The Mondrian Process

Speaker:	Dr. Yee Whye TEH
		Gatsby Computational Neuroscience Unit
		University College London

Title:		"The Mondrian Process"

Date:		Friday, 16 Jan 2009

Time:		4:00pm - 5:00pm

Venue:		Lecture Theater G
		(Chow Tak Sin Lecture Theater, near lifts 25/26)
		HKUST

Abstract:

We describe a novel stochastic process that can be used to construct a
multidimensional generalization of Sethuraman's stick-breaking
construction. These Mondrian processes operate by starting with a
multidimensional space, and recursively partitioning up the space using
axis-aligned cuts. We describe how Mondrian processes can be applied to
relational data modeling using the Aldous-Hoover representation for
infinitely exchangeable arrays.


********************
Biography:

Yee Whye Teh studied Computer Science and Mathematics at the University of
Waterloo, obtaining a B.Math. in 1997. He embarked on graduate studies at
the University of Toronto under the tutelage of Geoffrey Hinton. Between
1999 and 2001 he spent two years in London England at the Gatsby
Computational Neuroscience Unit where Hinton was the founding director. In
2001 they both returned to Toronto and he finished his Ph.D. in 2003. He
started a postdoc with Michael Jordan at UC Berkeley, finishing in 2004,
and started a stint as a Lee Kuan Yew postdoctoral fellow in August 2005.
Since January 2007, he has been a lecturer back at the Gatsby Unit.