An efficient proposal distribution for Metropolis-Hastings using a B-splines technique

Authors:
Wei Shao;Guangbao Guo;Fanyu Meng;Shuqin Jia
Affiliations:
School of Mathematics, Shandong University, Jinan 250100, China;Department of Statistics, Shandong University of Technology, Zibo 255049, Shandong, China;School of Mathematics, Shandong University, Jinan 250100, China;School of Mathematics, Shandong University, Jinan 250100, China
Venue:
Computational Statistics & Data Analysis
Year:
2013

Citing 2
Cited 0

Monte Carlo Strategies in Scientific Computing

Monte Carlo Strategies in Scientific Computing
Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images

IEEE Transactions on Pattern Analysis and Machine Intelligence

Quantified Score

Hi-index	0.03

Visualization

Abstract

In this paper, we proposed an efficient proposal distribution in the Metropolis-Hastings algorithm using the B-spline proposal Metropolis-Hastings algorithm. This new method can be extended to high-dimensional cases, such as the B-spline proposal in Gibbs sampling and in the Hit-and-Run (BSPHR) algorithm. It improves the proposal distribution in the Metropolis-Hastings algorithm by carrying more information from the target function. The performance of BSPHR was compared with that of other Markov Chain Monte Carlo (MCMC) samplers in simulation and real data examples. Simulation results show that the new method performs significantly better than other MCMC methods.