• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 2001.01a
• /
• pp.122-127
• /
• 2001

This study investigates the effectiveness of time delay neural networks(TDNN) for the time dependent prediction domain. Although it is well-known fact that the back-propagation neural network(BPN) performs well in pattern recognition tasks, the method has some limitations in that it can only learn an input mapping of static (or spatial) patterns that are independent of time of sequences. The preliminary results show that the accuracy of TDNN is higher than the standard BPN with time lag. Our proposed approaches are demonstrated by the stork market prediction domain.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2004.05a
• /
• pp.143-146
• /
• 2004

With the rapid growth of e-commerce, marketers are able to observe not only purchasing behavior on what and when customers purchased, but also the individual Web usage behavior that affect purchasing. The richness of this information has the potential to provide marketers with an in-depth understanding of customer. Using commonly available Web log data, this paper examines Web usage behaviors at the individual level. By decomposing the buying process into a pattern of visits and purchase conversion at each visit, we can better understand the relationship between Web usage behavior and purchase decision. This allows us to more accurately forecast a shopper's future purchase decision at the site and hence determine the value of individual customers to the siteAccording to our research, not only information seeking behavior but also visiting duration of a customer and participative behavior such as participation in event should be considered as important predicators of purchase decision of customer in a cosmetic internet shopping mall.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.47 no.7
• /
• pp.13-22
• /
• 2005

Main purpose of this article is to classify reservoirs corresponding to their physical characteristics, for example, dam height, dam width, age, repair-works history. First of all, data set of 13,976 reservoirs was analyzed using k means and self organized maps. As a result of these analysis, lots of reservoirs have been classified into four clusters. Factors and their critical values to classify the reservoirs into four groups have been founded by generating a decision tree. The path rules to each group seem reasonable since their survivor function showed unique pattern.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2007.11a
• /
• pp.288-305
• /
• 2007

본 논문은 데이터 마이닝의 기법 중 가장 잘 알려진 지도학습 기법의 성능 향상을 위한 새로운 Hybrid 및 Combined 기법인 불일치 패턴 모델(오차 패턴 모델)에 대한 연구 논문이다. 불일치 패턴 모델이란 2개 이상의 기법 중 향후 더 레코드별로 더 잘 맞출 수 있는 기법을 메타 분류하는 불일치 패턴 모델을 개발하여, 최종적으로는 기존의 기법보다 더 좋은 분류 정확도 및 예측 향상율을 기대하기 위한 기법을 의미한다. 본 논문에서는 의사 결정나무 추론 기법인 C5.0과 C&RT 그리고 신경망 분석, 그리고 로지스틱 회귀분석과 같은 대표적인 데이터 마이닝의 지도학습 기법을 이용하여 불일치 패턴 모델을 생성하여 보고, 이들이 기존 단일 기법과 기존의 Combined 모델인 Bagging, Boosting 그리고 Stacking 기법보다 성능이 우수함을 23개의 실제 데이터 및 공신력 있는 공개 데이터를 이용하여 증명하여 보였다. 또한 데이터의 특성에 따라서 불일치 패턴 모델의 성능의 변화 및 더 우수해 지는지를 알아보기 위한 연구포 같이 수행을 하여 본 모델의 활용성을 높이고자 하였다.

• Journal of Computing Science and Engineering
• /
• v.7 no.4
• /
• pp.220-230
• /
• 2013

In recent years, progress in hardware technology has resulted in the possibility of monitoring many events in real time. The volume of incoming data may be so large, that monitoring all individual data might be intractable. Revisiting any particular record can also be impossible in this environment. Therefore, many database schemes, such as aggregation, join, frequent pattern mining, and indexing, become more challenging in this context. This paper surveys the previous efforts to resolve these issues in processing data streams. The emphasis is on specifying and processing sliding window queries, which are supported in many stream processing engines. We also review the related work on stream query processing, including synopsis structures, plan sharing, operator scheduling, load shedding, and disorder control.

• Proceedings of the Korean Information Science Society Conference
• /
• 2005.07b
• /
• pp.109-111
• /
• 2005

비즈니스 프로세스를 연구하는데 있어 최근의 이슈는 비즈니스 프로세스가 복잡화되고 대량화됨에 따라, 비즈니스 프로세스를 좀더 효율적으로 개선하고자 하는 데에 맞춰져 있으며, 이러한 동향에 힘입어 워크플로우 마이닝이라는 연구분야가 생겨나게 되었다. 이에, 본 논문에서는 워크플로우 패턴을 기반으로 한 워크플로우 마이닝 기법에 관하여 연구하고, 기술한다. 첫째로, 기본적으로 워크플로우 마이닝 기술과 그의 근원이 되는 워크플로우 로그에 관하여 간략히 설명한다. 그리고, 패턴기반 워크플로우 마이닝을 하기 위한 기본적인 워크플로우 패턴에 관하여 기술하고, 패턴 기반 워크플로우 마이닝을 하기 위한 기법으로 몇 가지 알고리즘을 제시한다.

• Journal of Computing Science and Engineering
• /
• v.11 no.4
• /
• pp.130-141
• /
• 2017

Feature selection is one of the most challenging problems of pattern recognition and data mining. In this paper, a feature selection algorithm based on an improved version of binary differential evolution is proposed. The method simultaneously optimizes two feature selection criteria, namely, set approximation accuracy of rough set theory and relational algebra based derived score, in order to select the most relevant feature subset from an entire feature set. Superiority of the proposed method over other state-of-the-art methods is confirmed by experimental results, which is conducted over seven publicly available benchmark datasets of different characteristics such as a low number of objects with a high number of features, and a high number of objects with a low number of features.

• International Journal of Industrial Entomology and Biomaterials
• /
• v.15 no.2
• /
• pp.115-122
• /
• 2007

As the genome of B.mori is available in GenBank and the EST database of B.mori is expanding, identification of novel genes of B.mori is conceivable by data-mining techniques. We used the in silico cloning method to get the vacuolar-type $H^+-ATP$ synthetase (V-ATPase) c subunit (16 kDa proteolipid subunit) gene of B.mori and analysed with bioinformatics tools. The result was confirmed by RT-PCR and sequencing. The V-ATPase c subunit cDNA contains a 468 bp ORF. The ORF encoded a 155-residue protein that showed extensive homology with V-ATPase c subunits from other 15 species and contained four membrane-spanning helices. Tissue expression pattern analysis revealed that V-ATPase c expressed strongly in Malpighian tubules, not in fat body. This gene has been registered in GenBank under the accession number EU082222.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2013.07a
• /
• pp.185-187
• /
• 2013

유비쿼터스 컴퓨팅이 생활의 일부가 되어가면서 정보의 양도 급속도로 늘어나고 있으며, 이로 인해 많은 데이터 속에서 정보를 찾아내는 기술이 부각되고 있다. 고객 기반의 협력적 필터링을 이용한 고객 선호도 예측 방법에서는 아이템에 대한 사용자의 선호도를 기반으로 이웃 선정 방법을 사용하므로 아이템에 대한 내용을 반영하지 못할 뿐만 아니라 희박성 문제를 해결하지 못하고 있다. 그리고 비슷한 선호도를 가진 일부 아이템의 정보를 바탕으로 하기 때문에 아이템의 속성은 무시하는 경향이 있다. 본 논문에서는 유비쿼터스 상거래에서 RFM(Recency, Frequency, Monetary) 분석 기반의 SOM을 이용한 군집방법을 제안한다. 제안 방법은 고객의 구매 데이터 기반의 유사한 속성의 데이터끼리의 클러스터링을 통해 보다 빠른 시간 내에 고객 성향에 맞는 추천이 가능한 구매 패턴 추출이 가능하다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.1
• /
• pp.107-113
• /
• 2003

A new GA-based methodology using information granules is suggested for the construction of fuzzy classifiers. The proposed scheme consists of three steps: selection of information granules, construction of the associated fuzzy sets, and tuning of the fuzzy rules. First, the genetic algorithm (GA) is applied to the development of the adequate information granules. The fuzzy sets are then constructed from the analysis of the developed information granules. An interpretable fuzzy classifier is designed by using the constructed fuzzy sets. Finally, the GA are utilized for tuning of the fuzzy rules, which can enhance the classification performance on the misclassified data (e.g., data with the strange pattern or on the boundaries of the classes). To show the effectiveness of the proposed method, an example, the classification of the Iris data, is provided.