• Journal of the Korean Statistical Society
• /
• 제30권2호
• /
• pp.179-192
• /
• 2001

When a statistical model has a hierarchical structure such as multilayer perceptrons in neural networks or Gaussian mixture density representation, the model includes distribution with unidentifiable parameters when the structure becomes redundant. Since the exact structure is unknown, we need to carry out statistical estimation or learning of parameters in such a model. From the geometrical point of view, distributions specified by unidentifiable parameters become a singular point in the parameter space. The problem has been remarked in many statistical models, and strange behaviors of the likelihood ratio statistics, when the null hypothesis is at a singular point, have been analyzed so far. The present paper studies asymptotic behaviors of the maximum likelihood estimator and the Bayesian predictive estimator, by using a simple cone model, and show that they are completely different from regular statistical models where the Cramer-Rao paradigm holds. At singularities, the Fisher information metric degenerates, implying that the cramer-Rao paradigm does no more hold, and that he classical model selection theory such as AIC and MDL cannot be applied. This paper is a first step to establish a new theory for analyzing the accuracy of estimation or learning at around singularities.

• 한국경영과학회:학술대회논문집
• /
• 한국경영과학회 1996년도 추계학술대회발표논문집; 고려대학교, 서울; 26 Oct. 1996
• /
• pp.117-120
• /
• 1996

The objective of this study is to suggest modified Covariance Structure Analysis that combine with existing Multivariate Statistical Method which is used Casual Analysis Method in Management Information. For this purpose, we'll consider special feature and limitation about Correlation Analysis, Regression Analysis, Path Analysis and connect Covariance Structure Analysis with Statistical Factor Analysis so that theoretical casual model compare with variables structure in collecting data. A example is also presented to show the practical applicability of this approach.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제15권3호
• /
• pp.191-199
• /
• 2011

In this paper, we adopt a position specific scoring matrix as an abstraction of amino acid type to derive two new statistical potentials for protein structure prediction, and investigated its effect on the quality of the potentials compared to that derived using residue specific amino acid identity. For stringent test of the potential quality, we carried out folding simulations of 91 residue A chain of protein 2gpi, and found unexpectedly that the abstract amino acid type improved the quality of the one-body type statistical potential, but not for the two-body type statistical potential which describes long range interactions. This observation could be effectively used when one develops more accurate potentials for structure prediction, which are usually involved in merging various one-body and many-body potentials.

• 한국경영과학회지
• /
• 제33권2호
• /
• pp.153-174
• /
• 2008

There have been two streams of the studies on development structure : conceptual model approach and statistical analysis approach. But In these days, the latter has been becoming the main approach owing to the development of multivariate statistical methods and statistical packages. In this study, we examine methodologies and results of the leading researches related to development structure based on statistical analysis and propose the future research directions. This analysis would be expected to contribute toward the construction of long-range development policies on each country.

• Structural Engineering and Mechanics
• /
• 제46권4호
• /
• pp.459-486
• /
• 2013

A novel damage classification method based on wavelet packet transform and statistical analysis is developed in this study for structural health monitoring. The response signal of a structure under an impact load is normalized and then decomposed into wavelet packet components. Energies of these wavelet packet components are then calculated to obtain the energy distribution. Statistical similarity comparison based on an F-test is used to classify the structure from changes in the wavelet packet energy distribution. A statistical indicator is developed to describe the damage extent of the structure. This approach is applied to the test results from simply supported reinforced concrete beams in the laboratory. Cases with single and two damages are created from static loading, and accelerations of the structure from under impact loads are analyzed. Results show that the method can be used with no reference baseline measurement and model for the damage monitoring and assessment of the structure with alarms at a specified significance level.

• Journal of the Korean Statistical Society
• /
• 제30권3호
• /
• pp.407-418
• /
• 2001

Buhlmann(1944) established the validity of the block bootstrap proposed by Kunsch when it is applied to p-dimensional $\alpha$-mixing dependent sequence. But his result requires a rather restrictive condition on p in the sense that p is entangled with dependence structure. We address that such restriction on p(or complication of dependence structure with p) could be removed completely when the underlying dependence structure is replace by more weakly dependent structure such as ø-mixing.

• 한국정보과학회논문지:컴퓨팅의 실제 및 레터
• /
• 제15권12호
• /
• pp.993-997
• /
• 2009

클러스터링 기법은 다양한 분야에서 이용되어 왔으며, 통계 기반 기계번역에서도 익히 사용된 기법이다. 그러나 기존의 연구에서는 깊이 있는 문법적인 분석 없이 기계학습 기법을 사용하거나, 문장구조의 정보를 사용하더라도 정규식을 이용하여 판별하는 선에서 그치는 경우가 많았다. 본 논문에서는 각 문장의 의존관계 문법에 따른 구조와 조사 등의 품사 정보를 사용하여 문장구조를 파악하고 유형별로 분류하여 각각에 특화된 언어모델을 획득하는 방법과, 이를 구 기반 통계기계번역에 추가적인 정보로 사용하여 번역성능을 향상하는 데 이용하는 방법을 제안한다.

• 호남수학학술지
• /
• 제46권2호
• /
• pp.224-236
• /
• 2024

This study inspects the structure of CR-product of a holomorphic statistical manifold. Findings concerning geodesic submanifolds and totally geodesic foliations in the context of dual connections have been demonstrated. The integrability of distributions in CR-statistical submanifolds has been characterized. The statistical version of CR-product in the holomorphic statistical manifold has been researched. Additionally, some assertions for curvature tensor field of the holomorphic statistical manifold have been substantiated.

• 한국통계학회:학술대회논문집
• /
• 한국통계학회 2003년도 추계 학술발표회 논문집
• /
• pp.63-68
• /
• 2003

One of the widely accepted assumptions in many statistical problem is that the underlying distribution is symmetric. Though a large number of nonparametric test are available in the literature for this problem, very few procedures focuses on the distributional structure when the symmetry assumption is rejected. Yanagimoto and Sibuya (1972) provided the various types of asymmetric distributional structure, positive biasedness, namely. In this paper we consider the test of symmetry against several new positive biasedness restrictions which are stronger than Yanagimoto and Sibuya's type II bias but weaker than type IV (III) bias.

• Communications for Statistical Applications and Methods
• /
• 제26권4호
• /
• pp.359-370
• /
• 2019

Grouping structures in covariates are often ignored in regression models. Recent statistical developments considering grouping structure shows clear advantages; however, reflecting the grouping structure on the quantile regression model has been relatively rare in the literature. Treating the grouping structure is usually conducted by employing a group penalty. In this work, we explore the idea of group penalty to the quantile regression models. The grouping structure is assumed to be known, which is commonly true for some cases. For example, group of dummy variables transformed from one categorical variable can be regarded as one group of covariates. We examine the group quantile regression models via two real data analyses and simulation studies that reveal the beneficial performance of group quantile regression models to the non-group version methods if there exists grouping structures among variables.