[Ml-stat-talks] Statistics Seminar this Wednesday Oct 6th
Jelena Bradic
jbradic at Princeton.EDU
Mon Oct 4 10:32:40 EDT 2010
Dear all,
This Wednesday at 12:30pm we will be having a Statistics Seminar talk
with Richard Samworth from Cambridge University.
Title: "Title: Optimal weighted nearest neighbor classifiers"
Abstract:
Classifiers based on nearest neighbors are perhaps the simplest and
most intuitively appealing of all nonparametric classifiers. Arguably
the most obvious defect with the $k$-nearest neighbor classifier is
that it places equal weight on the class labels of each of the $k$
nearest neighbors to the point being classified. Intuitively, one
would expect improvements in terms of the misclassification rate to
be possible by putting decreasing weights on the class labels of the
successively more distant neighbors. In this talk, we determine the
optimal weighting scheme, and quantify the benefits attainable.
Remarkably, the improvements depend only on the dimension of the
data, not on the underlying population densities.
Best,
Jelena
--------------------------------------
Jelena Bradic
PhD Candidate
Department of Operations Research and Financial Engineering
Princeton University, Princeton, NJ 08544
jbradic at princeton.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/ml-stat-talks/attachments/20101004/ecb3c4e7/attachment.html>
More information about the Ml-stat-talks
mailing list