|
Activity Number:
|
410
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #306740 |
|
Title:
|
Spatially Adaptive Bayesian Thin-Plate Splines
|
|
Author(s):
|
Yu Yue*+
|
|
Companies:
|
University of Missouri-Columbia
|
|
Address:
|
2303 Whitegate Drive, Columbia, MO, 65202,
|
|
Keywords:
|
function estimation ; spatial adaptation ; thin-plate splines ; Gibbs sampling
|
|
Abstract:
|
There is a well-known Bayesian interpretation for function estimation by spline smoothing. In this paper, we derive a Bayesian hierarchical representation for a discretized thin-plate spline with a spatially adaptive smoothness penalty. We show how this allows the fitted model to adapt to various degrees of smoothness in the data. We propose a convenient form of the discretized prior to adaptively model the precision parameter and discuss efficient computational algorithms for implementing the Gibbs sampler.
|
