Activity Number:
|
282
- Sampling and Ensembling in Statistical Computing
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 9, 2022 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistical Computing
|
Abstract #323185
|
|
Title:
|
A Non-Degrading Streaming Sampler for Recursive Bayesian Inference
|
Author(s):
|
Ian Taylor* and Andee Kaplan and Brenda Betancourt
|
Companies:
|
Colorado State University and Colorado State University and University of Florida
|
Keywords:
|
Markov chain monte carlo;
Recursive Bayesian inference;
record linkage;
streaming data;
filtering
|
Abstract:
|
Recursive Bayesian inference is an important tool for applications in which data arrives sequentially and updated parameter estimates are desired each time data arrives. Models for which the posterior distribution is estimated via Markov Chain Monte Carlo can use Prior-Proposal-Recursive Bayes (PPRB) to resample existing posterior samples using the likelihood of the new data. Like all filtering methods, PPRB will eventually converge to sampling from a degenerate distribution, limiting its usefulness for repeated application in longitudinal data settings. We present a recursive Bayes sampling strategy that extends PPRB to avoid the eventual tendency towards degeneracy by the addition of a transition kernel step run in parallel on each filtered sample. We prove that this sampler produces samples from the target posterior distribution and demonstrate that it avoids degenerate sampling via simulation. We compare this sampler to other streaming samplers for recursive inference and present an application to streaming record linkage.
|
Authors who are presenting talks have a * after their name.