----------------------------------------------------------------------
ARTIFICIAL INTELLIGENCE RESEARCH SEMINAR
Department of Computer Science
Worcester Polytechnic Institute
SPEAKER: James Kilian
Machine Vision Laboratory, ECE
TITLE: "Introduction to Support Vector Learning"
DATE: Thursday, Dec. 2, 1999
TIME: 11 am
ROOM: FL246, Beckett Conference Room.
ABSTRACT:
The support vector machine (SVM) is a relatively new technique for solving
a variety of learning and function approximation problems. Rather than
training to minimize empirical risk, as is often done with neural networks
and radial basis function networks, SVM minimize an upper bound on the
generalization error -- a special case of an inductive principle called
structural risk minimization (also referred to as Vapnik-Chervonenkis (VC)
theory). The pragmatic basis of support vector learning is a nonlinear
mapping of a learning problem to a high dimensional reproducing kernel
Hilbert space followed by a linear learning step (e.g., classification,
regression, approximation). This talk summarizes the content of an
independent study course, conducted this semester, on statistical
learning theory and support vector machines. The underlying theory of SVM
will be reviewed, followed by discussions of the support vector learning
procedure, comparisons with other techniques, and an application to a
synthetic classification problem. The talk will conclude with open
questions on SVM raised in the course of the independent study course.
----------------------------------------------------------------------
Everyone is welcome to attend!