• Journal of the Korean Data and Information Science Society
• /
• 제27권2호
• /
• pp.381-394
• /
• 2016

그래피컬 모형은 변수들 사이의 조건부 종속성을 노드와 연결선을 통하여 그래프로 나타낸다. 변수들 사이의 복잡한 연관성을 표현하기 위하여 그래피컬 모형은 물리학, 경제학, 생물학을 포함하여 다양한 분야에 적용되고 있다. 조건부 종속성은 공분산 행렬의 역행렬의 비대각 성분이 0인 것과 대응하는 두 변수의 조건부 독립이 동치임에 기반하여 공분산 행렬의 역행렬로부터 추정될 수 있다. 본 논문은 공분산 행렬의 역행렬을 희박하게 추정하는 유사가능도 기반의 CONCORD (convex correlation selection method) 방법에 대하여 기존의 BCD (block coordinate descent) 알고리즘을 랜덤 치환을 활용한 갱신 규칙과 그래픽 처리 장치 (graphics processing unit)의 병렬 연산을 활용하여 고차원 자료에 대하여 보다 효율적인 BCDR (block coordinate descent with random permutation) 알고리즘을 제안하였다. 두 종류의 네트워크 구조를 고려한 모의실험에서 제안하는 알고리즘의 효율성을 수렴까지의 계산 시간을 비교하여 확인하였다.

• 응용통계연구
• /
• 제29권1호
• /
• pp.171-180
• /
• 2016

바젤 위원회는 시장위험의 측정 도구로 Value-at-Risk(VaR)와 expected shortfall(ES)을 사용할 것을 제안하였다. 여러 문헌에서 VaR와 ES의 다양한 추정 방법들이 연구 되었다. 본 연구에서는 준모수적인 방법인 conditional autoregressive value at risk(CAViaR), conditional autoregressive expectile(CARE) 방법들, 그리고 Gaussian 준최대가능도 추정량(QMLE)를 이용한 방법을 사후 검정을 통해서 비교하고자 한다. 각 방법의 타당성을 확인하기 위해서, VaR에 대한 사후 검정은 unconditional coverage(UC)와 conditional coverage(CC) 검정을 사용하고 ES에 대한 검정은 붓스트랩 방법을 사용한다. S&P500 지수와 현대 자동차 주식가격 지수에 대하여 실증 자료 분석이 수행되었다.

• 한국신뢰성학회지:신뢰성응용연구
• /
• 제1권2호
• /
• pp.183-192
• /
• 2001

Jelinski-Moranda model and modified Jelinski-Moranda model in software reliability are studied and we consider maximum likelihood estimator and Bayes estimates of the number of faults and the fault-detection rate per fault. A gibbs sampling approach is employed to compute the Bayes estimates, future survival function is examined. Model selection based on prequential likelihood of the conditional predictive ordinates. A numerical example with simulated data set is given.

• Journal of Communications and Networks
• /
• 제8권2호
• /
• pp.205-211
• /
• 2006

This paper considers the maximum likelihood (ML) detection of 16-ary differential amplitude and phase shift keying (DAPSK) in Rayleigh fading channels. Based on the conditional likelihood function, two new receiver structures, namely ML symbol-by-symbol receiver and ML sequence receiver, are proposed. For the symbol-by-symbol detection, the conventional DAPSK detector is shown to be sub-optimum due to the complete separation in the phase and amplitude detection, but it results in very close performance to the ML detector provided that its circular amplitude decision thresholds are optimized. For the sequence detection, a simple Viterbi algorithm with only two states are adopted to provide an SNR gain around 1 dB on the amplitude bit detection compared with the conventional detector.

• Communications for Statistical Applications and Methods
• /
• 제16권3호
• /
• pp.479-485
• /
• 2009

When a part of data is unobserved the marginal likelihood of parameters given the observed data often involves analytically intractable high dimensional integral and hence it is hard to find the maximum likelihood estimate of the parameters. Simulated maximum likelihood(SML) method which estimates the marginal likelihood via Monte Carlo importance sampling and optimize the estimated marginal likelihood has been used in many applications. A key issue in SML is to find a good proposal density from which Monte Carlo samples are generated. The optimal proposal density is the conditional density of the unobserved data given the parameters and the observed data, and attempts have been given to find a good approximation to the optimal proposal density. Algorithms which adaptively improve the proposal density have been widely used due to its simplicity and efficiency. In this paper, we describe a fully adaptive algorithm which has been used by some practitioners but has not been well recognized in statistical literature, and evaluate its estimation performance and robustness via a simulation study. The simulation study shows a great improvement in the order of magnitudes in the mean squared error, compared to non-adaptive or partially adaptive SML methods. Also, it is shown that the fully adaptive SML is robust in a sense that it is insensitive to the starting points in the optimization routine.

• Journal of the Korean Data and Information Science Society
• /
• 제7권1호
• /
• pp.21-35
• /
• 1996

In this paper, estimation and prediction procedures are discussed for grneral situation in which the failure time follows the independent density $f_{i}({\varepsilon}_{i})$ for the accelerated life testing under Type II censoring. In the context of accelerated life test experiment, procedures are given for estimating the parameters in the Eyring model, and for estimating mean life at a given future stress level. The procedures given are conditional confidence interval procedures, obtained by conditioning on ancillary statistics. A comparison is made of these procedures and procedures based on asymptotic properties of the maximum, likelihood estimates.

• ETRI Journal
• /
• 제34권2호
• /
• pp.184-189
• /
• 2012

In this paper, we propose a novel approach to statistical model-based voice activity detection (VAD) that incorporates a second-order conditional maximum a posteriori (CMAP) criterion. As a technical improvement for the first-order CMAP criterion in [1], we consider both the current observation and the voice activity decision in the previous two frames to take full consideration of the interframe correlation of voice activity. This is clearly different from the previous approach [1] in that we employ the voice activity decisions in the second-order (previous two frames) CMAP, which has quadruple thresholds with an additional degree of freedom, rather than the first-order (previous single frame). Also, a soft-decision scheme is incorporated, resulting in time-varying thresholds for further performance improvement. Experimental results show that the proposed algorithm outperforms the conventional CMAP-based VAD technique under various experimental conditions.

• International Journal of Reliability and Applications
• /
• 제15권2호
• /
• pp.111-123
• /
• 2014

A new class of bivariate distributions is derived by specifying its conditionals as the exponential and gamma distributions. Some properties and relations with other distributions of the new class are studied. In particular, the estimation of parameters is considered by the methods of maximum likelihood and pseudolikelihood of a special case of the new class. An application using a real bivariate data is given for illustrating the flexibility of the new class in this context, and, also, for comparing the estimation results obtained by the maximum likelihood and pseudolikelihood methods.

• Journal of the Korean Statistical Society
• /
• 제26권2호
• /
• pp.199-210
• /
• 1997

Using regression methods based on quasi-likelihood equation, one only needs to specify the conditional mean and variance functions for the response variable in the analysis. In this paper, an omnibus lack-of-fit test is proposed to test the validity of these two functions. Our test is consistent against the alternative under which either the mean or the variance is not the one specified in the null hypothesis. The large-sample null distribution of our test statistics can be approximated through simulations. Extensive numerical studies are performed to demonstrate that the new test preserves the prescribed type I error probability. Power comparisons are conducted to show the advantage of the new proposal.

• Communications for Statistical Applications and Methods
• /
• 제28권6호
• /
• pp.627-641
• /
• 2021

An asymmetric least squares estimation method has been employed to estimate linear models for percentile regression. An asymmetric maximum likelihood estimation (AMLE) has been developed for the estimation of Poisson percentile linear models. In this study, we propose generalized nonlinear percentile regression using the AMLE, and the use of the parametric bootstrap method to obtain confidence intervals for the estimates of parameters of interest and smoothing functions of estimates. We consider three conditional distributions of response variables given covariates such as normal, exponential, and Poisson for three mean functions with one linear and two nonlinear models in the simulation studies. The proposed method provides reasonable estimates and confidence interval estimates of parameters, and comparable Monte Carlo asymptotic performance along with the sample size and quantiles. We illustrate applications of the proposed method using real-life data from chemical and radiation epidemiological studies.