View Single Post
Old 10-03-2012, 11:42 AM
yaser's Avatar
yaser yaser is offline
Join Date: Aug 2009
Location: Pasadena, California, USA
Posts: 1,478
Default Re: expected value for the SVM generalization error

Originally Posted by Andrs View Post
In lecture 14 slide 20 there is the following statement:
Expected value(E_out) = < Expected val(#of support vectors/N-1)
That is, the exp val for the upper bound for the E_out is limited by the expected value of the #number of support vectors divided by the number of data points minus 1.

I am trying to find some reference to a book or article that presents this upper bound for the E_out for SVM but I am having a really hard time to find anything. Unfortunatly the course book does not cover SVM at all. Any suggestion about articles that discuss this upper bound for SVM more or less clearly?
The result is in Vapnik's short book "The Nature of Statistical Learning Theory" and his more technical book on the subject.
Where everyone thinks alike, no one thinks very much
Reply With Quote