[Colloquium] [Staff] Theory Seminars at Computer Science

[Donna Brooms]

Date:         Tuesday, April 24,  2012
Time:        3 p.m.
Place:        Ryerson 251, 1100 E. 58th Street
_________________________ 

Speaker:    Nathan Srebro
From:        University of Chicago:TTI-C
Title:         Fat Shattering, Learning, and Lower Bounds on Convex Optimization
Abstract:   We show how the fat-shattering dimension can be used to obtain lower bounds for a generic class of convex optimization problems under the local access model. This in turn implies that the sample complexity for learning is upper bounded by the optimization runtime (with only local black-box accesses) of the corresponding empirical optimization problem.
Joint work with Karthik Sridharan 

*Refreshments will be served prior to the talk at 2:30 in Ryerson 255*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20120423/d20bdce9/attachment.htm