• 정보처리학회논문지D
• /
• 제8D권5호
• /
• pp.573-580
• /
• 2001

대용량의 고차원 데이터 집합은 고차원 데이터 고유 희소성에 의하여 상당한 양의 잡음을 포함하므로 효과적인 고차원 클러스터링에 어려움을 더한다. CLIP은 이와 같은 고차원 데이터의 특성을 지원하는 클러스터링 알고리즘으로 개발되었다. CLIP은 1차원 성형변환 프로젝션을 점진적으로 적용하여, 각 프로젝션 공간에서 얻어진 1차원 클러스터들의 곱집합을 찾는다. 이 집합은 클러스터를 포함할 뿐 아니라 잡음도 포함할 수 있다. 본 논문에서는 클러스터를 포함하는 곱집합을 정제하는 확장된 CLIP 알고리즘을 제안한다. 이미 CLIP에서 찾은 곱집합에 반복적인 2차원 프로젝션을 적용하여 클러스터의 고차원적 잡음을 제거한다. 확장된 알고리즘의 성능을 평가하기 위해 합성 데이터를 이용한 일련의 실험을 통하여 효과성을 증명한다.

• 정보기술과데이타베이스저널
• /
• 제8권2호
• /
• pp.91-101
• /
• 2001

Previous research efforts on performance evaluation of multidimensional indexes typically have used synthetic data sets distributed uniformly or normally over multidimensional space. However, recent research research result has shown that these hinds of data sets hardly reflect the characteristics of multimedia database applications. In this paper, we discuss issues on generating high dimensional data and query sets for resolving the problem. We first identify the features of the data and query sets that are appropriate for fairly evaluating performances of multidimensional indexes, and then propose HDDQ_Gen(High-Dimensional Data and Query Generator) that satisfies such features. HDDQ_Gen supports the following features : (1) clustered distributions, (2) various object distributions in each cluster, (3) various cluster distributions, (4) various correlations among different dimensions, (5) query distributions depending on data distributions. Using these features, users are able to control tile distribution characteristics of data and query sets. Our contribution is fairly important in that HDDQ_Gen provides the benchmark environment evaluating multidimensional indexes correctly.

• Communications for Statistical Applications and Methods
• /
• 제22권6호
• /
• pp.575-587
• /
• 2015

The development in data collection techniques results in high dimensional data sets, where discrimination is an important and commonly encountered problem that are crucial to resolve when high dimensional data is heterogeneous (non-common variance covariance structure for classes). An example of this is to classify microbial habitat preferences based on codon/bi-codon usage. Habitat preference is important to study for evolutionary genetic relationships and may help industry produce specific enzymes. Most classification procedures assume homogeneity (common variance covariance structure for all classes), which is not guaranteed in most high dimensional data sets. We have introduced regularized elimination in partial least square coupled with QDA (rePLS-QDA) for the parsimonious variable selection and classification of high dimensional heterogeneous data sets based on recently introduced regularized elimination for variable selection in partial least square (rePLS) and heterogeneous classification procedure quadratic discriminant analysis (QDA). A comparison of proposed and existing methods is conducted over the simulated data set; in addition, the proposed procedure is implemented to classify microbial habitat preferences by their codon/bi-codon usage. Five bacterial habitats (Aquatic, Host Associated, Multiple, Specialized and Terrestrial) are modeled. The classification accuracy of each habitat is satisfactory and ranges from 89.1% to 100% on test data. Interesting codon/bi-codons usage, their mutual interactions influential for respective habitat preference are identified. The proposed method also produced results that concurred with known biological characteristics that will help researchers better understand divergence of species.

• 한국융합학회논문지
• /
• 제4권1호
• /
• pp.43-48
• /
• 2013

Designing of similarity on high dimensional data was done. Similarity measure between high dimensional data was considered by analysing neighbor information with respect to data sets. Obtained result could be applied to big data, because big data has multiple characteristics compared to simple data set. Definitely, analysis of high dimensional data could be the pre-study of big data. High dimensional data analysis was also compared with the conventional similarity. Traditional similarity measure on overlapped data was illustrated, and application to non-overlapped data was carried out. Its usefulness was proved by way of mathematical proof, and verified by calculation of similarity for artificial data example.

• 한국CDE학회논문집
• /
• 제2권1호
• /
• pp.11-18
• /
• 1997

Scientific volume visualization addresses the representation, manipulation, and rendering of volumetric data sets, providing mechanisms for looking closely into structures and understanding their complexity and dynamics. In the past several years, a tremendous amount of research and development has been directed toward algorithms and data modeling methods for a scientific data visualization. But there has been very little work on developing a mathematical volume model that feeds this visualization. Especially, in flow visualization, the volume model has long been required as a guidance to display the very large amounts of data resulting from numerical simulations. In this paper, we focus on the mathematical representation of volumetric data sets and the method of extracting meaningful information from the derived volume model. For this purpose, a B-spline volume is extended to a high dimensional trivariate model which is called as a flow visualization model in this paper. Two three-dimensional examples are presented to demonstrate the capabilities of this model.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2008년도 하계종합학술대회
• /
• pp.877-878
• /
• 2008

We suggest a feature reduction method to classify mouse function data sets, which integrate several biological data sets represented as high dimensional vectors. To increase classification accuracy and decrease computational overhead, it is important to reduce the dimension of features. To do this, we employed Hybrid Huberized Support Vector Machine with kernels used for a kernel logistic regression method. When compared to support vector machine, this a pproach shows the better accuracy with useful features for each mouse function.

• Communications for Statistical Applications and Methods
• /
• 제3권1호
• /
• pp.11-19
• /
• 1996

Canonical correlation analysis is a multivariate technique for identifying and quantifying the statistical relationship between two sets of variables. Like most multivariate techniques, the main objective of canonical correlation analysis is to reduce the dimensionality of the dataset. It would be particularly useful if high dimensional data can be represented in a low dimensional space. In this study, we will construct statistical graphs for paired sets of multivariate data. Specifically, plots of the observations as well as the variables are proposed. We discuss the geometric interpretation and goodness-of-fit of the proposed plots. We also provide a numerical example.

• Journal of the Korean Data and Information Science Society
• /
• 제19권3호
• /
• pp.751-759
• /
• 2008

Mixtures of factor analyzers(MFA) is useful to model the distribution of high-dimensional data on much lower dimensional space where the number of observations is very large relative to their dimension. Mixtures of common factor analyzers(MCFA) can reduce further the number of parameters in the specification of the component covariance matrices as the number of classes is not small. Moreover, the factor scores of MCFA can be displayed in low-dimensional space to distinguish the groups. We propose the factor scores of MCFA as new low-dimensional features for classification of high-dimensional data. Compared with the conventional dimension reduction methods such as principal component analysis(PCA) and canonical covariates(CV), the proposed factor score was shown to have higher correct classification rates for three real data sets when it was used in parametric and nonparametric classifiers.

• 한국정보과학회논문지:데이타베이스
• /
• 제28권4호
• /
• pp.568-576
• /
• 2001

대부분의 클러스터링 알고리즘들은 고차원 공간에서 성능이 급격히 저하되는 경향이 있다. 더욱이 고차원 데이타는 상당한 양의 잡음 데이타를 포함하고 있으므로 알고리즘의 추가적인 효과성 문제를 야기한다. 그러므로 고차원 데이타의 구조와 특성을 지원하는 적합한 클러스터링 기법이 개발되어야 한다. 본 논문에서는 선형변환 프로젝션을 이용한 클러스터링 알고리즘 CLIP을 제안한다. CLIP은 고차원 클러스터링의 효율성 및 효과성 문제를 극복하기 위해 개발되었으며, 클러스터 형성에 밀접하게 연관된 부분 공간에서 클러스터를 탐사하는 기법이다. 알고리즘의 주요 사상은 각1차원적 부분공간에서의 클러스터링에 기본을 두고 있지만. 점진적인 프로젝션을 이용하여 고차원 클러스터를 탐사한 뿐만 아니라 연산을 획기적으로 줄인다. CLIP의 성능을 평가하기 위해 합성 데이타를 이용한 일련의 실험을 통하여 효율성 및 효과성을 증명한다

• 한국생물정보학회:학술대회논문집
• /
• 한국생물정보시스템생물학회 2005년도 BIOINFO 2005
• /
• pp.23-28
• /
• 2005

The high-dimensionality and insufficiency of gene expression profiles and proteomic profiles makes feature selection become a critical step in efficiently building accurate models for cancer problems based on such data sets. In this paper, we use a method, called Discrete Function Learning algorithm, to find discriminatory feature vectors based on information theory. The target feature vectors contain all or most information (in terms of entropy) of the class attribute. Two data sets are selected to validate our approach, one leukemia subtype gene expression data set and one ovarian cancer proteomic data set. The experimental results show that the our method generalizes well when applied to these insufficient and high-dimensional data sets. Furthermore, the obtained classifiers are highly understandable and accurate.