A reminder of next week's meeting:

              THE CAMBRIDGE STATISTICS DISCUSSION GROUP



                  Wednesday 14th March 2007 7:15 for 7:45
                   (PLEASE NOTE CHANGE TO ORIGINAL DATE)



                 Small (Tom Ap Rees) Lecture Theatre,
                    Department of Plant Sciences,
                          Downing Site,
                            Cambridge






   Retrieving Information using a Bayesian Model of Generalization





                        Zoubin Ghahramani
                       Dept of Engineering
                     University of Cambridge



Abstract: 
Humans readily learn new concepts after observing a few examples and
show extremely good generalization to new instances. In contrast,
search tools on the internet exhibit little or no learning and
generalization.  Here we present a  new framework for retrieving 
information based on principles governing how humans learn new concepts and 
generalize. Given a query consisting of a set of items representing some 
concept, our method automatically infers which other items are relevant to 
that concept and retrieves them. Unlike previous such tools, which are 
based on simple text queries, our method leverages the rich and subtle
information provided by a query consisting of a set of items. Moreover our 
work shows  that retrieval can be firmly grounded in a Bayesian statistical 
model of human learning and generalization. Finally, the underlying 
computations reduce to an extremely efficient sparse linear equation, 
making it practical for large scale retrieval problems.  We show five example
applications including searches for scientific articles, images, proteins, 
and movies.
(Joint work with Katherine A Heller, University College London)


Speaker: 
Zoubin Ghahramani is Professor of Information Engineering at the University
of Cambridge, UK. He obtained BA and BSE degrees from the University of
Pennsylvania, and a PhD in 1995 from MIT working with Prof Mike Jordan. He
was a Postdoctoral Fellow in Computer Science at the University of Toronto
working with Prof Geoff Hinton.

His work has included research on human sensorimotor control, cognitive
science, statistics, and machine learning. His current focus is on Bayesian
approaches to statistical machine learning, with applications to
bioinformatics, information retrieval, and other areas. He has published
over 100 peer reviewed papers, and serves on the editorial boards of several
leading journals in the field, including JMLR, JAIR, IEEE PAMI, Machine
Learning, and Bayesian Analysis. He also serves on the Board of the
International Machine Learning Society.

Zoubin Ghahramani is also Associate Research Professor of Machine Learning
at Carnegie Mellon University.




Directions: 
The Department of Plant Sciences is located on the Downing site situated 
between Downing Street and Tennis Court Road. Go through the second door you 
come to in the building immediately on your left as you enter from Tennis 
Court Road. Once inside the building follow the arrows. 
It is not possible to park on the Downing site in the evening. The nearest 
car park is at Lion Yard. Arrivals after 7:45pm can gain admittance by 
contacting the secretary on 07761769436.




Next Meetings:
4th April - Sue Welham (Rothamsted Research) on 
'An Introduction to Mixed Model Splines'.
3rd May - Brian Tom (MRC Biostatistics). 
9th October - Matt Whiley (Amgen).


Supper: Some members eat regularly in the University Centre before
each talk meeting in the downstairs bar at 5-45pm. Feel free to join them.

Subscriptions:  of 1 pound are now due for attending the 2006-2007 session.

Secretary: Peter Watson, MRC Cognition and Brain Sciences Unit, 
15 Chaucer Road, Cambridge CB2 7EF;telephone 01223 355294 Extension 801; 
E-mail [log in to unmask]