||Sala conferenze- num. 322 - 3 piano
||Advances in Statistical Learning from Near-neighbors
||Maya Gupta, Asst. Professor, EE, University of Washington
||University of Washington, College of Engineering
||We consider the problem of classifying or labeling a test sample
based on a database of known samples and their labels. This is a
standard statistical learning problem often solved by neural nets,
support vector machines, Gaussian mixture models, or decision trees.
In this talk research is presented into methods which learn based on
near-neighbors; simple examples of this local learning approach are
k-nearest neighbor and linear interpolation. Theoretically, we
discuss how estimation bias can be reduced by using a convex
neighborhood of samples, and by using weights that solve a linear
interpolation and maximum entropy objective (LIME). Given weighted
neighbors, we show that Bayesian minimum expected risk estimates
will significantly outperform maximum likelihood estimates for
classification when costs are asymmetric, as is often the case in
medical, defense, and non-destructive evaluation applications.
Applications include protein structure prediction, the
non-destructive evaluation of pipeline integrity, and the automatic
creation of custom color enhancements.
Short Bio: Maya Gupta joined the University of Washington as an
Assistant Professor of Electrical Engineering in 2003. An NSF
Graduate Fellow, she completed the M.S. in EE in 1999 and the Ph.D.
in 2003 at Stanford University, working with Bob Gray and Richard
Olshen. From 1999-2003 she worked for Ricoh's California Research
Center as a color image processing research engineer, and is the
primary author on a number of patents in color image processing. She
took her BS in Electrical Engineering, and a BA in Economics from
Rice University, 1997. Gupta has also worked for AT&T Labs, NATO's
SACLANT Undersea Research Center, Hewlett Packard, and Microsoft.
More information about her research is available at her group's