[Colloquium] Machine Learning Seminar: Andy Cotter
Dawn Ellis
dellis at ttic.edu
Wed Feb 27 11:29:36 CST 2013
When: Thursday, at 12:00
Where: TTIC, 6045 S Kenwood Avenue, 5th Floor, Room #526
Who: Andy Cotter
Title: Fast Kernel SVM Training and Classification
Abstract:
I will discuss two algorithms related to kernel SVM optimization. The
first, the kernelized stochastic batch perceptron (SBP), is a training
algorithm which achieves a runtime guarantee better than that of any other
known kernelized SVM optimization approach. The second is an algorithm for
taking an existing solution to the kernel SVM optimization problem and
"sparsifying it", resulting in a new classifier with the same
generalization performance, but a smaller support size. This method is very
simple, very fast, and, due to the sparsity of its result, gives solutions
which classify previously-unseen points efficiently.
Title:
Host: Samory Kpotufe, samory at ttic.edu
--
Dawn Ellis
Administrative Assistant
773-834-1757
dellis at ttic.edu
TTIC
6045 S. Kenwood Ave.
Chicago, IL. 60637
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20130227/e7d75a52/attachment.htm
More information about the Colloquium
mailing list