• The Journal of Engineering Research
• /
• v.3 no.1
• /
• pp.65-73
• /
• 1998

One approach to improved image restoration methods has been the incorporation of additional source information via Gibbs priors that assume a source that is piecewise smooth. A natural Gibbs prior for expressing such constraints is an energy function defined on binary valued line processes as well as source intensities. However, the estimation of both continuous variables and binary variables is known to be a difficult problem. In this work, we consider the application of the deterministic annealing method. Unlike other methods, the deterministic annealing method offers a principled and efficient means of handling the problems associated with mixed continuous and binary variable objectives. The application of the deterministic annealing method results in a sequence of objective functions (defined only on the continuous variables) whose sequence of solutions approaches that of the original mixed variable objective function. The sequence is indexed by a control parameter (the temperature). The energy functions at high temperatures are smooth approximations of the energy functions at lower temperatures. Consequently, it is easier to minimize the energy functions at high temperatures and then track the minimum through the variation of the temperature. This is the essence of a continuation method. We show experimental results, which demonstrate the efficacy of the continuation method applied to a Bayesian restoration model.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.8
• /
• pp.2060-2071
• /
• 1999

The complicate software failure system is defined to the superposition of the points of failure from several component point process. Because the likelihood function is difficulty in computing, we consider Gibbs sampler using iteration sampling based method. For each observed failure epoch, we applied to latent variables that indicates with component of the superposition mode. For model selection, we explored the posterior Bayesian criterion and the sum of relative errors for the comparison simple pattern with superposition model. A numerical example with NHPP simulated data set applies the thinning method proposed by Lewis and Shedler[25] is given, we consider Goel-Okumoto model and Weibull model with GOS, inference of parameter is studied. Using the posterior Bayesian criterion and the sum of relative errors, as we would expect, the superposition model is best on model under diffuse priors.

• Asian-Australasian Journal of Animal Sciences
• /
• v.24 no.3
• /
• pp.330-335
• /
• 2011

The mode of inheritance of total milk yield and its genetic parameters were investigated in Churra dairy sheep through segregation analyses using a Monte Carlo Markov Chains (MCMC) method. Data which consisted of 7,126 lactations belonging to 5,154 ewes were collected between 1999 and 2002 from 15 Spanish Churra dairy flocks. A postulated major gene was assumed to be additive and priors used for variance components were uniform. Based on 50 000 Gibbs samples from ten replicates chains of 100,000 cycles, the estimated marginal posterior means${\pm}$posterior standard deviations of variance components of milk yield were $23.17{\pm}18.42$, $65.20{\pm}25.05$, $120.40{\pm}42.12$ and $420.83{\pm}40.26$ for major gene variance ($\sigma_G^2$), polygenic variance ($\sigma_u^2$), permanent environmental variance ($\sigma_{pe}^2$) and error variance ($\sigma_e^2$), respectively. The results of this study showed the postulated major locus was not significant, and the 95% highest posterior density regions ($HPDs_{95%}$) of most major gene parameters included 0, and particularly for the major gene variance. The estimated transmission probabilities for the 95% highest posterior density regions ($HPDs_{95%}$) were overlapped. These results indicated that segregation of a major gene was unlikely and that the mode of inheritance of total milk yield in Churra dairy sheep is purely polygenic. Based on 50,000 Gibbs samples from ten replicates chains of 100,000 cycles, the estimated polygenic heritability and repeatability were $h^2=0.20{\pm}0.05$ and r=$0.34{\pm}0.06$, respectively.

• The Korean Journal of Applied Statistics
• /
• v.27 no.6
• /
• pp.1069-1076
• /
• 2014

Rainfall weather patterns have changed due to global warming and sudden heavy rainfalls have become more frequent. Economic loss due to heavy rainfall has increased. We study the generalized Pareto distribution for modelling rainfall in Seoul based on data from 1973 to 2008. We use several priors including Jeffrey's noninformative prior and Gibbs sampling method to derive Bayesian posterior predictive distributions. The probability of heavy rainfall has increased over the last ten years based on estimated posterior predictive distribution.

• The Korean Journal of Applied Statistics
• /
• v.37 no.1
• /
• pp.103-118
• /
• 2024

The horseshoe prior is notably one of the most popular priors in sparse regression models, where only a small fraction of coefficients are nonzero. The parameter space of the horseshoe prior is much smaller than that of the spike and slab prior, so it enables us to efficiently explore the parameter space even in high-dimensions. However, on the other hand, the horseshoe prior has a high computational cost for each iteration in the Gibbs sampler. To overcome this issue, various MCMC algorithms for the horseshoe prior have been proposed to reduce the computational burden. Especially, Johndrow et al. (2020) recently proposes an approximate algorithm that can significantly improve the mixing and speed of the MCMC algorithm. In this paper, we compare (1) the traditional MCMC algorithm, (2) the approximate MCMC algorithm proposed by Johndrow et al. (2020) and (3) its variant in terms of computing times, estimation and variable selection performance. For the variable selection, we adopt the sequential clustering-based method suggested by Li and Pati (2017). Practical performances of the MCMC methods are demonstrated via numerical studies.