Michael Jordan: Applied Bayesian Nonparametrics

Professor Michael Jordan

EECS Berkeley

4.30 pm, Thursday, 4 March 2010

Room G07, The Informatics Forum

10 Crichton Street

Computer Science has historically been strong on data structures and weak on inference from data, whereas Statistics has historically been weak on data structures and strong on inference from data. One way to draw on the strengths of both disciplines is to pursue the study of “inferential methods for data structures''; i.e., methods that update probability distributions on recursively-defined objects such as trees, graphs, grammars and function calls. This is accommodated in the world of ''Bayesian nonparametrics,'' where prior and posterior distributions are allowed to be general stochastic processes. Both statistical and computational considerations lead one to certain classes of stochastic processes, and these tend to have interesting connections to combinatorics.

I will focus on Bayesian nonparametric modeling based on Dirichlet processes and completely random processes, giving examples of how recursions based on these processes lead to useful models in several applied problem domains, including natural language parsing, computational vision, statistical genetics and protein structural modelling.

Sharon Goldwater

We are delighted to announce that Sharon Goldwater has accepted a lectureship in Informatics at Edinburgh, with effect from October.

Sharon graduated from Brown University in Providence, RI, in May '98 with an Sc.B. in mathematics- computer science and a strong interest in linguistics. From 1998-2000, she worked as a researcher in the Artificial Intelligence Laboratory at Stanford Research International (SRI), where she developed telephone-based and multi-modal dialogue systems.

Sharon then returned to Brown, where she received her Sc.M. (2005) in Computer Science, and Ph.D. (2006). Her thesis, supervised by Mark Johnson in the Department of Cognitive and Linguistic Sciences, developed non-parametric Bayesian models for unsupervised learning of linguistic structure.

In 2006 she joined the Stanford natural language processing group as a visiting post-doctoral scholar. There she has continued her work on unsupervised language learning and cognitive modelling, as well as investigating the effects of prosody on speech recognizer errors.

Sharon's current research interests include unsupervised learning, computational modelling of human language acquisition (especially phonology and morphology), and Bayesian models of language.