Research on and implications of Support Vector Machines
			    John Langford

The goal of this talk is to give an overview of Support Vector
Machines.  I will briefly review Support Vector Machines and compare
the strengths and weaknesses of Support Vector Machines with Neural
Nets.  I'll cover several current thrusts in SVM research including
virtual support vectors, combination with decision trees, reduced set
SVM's, k-means algorithm vs. SVM, and support vector ubiquity.  Then,
I'll discuss some implications of the SVM research.