|
Activity Number:
|
562
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Learning and Data Mining
|
| Abstract - #304049 |
|
Title:
|
Nonlinear Regression Modeling via Bayesian Regularization with Lasso-Type Penalties
|
|
Author(s):
|
Shohei Tateishi*+ and Sadanori Konishi
|
|
Companies:
|
Kyushu University and Kyushu University
|
|
Address:
|
6-10-1 Hakozaki, Higashi-ku, Fukuoka, 812-8581, Japan
|
|
Keywords:
|
Basis expansion ; Information criterion ; Lasso ; Nonlinear regression ; Regularization
|
|
Abstract:
|
We consider the problem of constructing nonlinear regression models with Gaussian basis functions, using the lasso-type regularization. Regularization with the lasso penalty is one of the attractive procedures in that it shrinks some unknown parameters towards exactly zero in linear regression models. We propose imposing weighted lasso penalty on the nonlinear regression model and selecting the number of basis functions effectively. In order to select tuning parameters involved in the regularization method, we introduce a model selection criterion obtained from information-theoretic and Bayesian viewpoints. The proposed nonlinear modeling procedure is investigated through Monte Carlo simulations. Numerical results show the effectiveness of the proposed method in prediction accuracy.
|
