• The KIPS Transactions:PartB
• /
• v.8B no.6
• /
• pp.657-668
• /
• 2001

It is difficult that web documents are classified with exact user intention because existing document classification systems are based on word frequency number using single keyword. To improve this defect, first, we use keyword, a query, domain knowledge. Like explanation based learning, first, query is analyzed with knowledge based information and then structured user intention information is extracted. We use this intention tree in the course of existing word frequency number based document classification as user information and constraints. Thus, we can classify web documents with more exact user intention. In classifying document, structured user intention information is helpful to keep more documents and information which can be lost in the system using single keyword information. Our hybrid approach integrating user intention information with existing statistics and probability method is more efficient to decide direction and range of document category than existing word frequency approach.

• Journal of the Korea Society of Computer and Information
• /
• v.6 no.2
• /
• pp.20-26
• /
• 2001

The research into the conditions of the active rules is a very important element in the enhancing of the performance of the active database, and the processing time of the calculation generated from the conditions must be minimized in order to improve performance. In this paper, we propose the conditions processing system with the preprocessor which determines the delta tree structure and constructs the classification tree. Due to the characteristics of the active database through which the active rules can be comprehended beforehand. the preprocessor can be introduced. In this paper, the delta tree and classification tree which can effectively Process the join and selection operations instead of the delta relation is proposed. enhancing the condition evaluation performance.

• The KIPS Transactions:PartD
• /
• v.10D no.1
• /
• pp.23-32
• /
• 2003

Since the data is collected from disparate sources in many actual data mining environments, it is common to have data values in different abstraction levels. This paper shows that such multiple abstraction levels of data can cause undesirable effects in decision tree classification. After explaining that equalizing abstraction levels by force cannot provide satisfactory solutions of this problem, it presents a method to utilize the data as it is. The proposed method accommodates the generalization/specialization relationship between data values in both of the construction and the class assignment phase of decision tree classification. The experimental results show that the proposed method reduces classification error rates significantly when multiple abstraction levels of data are involved.

• Journal of KIISE:Software and Applications
• /
• v.34 no.1
• /
• pp.18-29
• /
• 2007

In this paper we present a system that classifies brain MR images by using 2 level decision tree learning. There are two kinds of information that can be obtained from images. One is the low-level features such as size, color, texture, and contour that can be acquired directly from the raw images, and the other is the high-level features such as existence of certain object, spatial relations between different parts that must be obtained through the interpretation of segmented images. Learning and classification should be performed based on the high-level features to classify images according to their semantic meaning. The proposed system applies decision tree learning to each level separately, and the high-level features are synthesized from the results of low-level classification. The experimental results with a set of brain MR images with tumor are discussed. Several experimental results that show the effectiveness of the proposed system are also presented.

• Proceedings of the Safety Management and Science Conference
• /
• 2000.05a
• /
• pp.411-422
• /
• 2000

본 연구에서는 결정트리 분야에서 각 분류알고리즘을 살펴보고 통계학의 판별분석과 기계학습(Machine Learning)분야에서 분류알고리즘을 비교하고, 자료에 따라 오분류율을 분석 하였다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.04b
• /
• pp.88-90
• /
• 2004

주기억 징치 DBMS성능에 캐쉬 미스가 중요한 요소이다. 그래서 캐쉬 미스를 줄여주는 캐쉬 인식 트리(chash consclous trees)들이 개발되어 왔다. 캐쉬 인식 트리에서 사용한 기법들은 포인터 압축, 키 압축 개념으로 일반화 할 수 있다. 포인터 압축은 CS$B^{＋}$-트리처럼 노드에 각 자식 노드를 가리키는 포인터를 제거하고 대신 세그먼트에 저장된 자식 노드들 중 첫 번째 자식 노드를 가리키는 포인터를 저장하는 개념이다. 키 압축은 pkB-트리, R-트리처럼 키 길이를 출이는 개념이다. 본 논문에서는 키 압축 개념과 포인터 압축 개념을 동싱에 지원하고, 디스크 기반의 GiST를 캐쉬 인식하도록 확장한 CC-GiST를 제안한다. 본 논문의 공헌은 다음과 같이 요약된다. 1)기존의 캐쉬 인식 트리들의 기법을 분류하고 분석함으로써, 캐쉬 인식 트리에 적용할 수 있는 일반적인 방법을 도출하였다. 2)포인터 압축을 위해 세그먼트의 개념을 키 압축을 위하여 베이스 키의 개념을 CC-GiST에 도입하였다. 3)디스크 기반의 GiST를 위해 기정의된 메소드들을 캐쉬 인식하도록 완전하게 수정하였다. 4) 제안한 CC-GiST를 이용하여 기존의 대표적인 캐쉬 인식 트리인 CSB＋-트리와 CR-트리를 구현하는 방법을 기술하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.11a
• /
• pp.17-18
• /
• 2008

다변량 퍼지 의사결정트리(이하 MFDT)는 학습 모델의 구조가 간소하고 분류율이 높다는 장점 때문에 일반 퍼지 의사결정트리를 대신해 손동작 인식 시스템의 분류기로 사용되었다. 다양한 사용자의 손동작 특성을 분류하기 위해 여러 개의 인식 모델을 만들고 새로운 사용자에게 가장 적합한 모델을 선택해 사용하는 모델 선택 기법도 손동작 인식에 적용되었다. 모델 선택 과정을 통해 선택된 모델은 기존 모델 중에서 새로운 사용자의 특성에 가장 가깝지만 해당 사용자에 최적화된 모델이라고는 할 수 없다. 이 논문에서는 MFDT 모델을 새로 입력된 데이터를 이용해 적응시키는 방법을 설명하고 실험 결과를 통해 적응 성능을 검증한다.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.12
• /
• pp.855-864
• /
• 2013

Decision tree classification is one of the important problems in data mining fields and data minings have been important tasks in the fields of large database technologies. Therefore the coupling efforts of data mining systems and database systems have led the developments of database primitives supporting data mining functions such as decision tree classification. These primitives consist of the special database operations which support the SQL implementation of decision tree classification algorithms. These primitives have become the consisting modules of database systems for the implementations of the specific algorithms. There are two aspects in the developments of database primitives which support the data mining functions. The first is the identification of database common primitives which support data mining functions by analysis. The other is the provision of the extended mechanism for the implementations of these primitives as an interface of database systems. In data mining, some primitives want be stored in DBMS is one of the difficult problems. In this paper, to solve of the problem, we describe the database primitives which construct and apply the optimized decision tree classifiers. Then we identify the useful operations for various classification algorithms and discuss the implementations of these primitives on the commercial DBMS. We implement these primitives on the commercial DBMS and present experimental results demonstrating the performance comparisons.

• Journal of Digital Contents Society
• /
• v.18 no.8
• /
• pp.1635-1640
• /
• 2017

In this paper, a hierarchical image tree model for weather classification is defined in a single outdoor image, and a weather classification algorithm using image intensity and k-mean segmentation image is proposed. In the first level of the hierarchical image tree model, the indoor and outdoor images are distinguished. Whether the outdoor image is daytime, night, or sunrise/sunset image is judged using the intensity and the k-means segmentation image at the second level. In the last level, if it is classified as daytime image at the second level, it is finally estimated whether it is sunny or foggy image based on edge map and fog rate. Some experiments are conducted so as to verify the weather classification, and as a result, the proposed method shows that weather features are effectively detected in a given image.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.11b
• /
• pp.999-1002
• /
• 2003

최근 인터넷 환경의 향상에 따라 네트워크는 멀티미디어 서비스 범위 확대와 효율적인 네트워크 자원 사용을 위해서 멀티캐스팅 기술을 지원하는 것이 바람직하다. 멀티캐스트 라우팅 프로토콜은 소스 기반 트리와 코어 기반 트리 방식으로 분류할 수 있다. 소스 기반 트리 방식은 멀티캐스트 그룹에서 소스 당 하나의 트리를 형성하게 되는 반면에 코어 기반 트리 방식은 그룹의 모든 노드가 하나의 트리를 공유한다. 이런 코어 기반 트리 방식에서는 코어 혹은 센터의 위치에 따라 멀티캐스트 라우팅 트리의 모양이 영향을 받으며 그에 따라 라우팅의 성능에 중대한 영향을 주게 된다. 본 논문에서는 공유 트리를 형성하는 라우팅 아키텍쳐에서 코어의 위치를 결정하기 위한 기존 방법들을 분석하며 성능 향상을 고려한 방안을 제안한다.