Seminar Details
| Date |
30-10-2004 |
| Time |
11:00 |
| Room/Location |
Università degli studi di Genova, Aula Magna, Via Balbi 5 |
| Title |
Analytic Methods in Learning Theory |
| Speaker |
S. Smale - Toyota Technological Institute at Chicago; T. Poggio - MIT; Alessandro Verri - Universita |
| Affiliation |
Toyota Technological Institute at Chicago, Universita' di Genova, MIT, Universita' di Modena e Reggi |
| Link |
http://www.disi.unige.it/person/RosascoL/amlt/index1.htm
|
| Abstract |
Summary:
Learning from examples is a possible gateway to understanding
intelligence in brains and machines. Classical theoretical
approaches in learning theory strongly rely on results about
empirical processes.
In recent years, some works proposed a functional analytical
approach based on the notion of stability. This framework
provides a general way to study the problem of generalization
(that is, the capability of learning).
In particular, this new approach makes it clear that the problem
of learning is strongly related to some traditional research
areas like inverse problems and sampling theory. Moreover, it
poses the challenge of exploiting the stability property to
design novel and effective algorithms. The goal of this workshop
is to bring together some recent results in this direction.
Organizing Committee:
T. Poggio, Massachusetts Institute of Technology
L. Rosasco, Universita' di Genova
A. Verri, Universita' di Genova |
|
|
 |
