|
Activity Number:
|
410
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #309657 |
|
Title:
|
Evolutionary Markov Chain Monte Carlo Algorithms for Expected Utility Maximization
|
|
Author(s):
|
Marco Ferreira*+ and Ramiro Ruiz and Alexandra Schmidt
|
|
Companies:
|
University of Missouri-Columbia and Universidade Federal do Rio de Janeiro and Universidade Federal do Rio de Janeiro
|
|
Address:
|
Dept Statistics, Columbia, MO, 65211-6100,
|
|
Keywords:
|
Optimal Bayesian decisions ; genetic algorithm ; optimal design ; environmental network design
|
|
Abstract:
|
We propose an evolutionary Markov chain Monte Carlo (EMCMC) framework for expected utility maximization. This is particularly useful when the optimal decision cannot be obtained analytically. Our proposed framework simultaneously computes the expected utility and maximizes it over the decision space. We develop an algorithm that simulates a population of Markov chains, each having its own temperature. The population evolves according to genetic operators, allowing the chains to explore the decision space both locally and globally. As a result, the algorithm explores the decision space very effectively. We illustrate that with two applications. First, we perform optimal design of a network of monitoring stations for spatio-temporal ground-level ozone. Second, we develop estimation of quantitative trait loci (QTL).
|
