Speaker: Martin Anthony

Title:  Using hyperplanes iteratively for classification

Abstract: 
We consider the generalization accuracy of classification methods based 
on the iterative use of linear classifiers. The resulting classifiers, 
called threshold decision lists act as follows. Some points of the data 
set to be classified are given a particular classification according to 
a linear threshold function (or hyperplane). These are then removed from 
consideration, and the procedure is iterated until all points are classified. 
Geometrically, we can imagine that points of the same classification are 
successively chopped off from the data set by a hyperplane. 
We analyse theoretically the generalization properties of data classification 
techniques that are based on the use of threshold decision lists and on the 
special subclass of multilevel threshold functions. We present bounds on the 
generalization error in a standard probabilistic learning framework.


Short bio: 
Martin Anthony received the B.Sc degree in Mathematics from Glasgow University, 
Scotland, in 1988 and the PhD degree in Mathematics from the University 
of London in 1991. He has been on the faculty of the Mathematics Department 
at the London School of Economics since 1990, where he is currently a Reader 
in Mathematics. His research interests are in computational learning theory, 
discrete mathematics and neural networks.