Learning from Multiple Measurements
Speaker: Sam Roweis, New York University
Location: Warren Weaver Hall 202
Date: September 25, 2009, 11:30 a.m.
Host: Dennis Shasha
A common situation arising in everything from scientific experiments to perceptual systems is when several uncharacterized sensors all make measurements of the same unknown signal. Ideally, we would like to simultaneously combine all the measurements into an estimate of true underlying source ("fusion") and also learn the properties of the individual sensors ("identification"). In this talk I'll describe an approach to solving such problem based on learning a probabilistic generative model whose parameters capture both the underlying source being measured and the observational effect of the each sensor. I will show examples of these models successfully performing time alignment of repeated traces, exposure calibration of image brackets, microphone deconvolution for audio recordings and coordinated mapping for mobile robots.
Sam Roweis is an Associate Professor in the Department of Computer Science at the New York University. His research interests are in machine learning, data mining, and statistical signal processing. Roweis did his undergraduate degree at the University of Toronto in the Engineering Science program and earned his doctoral degree in 1999 from the California Institute of Technology working with John Hopfield. He did a postdoc with Geoff Hinton and Zoubin Ghahramani at the Gatsby Unit in London, and was a visiting faculty member at MIT in 2005. He has also worked at several industrial research labs including Google, Bell Labs, Whizbang! Labs and Microsoft. He is the holder of a Canada Research Chair in Statistical Machine Learning, a Sloan Research Fellowship, the winner of a Premier's Research Excellence Award, and a Scholar of the Canadian Institute for Advanced Research.
In-person attendance only available to those with active NYU ID cards. All individuals must show the Daily Screener green pass in order to gain entry to the building.