Nuclear norm regularization for overparametrized Hammerstein systems

In this paper we study the overparametrization scheme for Hammerstein systems in the presence of regularization. The quality of the convex approximation is analysed, that is obtained by relaxing the implicit rank one constraint. To obtain an improved convex relaxation we propose the use of nuclear n...

Full description

Saved in:
Bibliographic Details
Published in49th IEEE Conference on Decision and Control (CDC) pp. 7202 - 7207
Main Authors Falck, T, Suykens, J A K, Schoukens, J, De Moor, B
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2010
Subjects
Bandwidth
Convex functions
Correlation
Estimation
Fasteners
Least squares approximation
Online AccessGet full text
ISBN142447745X
9781424477456
ISSN0191-2216
DOI10.1109/CDC.2010.5717892

Cover

More Information
Summary:In this paper we study the overparametrization scheme for Hammerstein systems in the presence of regularization. The quality of the convex approximation is analysed, that is obtained by relaxing the implicit rank one constraint. To obtain an improved convex relaxation we propose the use of nuclear norms, instead of using ridge regression. On several simple examples we illustrate that this yields a solution close to the best possible convex approximation. Furthermore the experiments suggest that ridge regression in combination with a projection step yield a generalization performance close to the one obtained by nuclear norms.
ISBN:142447745X
9781424477456
ISSN:0191-2216
DOI:10.1109/CDC.2010.5717892