[Colloquium] Show & Tell Series at TTI-C 11/30 @ 12:15

[Katherine Cumming]

TOYOTA TECHNOLOGICAL INSTITUTE 
SHOW AND TELL SERIES TALK
 
Speaker: Yasemin Altun
Speaker's homepage: http://www.tti-c.org/altun.html
 
 
Title: Kernel Methods for Graphical Models
 
Time: Tuesday, November 30th, 12:15pm
Place: TTI-C conference room (1427 E. 60th St. - 2nd Floor) Lunch
provided
 
Abstract: 
Kernel methods for graphical models combine advantages of two important
areas of machine learning: From graphical models, they inherit the
property of exploiting correlations between multiple labels which
constitute a structure. As kernel methods, they can also learn
non-linear discriminant functions and thus overcome limitations of
parametric methods which exhibit conceptual and computational
difficulties in high-dimensional input spaces. 
 
Recent work on this hybrid field can be classified as generalizations of
two kernel methods to structure learning: Maximum Margin Markov Networks
[TGK04] and Hidden Markov Support Vector Machines [ATH03] as
generalizations of SVMs and Kernel Conditional Random Fields [LZL04] and
Gaussian Process Sequence Classification [AHS04] as generalizations of
Gaussian Processes. These approaches as well as others such as
perceptron for structured domains [CD01] differ in their objective
functions, parameterizations and optimization methods.  In this talk, I
will present these methods and investigate their similarities and
differences.
 
 
If you have questions, or would like to meet the speaker, please contact
Katherine at 4-1994 or kcumming at tti-c.org. For information on future
TTI-C talks or events, please go to the TTI-C Events page:
http://www.tti-c.org/events.shtml
 
 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20041123/d34b7ecf/attachment.htm