Cornell workshops on structured Bayes 10/22, 10/23

[John T. Hale]

Dear Fellow Northeast Statistics enthusiasts -- I wanted to alert you straightaway to a pair of workshops coming up in Ithaca. The first session is in about a week and a half; they are open to everyone. The topics center around applications of Structured Bayes to cognitive science.

best,

-john

In cognitive science, there has been an explosion of interest recently in probabilistic modelling.

One important aspect of this work is an emerging set of ideas that could be called "structured Bayesian modelling."

This approach definines probabilistic generative models over richly structured representations such as graphs, trees, grammars and logics. In two hands-on workshops Timothy O'Donnell  (Harvard Psychology) and Noah Goodman (MIT BCS) will introduce some tools from the structured Bayesian toolbox and explore some applications to language from the recent literature including word-segmentation (e.g. the work of Goldwater and colleagues ) and word-learning (such as the work of Frank and colleagues).  The workshop teaches attendees how to specify their own models using "Church", a stochastic functional programming  language derived from Lisp, but with a probabilistic semantics.

The workshops are hosted by the Linguistics Department at Cornell University. They take place October 22/23 and November 12/13 in room B07 of Morrill Hall, at Cornell University in Ithaca NY.  Further schedule details available at   http://ling.cornell.edu/index.cfm/page/events.htm

Funded by a Small Grant from the Cornell Institute for the Social

Sciences and open to the public. 

A Bayesian framework for word segmentation: exploring the effects of context.

Sharon Goldwater, Thomas L. Griffiths, and Mark Johnson. Cognition,

112:1, pages 21-54. 2009.

Using speakers’ referential intentions to model early cross-situational word learning.

Michael C. Frank, Noah D. Goodman and Joshua B. Tenenbaum.

Psychological Science 20:5 pages 578-585. 2009