• The KIPS Transactions:PartB
• /
• v.13B no.4 s.107
• /
• pp.393-400
• /
• 2006

There are many feature extraction methods for face recognition. We need a new method to overcome the small sample problem that the number of feature variables is larger than the sample size for face image data. The paper considers partial least squares(PLS) as a new dimension reduction technique for feature vector. Principal Component Analysis(PCA), a conventional dimension reduction method, selects the components with maximum variability, irrespective of the class information. So, PCA does not necessarily extract features that are important for the discrimination of classes. PLS, on the other hand, constructs the components so that the correlation between the class variable and themselves is maximized. Therefore PLS components are more predictive than PCA components in classification. The experimental results on Manchester and ORL databases shows that PLS is to be preferred over PCA when classification is the goal and dimension reduction is needed.

• The KIPS Transactions:PartD
• /
• v.12D no.1 s.97
• /
• pp.111-120
• /
• 2005

Software development creates many product that class components, Class Diagram, form, object, and design pattern. So this Paper suggests Index/XML Sequence Relation information system for product abstraction and classification, the system of design product Sequence Relation abstraction which can store, reuse design patterns in the meta modeling database with pattern Relation information. This is Index/XML Sequence Relation system which can easily change various relation information of product for product abstraction and classification. This system designed to extract and classify design pattern efficiently and then functional indexing, sequence base indexing for standard pattern, code indexing to change pattern into code and grouping by Index-ID code, and its role information can apply by structural extraction and design pattern indexing process. and it has managed various products, class item, diagram, forms, components and design pattern.

• Journal of KIISE:Software and Applications
• /
• v.28 no.8
• /
• pp.560-569
• /
• 2001

지식 기반 정보검색 시스템에서의 질의 확장은 단어간의 의미 관계를 고려한 지식베이스를 필요로 한다. 기존의 단순 마이닝 기법은 사용자의 선호도를 고려하지 않은 채 연관 단어를 추출하므로 재현율은 향상되나 정확도는 저하된다. 본 논문에서는 단어간의 의미 관게를 고려한 연관 단어 중에서 사용자가 선호하는 연관 단어만을 포함하는 정확도가 향상된 최적화된 연관 단어 지식베이스 구축을 위한 방법을 제안한다. 이를 위해 컴퓨터 분야의 웹문서를 8개의 클래스로 분류하고, 각 클래스별 웹문서에서 명사를 추출한다. 추출된 명사를 대상으로 Apriori 알고리즘을 이용하여 연관 단어를 추출하고, 유전자 알고리즘을 이용하여 사용자가 선호하지 않은 연관 단어를 지식베이스의 구축 대상에서 제외시킨다. 본 논문에서 제안된 Apriori 알고리즘과 유전자 알고리즘의 성능을 평가하기 위하여 Apriori 알고리즘은 상호 정보량과 Rocchio 알고리즘과 비교하며, 유전자 알고리즘은 TF.IDF를 이용한 단어 정제 방법과 비교한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04a
• /
• pp.586-588
• /
• 2001

원전보호계통(RPS: Reactor Protection System)은 사고 시 치명적 피해를 입을 수 있다는 점에서 안전에 대한 중요도가 가장 높은 Safety 1E class로 분류되며, 이러한 보호계통을 디지털 라이즈 하는데 있어서 높은 신뢰도에 대한 보장이 필요하다. 따라서 본 논문에서는 DPPS(Digital Plant Protection System) 내에서 작동하는 내장형 소프트웨어의 높은 신뢰성을 보장하기 위한 테스팅 방법론을 제시하고자 한다. DPPS에서 작동하는 내장형 소프트웨어를 테스트하기 위한 방법은 크게 두 가지로 나누어진다. 첫 번째 단계는 절차중심의 프로그램에서 객체를 추출하고 이를 이용하여 클래스를 추출하는 제공학의 단계이다. 두 번째 단계는 이러한 클래스들을 이용하여 레벨별 테스팅을 수행하기 위한 테스트 아이템을 추출하고, 추출된 테스트 아이템을 이용하여 테스트 케이스를 선정하는 단계이다. 이렇게 각 레벨별로 선정된 테스트 케이스를 이용하여 단위 테스팅, 통합 테스팅, 시스템 테스팅 이렇게 3단계의 레벨별 테스팅을 수행한다.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.12
• /
• pp.3668-3674
• /
• 2009

Use case diagram is used for defining inter-action between users and systems in object-oriented modeling. It is very important to defining users' requirements for efficient software development. In this paper, we propose a object-oriented modeling process using hierarchical use case analysis for designing class model. First, We define many use case diagrams by several hierarchical modeling level. And next, we can also design class model using the use case diagrams. Our proposed modeling process provides interaction between use case model and class model. That can make us to check the modeling process during the software development. Using the proposed object-oriented modeling we can develop software based on users' requirements. It is very useful for class modeling.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.5 no.2
• /
• pp.28-43
• /
• 1995

This paper concerns with automatic generation of fuzzy rules which can be used for pattern classification. Feature space is recursively subdivided into hyperspheres, and each hypersphere is represented by its centroid and bounding distance. Fuzzy rules are then generated based on the constructed hyperspheres. The resulting fuzzy rules have very simple premise parts, and they can be organized into a hierarchical structure so that classification process can be implemented very rapidly. The experimented results show that the suggested method works very well compared to other methods.

• Annual Conference on Human and Language Technology
• /
• 2005.10a
• /
• pp.13-20
• /
• 2005

이 논문은 공학적인 접근 방법에 기반한 단계적인 전문분야 시소러스 구축 방법을 제안한다. 시소러스 구축 과정은 용어 추출 단계, 용어 분류 단계, 계층 구조 구축 의 3단계로 구성되고, 모든 단계에서 자동 처리와 전문가 검증 작업을 거친다. 추출된 용어를 미리 정해진 분류 체계에 따라 분리한 후 여러 개의 작은 시소러스를 구축하고, 마지막으로 전체 시소러스로 결합한다. 이 방법은 1) 시소러스를 구축하는 복잡도가 줄어들고, 2) 클래스 단위의 작은 시소러스가 다른 전문분야 시소러스에 쉽게 재사용 될 수 있으며, 3) 각 클래스에 포함된 용어들의 분포를 쉽게 판단할 수 있는 장점이 있다. 제안한 방법을 이용하여 한국어 정보기술 분야 시소러스를 구축하였다. 시소러스 구축에 사용된 용어들은 정보기술 분야의 최근의 한국어 신문과 특허 문서에서 추출하였기 때문에 한국에서 만들어진 신조어를 포함한다. 구축된 시소러스는 81 개의 상위 레벨클래스와 1,000개 이상의 용어로 구성된다.

• Journal of the Korea Convergence Society
• /
• v.13 no.3
• /
• pp.1-9
• /
• 2022

UML class diagrams are used to visualize the static aspects of a software system and are involved from analysis and design to documentation and testing. Software modeling using class diagrams is essential for software development, but it may be not an easy activity for inexperienced modelers. The modeling productivity could be improved with a dataset of class diagrams which are classified by domain categories. To this end, this paper provides a classification method for a dataset of class diagram images. First, real class diagrams are selected from collected images. Then, class names are extracted from the real class diagram images and the class diagram images are classified according to domain categories. The proposed classification model has achieved 100.00%, 95.59%, 97.74%, and 97.77% in precision, recall, F1-score, and accuracy, respectively. The accuracy scores for the domain categorization are distributed between 81.1% and 95.2%. Although the number of class diagram images in the experiment is not large enough, the experimental results indicate that it is worth considering the proposed approach to class diagram image classification.

• Journal of KIISE:Software and Applications
• /
• v.35 no.1
• /
• pp.41-49
• /
• 2008

Fingerprint classification is useful to reduce the matching time of a huge fingerprint identification system by categorizing fingerprints into predefined classes according to their global features. Although global features are distributed diversly because of the uniqueness of a fingerprint, previous fingerprint classification methods extract global features non-adaptively from the fixed region for every fingerprint. We propose an novel method that extracts features adaptively for each fingerprint in order to classify various fingerprints effectively. It extracts ridge directional values as feature vectors from the region after searching the feature region by calculating variations of ridge directions, and classifies them using support vector machines. Experimental results with NIST4 database show that we have achieved a classification accuracy of 90.3% for the five-class problem and 93.7% for the four-class problem, and proved the validity of the proposed adaptive method by comparison with non-adaptively extracted feature vectors.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.4
• /
• pp.492-497
• /
• 2005

Feature extraction is important to classify data with large dimension such as image data. The representative feature extraction methods lot feature extraction ate PCA, ICA, LDA and MLP, etc. These algorithms can be classified in two groups: unsupervised algorithms such as PCA, LDA, and supervised algorithms such as LDA, MLP. Among these two groups, supervised algorithms are more suitable to extract the features for classification because of the class information of input data. In this paper we suggest a new feature extraction algorithm PCA-FX which uses class information with PCA to extract ieatures for classification. We test our algorithm using Yale face database and compare the performance of proposed algorithm with those of other algorithms.