Applications of Bernstein polynomials in Statistics

Add to your list(s) Download to your calendar using vCal

• Sonia Petrone (University Bocconi Milan)
• Friday 05 November 2010, 16:00-17:00
• MR12, CMS, Wilberforce Road, Cambridge, CB3 0WB.

If you have a question about this talk, please contact Richard Nickl.

Bernstein polynomials are studied in analysis as a probabilistic approximation scheme, and they are well known to provide a constructive proof of Weirstrass theorem. In fact, their simple probabilistic nature make them also attractive in many statistical problems. In this talk I will give an overview of applications of Bernstein polynomials in statistics, in particular in Bayesian nonparametric inference. I will briefly review their use to construct a nonparametric prior, which provides a smoothing of a Dirichlet process, and allows to fairly easily incorporate prior information. I will discuss theoretical properties and show some applications, in particular to shrinkage estimation, comparing with the clustering properties of the Dirichlet process. Bernstein polynomials also arise in quantile estimation from a Dirichlet process. I will finally discuss extensions to data on a general subset of R and to multivariate data.

This talk is part of the Statistics series.

This talk is included in these lists:

• All CMS events
• All Talks (aka the CURE list)
• CMS Events
• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• DPMMS Lists
• DPMMS info aggregator
• DPMMS lists
• Guy Emerson's list
• MR12, CMS, Wilberforce Road, Cambridge, CB3 0WB
• Machine Learning
• School of Physical Sciences
• Statistical Laboratory info aggregator
• Statistics
• Statistics Group
• rp587

Note that ex-directory lists are not shown.