Marina Meila Visiting Scientist mmp - AT - csail Office 32-G478 Phone: 617-258-7621 Home page at UW

My interests are in machine learning and probabilistic modelling and their connections to mathematical optimization. Some of the topics that interest me currently are

• how to embed a directed graph in the plane? Does the problem even make sense, and if not, can one turn it into a problem that does?
• connections between compressed sensing and machine learning: can we replace experiment design (before the measurements) with machine learning (after the measurements are taken) and to what extent? This is the core question in a project of gravimetric inversion that I am involved in.
• probability distributions over rankings and partial orderings. How can we estimate them efficiently, and what do their parameters express? Can we use such knowledge to improve what a ranking machine offers a user?
• Foundations of clustering: Can we prove something about the output of a clustering algorithm?

• I was a graduate student in EECS 1992-1999 and spent several years in the former MIT AI Lab. My PhD thesis is "Learning with mixtures of trees"

• Visiting scientist CSAIL November 2003

Marina Meila

University of Washington

Department of Statistics

Box 354322

Seatlle WA 98195-4322

Phone:(206)543-8484