• Journal of KIISE:Databases
• /
• v.29 no.5
• /
• pp.367-380
• /
• 2002

Due to recent increase in applications requiring huge amount of data such as spatial data analysis and image analysis, clustering on large databases has been actively studied. In a hierarchical clustering method, a tree representing hierarchical decomposition of the database is first created, and then, used for efficient clustering. Existing hierarchical clustering methods mainly adopted the bottom-up approach, which creates a tree from the bottom to the topmost level of the hierarchy. These bottom-up methods require at least one scan over the entire database in order to build the tree and need to search most nodes of the tree since the clustering algorithm starts from the leaf level. In this paper, we propose a novel top-down hierarchical clustering method that uses multidimensional indexes that are already maintained in most database applications. Generally, multidimensional indexes have the clustering property storing similar objects in the same (or adjacent) data pares. Using this property we can find adjacent objects without calculating distances among them. We first formally define the cluster based on the density of objects. For the definition, we propose the concept of the region contrast partition based on the density of the region. To speed up the clustering algorithm, we use the branch-and-bound algorithm. We propose the bounds and formally prove their correctness. Experimental results show that the proposed method is at least as effective in quality of clustering as BIRCH, a bottom-up hierarchical clustering method, while reducing the number of page accesses by up to 26～187 times depending on the size of the database. As a result, we believe that the proposed method significantly improves the clustering performance in large databases and is practically usable in various database applications.

• Proceedings of the Korean Information Science Society Conference
• /
• 2000.04a
• /
• pp.463-465
• /
• 2000

인터넷을 통한 그룹 통신이 활발해지면서 신뢰성을 보장하는 멀티캐스트 서비스를 요구하는 응용들이 점점 다양해지고 증가하고 있다. 이에 따라 멀티캐스트에서의 신뢰성을 보장하기 위하여 다양한 기술들이 발표되었으며 특히 트리를 기반으로 오류 및 혼잡 제어를 수행하여 신뢰성 뿐만 아니라 확장성을 보장하는 효율적인 방법에 대한 연구가 지속되고 있다. 본 논문에서는 이러한 트리 기반의 프로토콜에서 효율적이고 동적인 트리 형성 및 관리를 위한 방안을 제안한다. 하나의 멀티캐스트 그룹을 여러 개의 지역 그룹으로 나누어 트리로 구성할 때 지역 그룹에 속할 수 있는 그룹 수신자 수에 임계값을 두고 지역 대표자는 이 임계값이 초과되지 않을 때까지 새로운 참여자를 받아들일 수 있도록 하였다. 또한 새로운 노드가 그룹에 참여할 때 지역 대표자 뿐 아니라 일반 수신자들도 지역 대표자에 대한 정보를 전송해 줌으로써 참여시간의 감소와 그룹 내의 메시지 오버헤드를 줄일 수 있도록 하였다. 동적인 트리 형성 및 관리 과정에서 지역 그룹 대표자에 연결된 노드수가 적고 깊이가 깊어질수록 트리 관리에 관한 오버헤드가 증가하는 문제를 해결하기 위해 간단한 트리 재조정을 수행함으로써 트리에서 발생하는 오버헤드를 감소시킬 수 있도록 하였다.

• Proceedings of the Korean Institute of IIIuminating and Electrical Installation Engineers Conference
• /
• 2005.11a
• /
• pp.359-362
• /
• 2005

클러스터 기반의 계층적 트리 구조는 대형화가 용이하면서 라우팅 프로토콜이 단순하다는 점에서 장점을 가진다. 본 논문에서는 무선 센스 네트워크에서 클러스터 트리 네트워크 관리를 효과적으로 지원할 수 있는 방안과 라우팅 프로토콜의 데이터 모음과 지연에 대해 분석한다. 수치 해석의 결과에서 보면 중간 노드에서의 데이터 모음을 통한 통신 메시지 수를 줄임으로 센스 노드의 에너지를 크게 절약할 수 있지만, 이에 따른 메시지 전달 지연이 커짐을 알 수 있었다.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.6
• /
• pp.113-120
• /
• 2024

With the advancement of information and communication technology, we can easily generate various forms of data in our daily lives. To efficiently manage such a large amount of data, systematic classification into categories is essential. For effective search and navigation, data is organized into a tree-like hierarchical structure known as a category tree, which is commonly seen in news websites and Wikipedia. As a result, various techniques have been proposed to classify large volumes of documents into the terminal nodes of category trees. However, document classification methods using category trees face a problem: as the height of the tree increases, the number of terminal nodes multiplies exponentially, which increases the probability of misclassification and ultimately leads to a reduction in classification accuracy. Therefore, in this paper, we propose a new node expansion-based classification algorithm that satisfies the classification accuracy required by the application, while enabling detailed categorization. The proposed method uses a greedy approach to prioritize the expansion of nodes with high classification accuracy, thereby maximizing the overall classification accuracy of the category tree. Experimental results on real data show that the proposed technique provides improved performance over naive methods.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.6
• /
• pp.49-56
• /
• 1999

In many continuous speech recognition systems based on HMMs, decision tree-based state tying has been used for not only improving the robustness and accuracy of context dependent acoustic modeling but also synthesizing unseen models. To construct the phonetic decision tree, standard method performs one-level pruning using just single Gaussian triphone models. In this paper, two novel approaches, two-level decision tree and multi-mixture decision tree, are proposed to get better performance through more accurate acoustic modeling. Two-level decision tree performs two level pruning for the state tying and the mixture weight tying. Using the second level, the tied states can have different mixture weights based on the similarities in their phonetic contexts. In the second approach, phonetic decision tree continues to be updated with training sequence, mixture splitting and re-estimation. Multi-mixture Gaussian as well as single Gaussian models are used to construct the multi-mixture decision tree. Continuous speech recognition experiment using these approaches on BN-96 and WSJ5k data showed a reduction in word error rate comparing to the standard decision tree based system given similar number of tied states.

• The KIPS Transactions:PartC
• /
• v.13C no.5 s.108
• /
• pp.627-634
• /
• 2006

Overlay network eliminates the need to change the application-layer tree when the underlying network changes and enables the overlay network to survive in environments where nonmember nodes do not support multicast functionality. An overlay protocol monitors group dynamics, while underlying unicast protocols track network dynamics, resulting in more stable protocol operation and low control overhead even in a highly dynamic environment. But, if overlay multicast protocols does not know the location information of node, this makes it very difficult to build an efficient multicasting tree. So, we propose a Hierarchical Overlay Multicast Architecture (HOMA) with the location information. Because proposed architecture makes static region-based dynamic group by multicast members, it is 2-tired overlay multicasts of application layer that higher layer forms overlay multicast network between members that represent group, and support multicast between multicast members belonging to region at lower layer. This use GPS, take advantage of geographical region, and realizes a region-sensitive higher layer overlay multicast tree which is impervious to the movements of nodes. The simulation results show that our approach solves the efficiency problem effectively.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.9
• /
• pp.2247-2258
• /
• 1997

Quadtree, which is hierarchical data structure, is a very important data structure to represent binary images. Since a linear quadtree representation as is a way to store a quadtree is efficient to save space compared with other representations. It is, however, complicated and takes a large amount of time to build the linear quadtree. In this paper, we present O(1) time a linear quadtree building algorithm for a $n{\times}n$ binary image using three-dimensional $n{\times}n{\times}n$ processors on RMESH structure. Our algorithm, by use of O(1) time collapsing algorithm and reported O(1) time sorting algorithm, is simpler and easier to understand than resently presented algorithm on PARBUS structure.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1995.04a
• /
• pp.32-40
• /
• 1995

본 연구에서는 2계층 정보통신 네트워크의 구축에 직접적으로 적용가능한 트리-스타형 2계층 네트워크 설계문제를 다루었다. 정보통신네트워크 설계에의 적용 가능성을 높이기 위하여, 기존의 연구와 달리 네트워크상의 마디와 가지에 설치해야 할 설비들의 다양한 용량제약을 고려할 수 있도록 정립하였다. 또한 네트워크상의 어느 마디쌍간의 물류(정보통신) 수요를 고려할 수 있도록 하였다. 의사결정기준은 설계시에 고려해야 할 네트워크 구축비용(마디 및 가지의 설비비) 및 운영비용(물류비)의 합을 최소화하는 것으로 설정하였다. 이 문제를 혼합 0-1 정수계획 모형으로 정립하였다. 이 모형의 유용성을 살펴보기 위하여 최근 많은 관심을 끌고 있는 CATV 네트워크의 일반형태인 광가입자 선로망의 설계문제를 본 연구모형으로 표현하여 보았다. 이를 토대로 보다 일반적인 다계층 네트워크 설계문제로 확장하는 방안에 대하여도 검토하였다. 마지막으로 본 연구 모형을 정보통신네트워크 설계시에 적용하기 위하여 휴리스틱 해법을 개발할 경우, 문제 특성상 기존의 연구에서 고려하는 해법의 효율성 외에도 통신네트워크의 특성, 비용구조등을 고려해야 할 당위성에 대하여 토론하였다.

• Journal of KIISE:Computer Systems and Theory
• /
• v.36 no.6
• /
• pp.521-531
• /
• 2009

Flash memory has physical characteristics different from hard disk where two costs of a read and write operations differ each other and an overwrite on flash memory is impossible to be done. In order to solve these restrictions with software, storage systems equipped with flash memory deploy FTL(Flash Translation Layer) software. Several FTL algorithms have been suggested so far and most of them prefer sequential write pattern to random write pattern. In this paper, we provide a new technique to efficiently store and maintain the B-tree index on flash memory. The operations like inserts, deletes, updates of keys for the B-tree generate random writes rather than sequential writes on flash memory, leading to inefficiency to the B-tree maintenance. In our technique, we convert random writes generated by the B-tree into sequential writes and then store them to the write-buffer on flash memory. If the buffer is full later, some sequential writes in the buffer will be issued to FTL. Our diverse experimental results show that our technique outperforms the existing ones with respect to the I/O cost of flash memory.

• Proceedings of the Korean Information Science Society Conference
• /
• 2007.06a
• /
• pp.139-140
• /
• 2007