.75. Accommodations are not included with the fees, but they can be booked during the registration process. Parking space can be arranged free of charge here.
Applied Bayesian nonparametric modeling, by David Dunson:
The focus of these lectures is on providing an applied-driven introduction to Bayesian nonparametrics, focusing initially on “canonical” classes of models, such as Dirichlet process mixtures and Gaussian processes. Several motivating applications will provide concrete context in introducing modeling details and computational algorithms. After providing basic details and illustrations on implementations and properties in basic cases, I will consider several more complex case studies ranging from reproductive epidemiology studies to social science surveys to brain connectomics. You can find a description of the topics, including relevant references, here.
Nonparametric Bayesian methods, by Harry van Zanten:
The ever increasing use of nonparametric Bayesian methods raises all kinds of interesting theoretical, mathematical questions. In my lectures I intend to address a number of these issues. It is by now well known that in Bayesian nonparametrics, priors need to be tuned rather carefully to ensure good performance. An unfortunate choice of the prior can easily lead to sub-optimal, or even misleading procedures. Mathematical theory can help to understand the performance and limitations of nonparametric Bayes procedures. It can shed light on how things like consistency, convergence rates, automatic adaptation, etc., are related to the fine properties of the prior. I will discuss several examples of results with this flavour and give some idea of the type of mathematics that is involved.
David Dunson is the Arts and Sciences Professor of Statistical Science, Mathematics and Electrical & Computer Engineering at Duke University. He has made broad contributions to Bayesian theory and methods motivated by applications to broad fields, with a particular focus on biomedical studies ranging from fertility to genomics to neurosciences. An emphasis has been on latent variable models, nonparametric approaches and geometrically based models.
Harry van Zanten obtained his PhD in mathematics in 2001 at the University of Amsterdam. He has gone through the academic ranks at the Vrije Universiteit in Amsterdam and the Eindhoven University of Technology. Since 2012 he is professor of Mathematical Statistics at the University of Amsterdam.
|Monday 4 April||09.30-10.30||Registration and coffee
|Tuesday 5 April||10.00-10.30||Coffee|
Evening reception and poster session
|Wednesday 6 April||10.00-10.30||Coffee|
Please send queries to the organisers: Mark Fiecas (m dot fiecas at warwick dot ac dot uk) and Joris Bierkens (j dot bierkens at warwick dot ac dot uk).