• Communications for Statistical Applications and Methods
• /
• 제15권3호
• /
• pp.333-342
• /
• 2008

회귀모형에서의 변수선택에 관한 정리를 공분산분석 모형으로 확장하였다. 공분산분석 모형에서 몇개의 회귀변수를 제거한 축소모형을 세우는 경우에 추정량의 변화를 알아본 결과, 회귀계수 뿐만아니라 분산분석계수도 추정량의 편차는 증가하지만 분산은 감소하며, 어떤 경우에는 평균제곱오차도 감소한다는 결론을 얻었다.

• 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.

• Structural Engineering and Mechanics
• /
• 제56권4호
• /
• pp.625-648
• /
• 2015

Modern super-tall buildings are more sensitive to strong winds. The evaluation of wind loads for the design of these buildings is of primary importance. A direct monitoring of wind forces acting on super-tall structures is quite difficult to be realized. Indirect measurements interpreted by inverse techniques are therefore favourable since dynamic response measurements are easier to be carried out. To this end, a Kalman filtering based inverse approach is developed in this study so as to estimate the wind loads on super-tall buildings based on limited structural responses. The optimum solution of Kalman filter gain by solving the Riccati equation is used to update the identification accuracy of external loads. The feasibility of the developed estimation method is investigated through the wind tunnel test of a typical super-tall building by using a Synchronous Multi-Pressure Scanning System. The effects of crucial factors such as the type of wind-induced response, the covariance matrix of noise, errors of structural modal parameters and levels of noise involved in the measurements on the wind load estimations are examined through detailed parametric study. The effects of the number of vibration modes on the identification quality are studied and discussed in detail. The made observations indicate that the proposed inverse approach is an effective tool for predicting the wind loads on super-tall buildings.

• Communications for Statistical Applications and Methods
• /
• 제23권2호
• /
• pp.105-117
• /
• 2016

In the paper, as a sequence of the first tutorial, we discuss sufficient dimension reduction methodologies used to estimate central subspace (sliced inverse regression, sliced average variance estimation), central mean subspace (ordinary least square, principal Hessian direction, iterative Hessian transformation), and central $k^{th}$-moment subspace (covariance method). Large-sample tests to determine the structural dimensions of the three target subspaces are well derived in most of the methodologies; however, a permutation test (which does not require large-sample distributions) is introduced. The test can be applied to the methodologies discussed in the paper. Theoretical relationships among the sufficient dimension reduction methodologies are also investigated and real data analysis is presented for illustration purposes. A seeded dimension reduction approach is then introduced for the methodologies to apply to large p small n regressions.

• Structural Engineering and Mechanics
• /
• 제28권2호
• /
• pp.129-152
• /
• 2008

This paper considers the solution of the stochastic differential equations (SDEs) with random operator and/or random excitation using the spectral SFEM. The random system parameters (involved in the operator) and the random excitations are modeled as second order stochastic processes defined only by their means and covariance functions. All random fields dealt with in this paper are continuous and do not have known explicit forms dependent on the spatial dimension. This fact makes the usage of the finite element (FE) analysis be difficult. Relying on the spectral properties of the covariance function, the Karhunen-Loeve expansion is used to represent these processes to overcome this difficulty. Then, a spectral approximation for the stochastic response (solution) of the SDE is obtained based on the implementation of the concept of generalized inverse defined by the Neumann expansion. This leads to an explicit expression for the solution process as a multivariate polynomial functional of a set of uncorrelated random variables that enables us to compute the statistical moments of the solution vector. To check the validity of this method, two applications are introduced which are, randomly loaded simply supported reinforced concrete beam and reinforced concrete cantilever beam with random bending rigidity. Finally, a more general application, randomly loaded simply supported reinforced concrete beam with random bending rigidity, is presented to illustrate the method.

• Quantitative Bio-Science
• /
• 제37권2호
• /
• pp.133-141
• /
• 2018

Graphical lasso is one of the most popular methods to estimate a sparse precision matrix, which is an inverse of a covariance matrix. The objective function of graphical lasso imposes an ${\ell}_1$-penalty on the (vectorized) precision matrix, where a tuning parameter controls the strength of the penalization. The selection of the tuning parameter is practically and theoretically important since the performance of the estimation depends on an appropriate choice of tuning parameter. While information criteria (e.g. AIC, BIC, or extended BIC) have been widely used, they require an asymptotically unbiased estimator to select optimal tuning parameter. Thus, the biasedness of the ${\ell}_1$-regularized estimate in the graphical lasso may lead to a suboptimal tuning. In this paper, we propose a two-staged bias-correction procedure for the graphical lasso, where the first stage runs the usual graphical lasso and the second stage reruns the procedure with an additional constraint that zero estimates at the first stage remain zero. Our simulation and real data example show that the proposed bias correction improved on both edge recovery and estimation error compared to the single-staged graphical lasso.

• Journal of the Korean Data and Information Science Society
• /
• 제24권5호
• /
• pp.1077-1088
• /
• 2013

High-dimensional data analysis arises from almost all scientific areas, evolving with development of computing skills, and has encouraged penalized estimations that play important roles in statistical learning. For the past years, various penalized estimations have been developed, and the least absolute shrinkage and selection operator (LASSO) proposed by Tibshirani (1996) has shown outstanding ability, earning the first place on the development of penalized estimation. In this paper, we first introduce a number of recent advances in high-dimensional data analysis using the LASSO. The topics include various statistical problems such as variable selection and grouped or structured variable selection under sparse high-dimensional linear regression models. Several unsupervised learning methods including inverse covariance matrix estimation are presented. In addition, we address further studies on new applications which may establish a guideline on how to use the LASSO for statistical challenges of high-dimensional data analysis.

• The Journal of the Acoustical Society of Korea
• /
• 제12권2E호
• /
• pp.47-62
• /
• 1993

SMI 방법은 수치적인 불안정성과 아울러 많은 계산량을 갖는다. 본 논문에서는 역 공분산 행렬의 Cholesky 분할을 이용하여 SMI 방법보다 효율적인 방법을 제안한다. 제안한 방법에서는 적응 빔 형상과 검출이 하나의 구조로 실현되며 이에 피룡한 역 공분산 행렬의 Cholesky factor는 secondary 입력으로부터 GS 프로세서를 이용하여 추정한다. 제안한 구조의 중요한 특징은 공분산 행렬과 Cholesky factor를 직접 구할 필요가 없다는 점이며, 또한 GS 프로세서의 장점을 이용한 systolic 구조를 사용함으로써 효율적인 계산을 수행할 수 있다. 모의 실험을 통하여 제안한 방법의 성능과 SMI 방법의 성능을 서로 비교하였다. 또한 nonhomogeneous 환경에서 동작하기 위한 방법이 제시되었으며, 아울러 계산량이 많은 GS 구조의 단점을 극복하기 위해 lattice-GS 구조를 이용하는 방법을 제안하였다.

• ETRI Journal
• /
• 제34권5호
• /
• pp.684-689
• /
• 2012

A mode-singular-value-decomposition (SVD) maximum likelihood (ML) estimation procedure is proposed for the source localization problem under an additive measurement error model. In a practical situation, the noise variance is usually unknown. In this paper, we propose an algorithm that does not require the noise covariance matrix as a priori knowledge. In the proposed method, the weight is derived by the inverse of the noise magnitude square in the ML criterion. The performance of the proposed method outperforms that of the existing methods and approximates the Taylor-series ML and Cram$\acute{e}$r-Rao lower bound.

• International journal of advanced smart convergence
• /
• 제5권4호
• /
• pp.66-70
• /
• 2016

When tooth grinding occurs, electrical stimulation is given at the same time, and tooth grinding stops on such stimulation. Electromyography signals are used as control signals of electrical stimulation to disturb tooth grinding. However because of the electrical stimulation, the M-waves are generated and mixed with spontaneous electromyogram. In this study, we designed an optimal filter to remove M-wave and conserve spontaneous electromyogram simultaneously. The inverse power method (IPM) showed that the optimal filter coefficient is the eigenvector corresponding to the minimum eigenvalue of the input covariance matrix. In order to evaluate the performance of the optimal filter, we compared using a conventional band pass filter and adaptive filter using least mean square algorithm. The experimental results show that the optimal filter can effectively remove the M-wave compared to the previously studied prediction error filter.