|
Activity Number:
|
261
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #307841 |
|
Title:
|
Asymptotic Equivalence for Nonparametric Regression Experiments with Random Design
|
|
Author(s):
|
Andrew Carter*+
|
|
Companies:
|
University of California, Santa Barbara
|
|
Address:
|
Dept of Statistics, Santa Barbara, CA, 93106-3110,
|
|
Keywords:
|
asymptotic equivalence ; nonparametric regression
|
|
Abstract:
|
We explore an extension of Brown and Low's (1996) result establishing the asymptotic equivalence of nonparametric regression experiments and the continuous Gaussian-process-with-drift experiment. Specifically, we consider the effect of an unknown design densities in the regression experiment. The inclusion of the additional parameter also induces changes in the limiting Gaussian experiments - they are now a hierarchical model with the information about the design as one component. We will demonstrate the connection between the nonparametric regression experiments and the hierarchical Gaussian models which suggests a connection to Gaussian deconvolution problems.
|
