[Colloquium] Reminder: Show & Tell Series at TTI-C (Today @ 12:15pm)

[Katherine Cumming]

TTI-C SHOW AND TELL SERIES TALK
 
Speaker: Adam Kalai, TTI-C
Speaker's home page: http://www.tti-c.org//kalai.html
 
Time: Tuesday, February 15th
Location: TTI-C Conference Room
Lunch/Refreshments Provided  @ 12:00pm 
Seminar @ 12:15p
Learning probabilities, learning theory, and regressing to regression
Abstract:
Much of the work in Computational Learning Theory has focused on hard
classification. One is given a set of labeled training data, for example
a set of job applications labeled by weather or not they led to a
successful hire, and one wants to develop a rule for automatically
predicting whether or not future candidates will be hired, based on
their applications. Clearly, it would be more useful to predict the
*probability* that a candidate will be hired.

This is now recognized as a central problem in Machine Learning. In
fact, this is a special case of regression, long studied by
statisticians. I will describe some of the basic models from
Computational Learning Theory. I will also relate this to some basic
models from statistics, both old and new. The "point" is that
Computational Learning Theory gives a fresh perspective on regression by
giving provable guarantees about efficiency and accuracy of regression
algorithms on various classes of functions. No background in learning is
assumed. 
 
-----------------
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.html
 
 
 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20050215/58bcda8e/attachment.htm