• Proceedings of the Korean Information Science Society Conference
• /
• 2011.06c
• /
• pp.109-111
• /
• 2011

대부분의 빈발 부분그래프를 이용한 그래프 분류 알고리즘들은 빈발 부분그래프를 마이닝하여 개별적인 빈발 부분그래프의 포함 여부를 특징 벡터로 구성하는 단계와 기계학습 알고리즘들을 훈련시켜 분류 모델을 수립하는 단계로 구성된다. 이와 같은 그래프 분류 알고리즘들은 부분그래프의 개별적인 존재 여부만을 이용하여 특징을 구성하기 때문에 변별력이 떨어지는 문제점이 있다. 본 논문에서는 빈발 부분그래프들이 동시 발생하는 특징 벡터의 변별력을 반영할 수 있는 특징선택 기법을 적용한 모델 기반 탐색트리 기법을 제안한다. 동시 발생 부분그래프를 특징으로 사용하여 변별력을 향상시킬 수 있으며, 모델기반 탐색 트리를 사용하여 제안하는 기법이 기존의 방법보다 더 높은 그래프 분류 성능을 보이는 것을 입증하였다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.06b
• /
• pp.229-231
• /
• 2006

웹 마이닝은 e-서비스 시스템에서 고객 활동을 분석하기 위하여 널리 보급된 방법 중 하나로서 궁극적인 목표는 새로운 고객을 얻고 기존 고객을 유지하면서 고객의 생산성을 증가시키는데 도움을 줄 수 있는 유용한 정보를 인식하는 것이다. 그러나 웹 로그 자료와 고객의 구매 패턴 사이에 직접적인 관계가 없고, 실험 데이터 집합이 적고 부정확 할 경우 실험 데이터의 적은 집합만으로 유용한 정보를 인식하는 것은 불충분하기 때문에 유용한 정보를 인식하는 것은 더욱 어렵게 된다. 본 논문에서는 기업들에게 유용한 패턴을 제공할 수 있는 독자적인 분류 방법을 사용하여 기존 고객의 보존력을 높일 수 있는 높은 정확도를 가지는 소형 트리를 구축할 수 있었다.

• Annual Conference of KIPS
• /
• 2006.11a
• /
• pp.645-648
• /
• 2006

생물학적 데이터 서열에는 크게 DNA 서열과 단백질 서열이 있다. 이들 서열 데이터들은 여러 데이터베이스에 걸쳐 매우 방대한 양을 가지고 있으며, 각각의 서열은 수백 또는 수천 개의 항목들을 가지고 있어 길이가 매우 길다. 일반적으로 유전적인 변형, 또는 변이로부터 보존된 영역이나 특정 패턴들을 서열 안에 포함하고 있는데 생물학적 서열 데이터에서 보존된 영역이나 패턴들은 계통발생학적 근거로 활용 될 수도 있으며 기능과 밀접한 관계를 가지기도 한다. 따라서 서열들로부터 빈번하게 발생하는 패턴을 발견하고자 하는 알고리즘 개발이 요구되고 있다. 초창기 Apriori 알고리즘을 변형하여 빈발 패턴을 발견하고자 하는 노력들로부터 근래에는 PrefixSpan 트리를 이용하여 효과적으로 성능을 개선하고 있지만 아직까지는 여러 번의 데이터베이스 접근이 요구되고 있어 성능저하가 발생한다. 이에 본 논문에서는 접미사 트리를 변형하여 데이터베이스 접근을 획기적으로 줄이고 많은 서열들로부터 빈번하게 발생하는 연속적인 서열을 효과적으로 발견하는 방법을 제안한다.

• The KIPS Transactions:PartB
• /
• v.12B no.4 s.100
• /
• pp.487-498
• /
• 2005

Recently, It has focused on decision tree algorithm that can handle large dataset. However, because most of these algorithms for large datasets process data in a batch mode, if new data is added, they have to rebuild the tree from scratch. h more efficient approach to reducing the cost problem of rebuilding is an approach that builds a tree incrementally. Representative algorithms for incremental tree construction methods are BOAT and ITI and most of these algorithms use a local discretization method to handle the numeric data type. However, because a discretization requires sorted numeric data in situation of processing large data sets, a global discretization method that sorts all data only once is more suitable than a local discretization method that sorts in every node. This paper proposes an incremental tree construction method that efficiently rebuilds a tree using a global discretization method to handle the numeric data type. When new data is added, new categories influenced by the data should be recreated, and then the tree structure should be changed in accordance with category changes. This paper proposes a method that extracts sample points and performs discretiration from these sample points to recreate categories efficiently and uses confidence intervals and a tree restructuring method to adjust tree structure to category changes. In this study, an experiment using people database was made to compare the proposed method with the existing one that uses a local discretization.

• Proceedings of the Korean Information Science Society Conference
• /
• 2000.10a
• /
• pp.189-191
• /
• 2000

데이터 마이닝의 방법론 중 클러스터링은 데이터베이스 객체들의 에트리뷰트 값에 근거하여 유사한 그룹으로 식별하는 기술적인 작업이다. 그러나 대부분 알고리즘들은 데이터의 차원이 증가할수록 형성된 전체 데이터 공간은 매우 방대하므로 의미있는 클러스터의 탐색이 더욱 어렵다. 따라서 효과적인 클러스터링을 위해서는 클러스터가 포함될 데이터 공간의 예측이 필요하다. 본 논문에서는 고차원 데이터에서 각 차원에 대한 점진적 프로젝션을 이용한 클러스터링 방법을 제안한다. 제안한 방법에서는 클러스터가 포함될 가능성이 있는 데이터공간의 후보영역을 결정하여, 이 영역에서 점들의 평균값을 중심으로 클러스터를 탐색한다.

• The KIPS Transactions:PartD
• /
• v.17D no.4
• /
• pp.253-258
• /
• 2010

Traditional frequent pattern mining considers equal profit/weight value of every item. Weighted Frequent Pattern (WFP) mining becomes an important research issue in data mining and knowledge discovery by considering different weights for different items. Existing algorithms in this area are based on fixed weight. But in our real world scenarios the price/weight/importance of a pattern may vary frequently due to some unavoidable situations. Tracking these dynamic changes is very necessary in different application area such as retail market basket data analysis and web click stream management. In this paper, we propose a novel concept of dynamic weight and an algorithm DWFPM (dynamic weighted frequent pattern mining). Our algorithm can handle the situation where price/weight of a pattern may vary dynamically. It scans the database exactly once and also eligible for real time data processing. To our knowledge, this is the first research work to mine weighted frequent patterns using dynamic weights. Extensive performance analyses show that our algorithm is very efficient and scalable for WFP mining using dynamic weights.

• Bioinformatics and Biosystems
• /
• v.1 no.2
• /
• pp.134-138
• /
• 2006

Depression is the most common and widespread mood disorder. About 20% of the population might suffer a major, incapacitating episode of depression during their lifetime. This disorder can be classified into two types: major depressive disorders and bipolar disorder. Since pharmaceutical treatments are different according to types of depression disorders, correct and fast classification is quite critical for depression patients. Yet, classical statistical method, such as minnesota multiphasic personality inventory (MMPI), have some difficulties in applying to depression patients, because the patients suffer from concentration. We used electroencephalogram (EEG) analysis method fer classification of depression. We extracted nonlinearity of information flows between channels and estimated approximate entropy (ApEn) for the EEG at each channel. Using these attributes, we applied two types of data mining classification methods: decision tree and possibilistic support vector machines (PSVM). We found that decision tree showed 85.19% accuracy and PSVM exhibited 77.78% accuracy for classification of depression, 30 patients with major depressive disorder and 24 patients having bipolar disorder.

• Journal of Digital Convergence
• /
• v.20 no.2
• /
• pp.217-223
• /
• 2022

Subspace clustering for online data streams requires a large amount of memory resources as all subsets of data dimensions must be examined. In order to track the continuous change of clusters for a data stream in a finite memory space, in this paper, we propose a grid-based subspace clustering algorithm that effectively uses memory resources. Given an n-dimensional data stream, the distribution information of data items in data space is monitored by a grid-cell list. When the frequency of data items in the grid-cell list of the first level is high and it becomes a unit grid-cell, the grid-cell list of the next level is created as a child node in order to find clusters of all possible subspaces from the grid-cell. In this way, a maximum n-level grid-cell subspace tree is constructed, and a k-dimensional subspace cluster can be found at the k^th level of the subspace grid-cell tree. Through experiments, it was confirmed that the proposed method uses computing resources more efficiently by expanding only the dense space while maintaining the same accuracy as the existing method.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.6
• /
• pp.1275-1282
• /
• 2014

Too much information exists in ubiquitous environment, and therefore it is not easy to obtain the appropriately classified information from the available data set. Decision tree algorithm is useful in the field of data mining or machine learning system, as it is fast and deduces good result on the problem of classification. Sometimes, however, a decision tree may have leaf nodes which consist of only a few or noise data. The decisions made by those weak leaves will not be effective and therefore should be excluded in the decision process. This paper proposes a method using a classifier, UChoo, for solving a classification problem, and suggests an effective method of decision process involving only the important leaves and thereby excluding the noisy leaves. The experiment shows that this method is effective and reduces the erroneous decisions and can be applied when only important decisions should be made.

• The KIPS Transactions:PartD
• /
• v.16D no.4
• /
• pp.621-630
• /
• 2009

In recent years, there have been many attempts to connect the latest RFID (Radio Frequency Identification) technology with EIS (Enterprise Information System) and utilize them. However, in most cases the focus is only on the simultaneous multiple reading capability of the RFID technology neglecting the management of massive data created from the reader. As a result, it is difficult to obtain time-related information such as flow prediction and analysis in process control. In this paper, we suggest a new method called 'procedure tree', an enhanced and complementary version of PathTree which is one of RFID data mining techniques, to manage massive RFID data sets effectively and to perform a real-time process control efficiently. We will evaluate efficiency of the proposed system after applying real-time process management system connected with the RFID-based EIS. Through the suggested method, we are able to perform such tasks as prediction or tracking of process flow for real-time process control and inventory management efficiently which the existing RFID-based production system could not have done.