|
Activity Number:
|
291
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #307550 |
|
Title:
|
Log-Density Functional ANOVA Model Estimation and Nonparametric Graphical Model Building
|
|
Author(s):
|
Yongho Jeon*+
|
|
Companies:
|
University of Wisconsin-Madison
|
|
Address:
|
104 Eagle Heights, Apt. C, Madison, WI, 53705,
|
|
Keywords:
|
density estimation ; functional ANOVA model ; graphical model
|
|
Abstract:
|
The (undirected) graphical model uses graphs to compactly display the conditional dependence among random variables and have become popular, but has mostly been studied in the parametric framework. To enhance the scope of applicability of the graphical model, we consider the building of nonparametric graphical model through its connection with log-density functional ANOVA model. We propose a new method for fitting the log-density ANOVA model based on a penalized M-estimation formulation with a novel loss function. With the smoothing spline type penalty, our method achieves the optimal rate in nonparametric estimation. With a sparsity-inducing penalty, we obtain a sparse solution in terms of function components, which provides a practical way to construct and estimate the nonparametric graphical model.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2006 program |
