• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.7 no.1
• /
• pp.81-98
• /
• 2013

Since Probabilistic Latent Semantic Analysis (PLSA) and Latent Dirichlet Allocation (LDA) were introduced, many revised or extended topic models have appeared. Due to the intractable likelihood of these models, training any topic model requires to use some approximation algorithm such as variational approximation, Laplace approximation, or Markov chain Monte Carlo (MCMC). Although these approximation algorithms perform well, training a topic model is still computationally expensive given the large amount of data it requires. In this paper, we propose a new method, called non-simultaneous sampling deactivation, for efficient approximation of parameters in a topic model. While each random variable is normally sampled or obtained by a single predefined burn-in period in the traditional approximation algorithms, our new method is based on the observation that the random variable nodes in one topic model have all different periods of convergence. During the iterative approximation process, the proposed method allows each random variable node to be terminated or deactivated when it is converged. Therefore, compared to the traditional approximation ways in which usually every node is deactivated concurrently, the proposed method achieves the inference efficiency in terms of time and memory. We do not propose a new approximation algorithm, but a new process applicable to the existing approximation algorithms. Through experiments, we show the time and memory efficiency of the method, and discuss about the tradeoff between the efficiency of the approximation process and the parameter consistency.

• The Korean Journal of Applied Statistics
• /
• v.18 no.1
• /
• pp.1-13
• /
• 2005

Latent Class model has been considered recently by many researchers and practitioners as a tool for identifying heterogeneous segments or groups in a population, and grouping objects into the segments. In this paper we consider data on prostate cancer patients from Korean National Cancer Institute and propose a method for grouping prostate cancer patients by using latent class Poisson model. A Bayesian approach equipped with a Markov chain Monte Carlo method is used to overcome the limit of classical likelihood approaches. Advantages of the proposed Bayesian method are easy estimation of parameters with their standard errors, segmentation of objects into groups, and provision of uncertainty measures for the segmentation. In addition, we provide a method to determine an appropriate number of segments for the given data so that the method automatically chooses the number of segments and partitions objects into heterogeneous segments.

• The Bulletin of The Korean Astronomical Society
• /
• v.41 no.2
• /
• pp.62.2-62.2
• /
• 2016

We present a newly developed algorithm based on a Bayesian method for 2D tilted-ring analysis of disk galaxies which operates on velocity fields. Compared to the conventional ones based on a chi-squared minimisation procedure, this new Bayesian-based algorithm less suffers from local minima of the model parameters even with high multi-modality of their posterior distributions. Moreover, the Bayesian analysis implemented via Markov Chain Monte Carlo (MCMC) sampling only requires broad ranges of posterior distributions of the parameters, which makes the fitting procedure fully automated. This feature is essential for performing kinematic analysis of an unprecedented number of resolved galaxies from the upcoming Square Kilometre Array (SKA) pathfinders' galaxy surveys. A standalone code, the so-called '2D Bayesian Automated Tilted-ring fitter' (2DBAT) that implements the Bayesian fits of 2D tilted-ring models is developed for deriving rotation curves of galaxies that are at least marginally resolved (> 3 beams across the semi-major axis) and moderately inclined (20 < i < 70 degree). The main layout of 2DBAT and its performance test are discussed using sample galaxies from Australia Telescope Compact Array (ATCA) observations as well as artificial data cubes built based on representative rotation curves of intermediate-mass and massive spiral galaxies.

• Smart Structures and Systems
• /
• v.26 no.5
• /
• pp.667-680
• /
• 2020

This study aims at reporting a systematic procedure for evaluating the static and dynamic structural performance of steel bridges based on a short-period structural health monitoring measurement. Sungsu bridge located in Korea is considered as a case study presenting the most recent tests carried out to examine the bridge condition. Short-period measurements of Structural Health Monitoring (SHM) system were used during the bridge testing phase. A novel symmetry index is introduced using statistical analyses of deflection and strain measurements. Frequency Domain Decomposition (FDD) is implemented to the strain measurements to estimate the bridge mode shapes and damping ratios. Furthermore, Markov Chain Monte Carlo (MCMC) is also implemented to examine the reliability of bridge performance while ambient design trucks are in static or moving at different speeds. Strain, displacement and acceleration were measured at selected locations on the bridge. The results show that the symmetry index can be an efficient and useful measure in assessing the steel bridge performance. The results from the used method reveal that the performance of the Sungsu bridge is safe under operational conditions.

• Korean Management Science Review
• /
• v.34 no.2
• /
• pp.15-33
• /
• 2017

This paper considers a Bayesian Network analysis for understanding the heterogeneous cross-category effects of different promotion activities and developing an efficient overall promotion strategy for a large retail store. More specifically we differentiate price reduction promotion and floor promotion and study their heterogeneous effect on consumer purchase behavior under a market basket setting. We then utilize Bayesian networks in identifying complex association structure in market basket dataset by analyzing the effects of different promotional activities and also include the effects of time, family income and size. We find from our Bayesian network analysis that the dominant cross-category promotion effect of price promotion is the indirect effect whereas the dominant cross-category promotion effect of floor promotion is the direct effect. Also, among the demographic variables we find that family size of the household is linked with more product categories compared to income and see that there are differences in the extent of the effects by product category. Finally, we also show the existence of products acting as a network hub and how they can be utilized by retailers faced with a limited marketing budget and suggest a more efficient promotion strategy.

• Proceedings of the Korean Statistical Society Conference
• /
• 2005.05a
• /
• pp.17-23
• /
• 2005

In this paper we consider the well-known semiparametric proportional hazards models for survival analysis. These models are usually used with few covariates and many observations (subjects). But, for a typical setting of gene expression data from DNA microarray, we need to consider the case where the number of covariates p exceeds the number of samples n. For a given vector of response values which are times to event (death or censored times) and p gene expressions(covariates), we address the issue of how to reduce the dimension by selecting the significant genes. This approach enables us to estimate the survival curve when n ${\ll}$p. In our approach, rather than fixing the number of selected genes, we will assign a prior distribution to this number. The approach creates additional flexibility by allowing the imposition of constraints, such as bounding the dimension via a prior, which in effect works as a penalty To implement our methodology, we use a Markov Chain Monte Carlo (MCMC) method. We demonstrate the use of the methodology to diffuse large B-cell lymphoma (DLBCL) complementary DNA (cDNA) data and Breast Carcinomas data.

• Structural Engineering and Mechanics
• /
• v.69 no.3
• /
• pp.317-326
• /
• 2019

This research aims to assess the tight seismic risk curve of the intake tower at Geumgwang reservoir by considering the recorded historical earthquake data in the Korean Peninsula. The seismic fragility, a significant part of risk assessment, is updated by using Bayesian inference to consider the uncertainties and computational efficiency. The reservoir is one of the largest reservoirs in Korea for the supply of agricultural water. The intake tower controls the release of water from the reservoir. The seismic risk assessment of the intake tower plays an important role in the risk management of the reservoir. Site-specific seismic hazard is computed based on the four different seismic source maps of Korea. Probabilistic Seismic Hazard Analysis (PSHA) method is used to estimate the annual exceedance rate of hazard for corresponding Peak Ground Acceleration (PGA). Hazard deaggregation is shown at two customary hazard levels. Multiple dynamic analyses and a nonlinear static pushover analysis are performed for deriving fragility parameters. Thereafter, Bayesian inference with Markov Chain Monte Carlo (MCMC) is used to update the fragility parameters by integrating the results of the analyses. This study proves to reduce the uncertainties associated with fragility and risk curve, and to increase significant statistical and computational efficiency. The range of seismic risk curve of the intake tower is extracted for the reservoir site by considering four different source models and updated fragility function, which can be effectively used for the risk management and mitigation of reservoir.

• The Journal of Asian Finance, Economics and Business
• /
• v.8 no.4
• /
• pp.665-672
• /
• 2021

The paper examines the impact of shadow economy and corruption, along with public expenditure, trade openness, foreign direct investment (FDI), inflation, and tax revenue on the economic growth of the BRICS countries. Data were collected from the World Bank, Transparency International, and Heritage Foundation over the 1991-2017 period. The Bayesian linear regression method is used to examine whether shadow economy, corruption and other indicators affect the economic growth of countries studied. This paper applies the normal prior suggested by Lemoine (2019) while the posterior distribution is simulated using Monte Carlo Markov Chain (MCMC) technique through the Gibbs sampling algorithm. The results indicate that public expenditure and trade openness can enhance the BRICS countries' economic growth, with the positive impact probability of 75.69% and 67.11%, respectively. Also, FDI, inflation, and tax revenue positively affect this growth, though the probability of positive effect is ambiguous, ranging from 51.13% to 56.36%. Further, the research's major finding is that shadow economy and control of corruption have a positive effect on the economic growth of the BRICS countries. Nevertheless, the posterior probabilities of these two factors are 62.23% and 65.25%, respectively. This result suggests that their positive effect probability is not high.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.2
• /
• pp.71.1-71.1
• /
• 2021

We perform disk-halo decomposition of the Large Magellanic Cloud (LMC) using a novel HI velocity field extraction method, aimed at better deriving its HI kinematics and thus mass distribution in the galaxy including both baryons and dark matter. We decompose all the line-of-sight velocity profiles of the combined HI data cube of the LMC, taken from the Australia Telescope Compact Array (ATCA) and Parkes radio telescopes with an optimal number of Gaussian components. For this, we use a novel tool, the so-called BAYGAUD which performs profile decomposition based on Bayesian MCMC techniques. From this, we disentangle turbulent non-ordered HI gas motions from the decomposed gas components, and produce an HI bulk velocity field which better follows the global circular rotation of the galaxy. From a 2D tilted-ring analysis of the HI bulk velocity field, we derive the rotation curve of the LMC after correcting for its transverse, nutation and precession motions. The dynamical contributions of baryons like stars and gaseous components which are derived using the Spitzer 3.6 micron image and the HI data are then subtracted from the total kinematics of the LMC. Here, we present the bulk HI rotation curve, the mass models of stars and gaseous components, and the resulting dark matter density profile of the LMC.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.2
• /
• pp.75.4-75.4
• /
• 2019

We present a quantitative analysis of the relationship between the gas dynamics and star formation history of DDO 210 which is an irregular dwarf galaxy in the local Universe. We perform profile analysis of an high-resolution neutral hydrogen (HI) data cube of the galaxy taken with the large Very Large Array (VLA) survey, LITTLE THINGS using newly developed algorithm based on a Bayesian Markov Chain Monte Carlo (MCMC) technique. The complex HI structure and kinematics of the galaxy are decomposed into multiple kinematic components in a quantitative way like 1) bulk motions which are most likely to follow the underlying circular rotation of the disk, 2) non-circular motions deviating from the bulk motions, and 3) kinematically cold and warm components with narrower and wider velocity dispersion. The decomposed kinematic components are then spatially correlated with the distribution of stellar populations obtained from the color-magnitude diagram (CMD) fitting method. The cold and warm gas components show negative and positive correlations between their velocity dispersions and the surface star formation rates of the populations with ages of < 40 Myr and 100~400 Myr, respectively. The cold gas is most likely to be associated with the young stellar populations. Then the stellar feedback of the young populations could influence the warm gas. The age difference between the populations which show the correlations indicates the time delay of the stellar feedback.