다차원 색인을 이용한 밀도 기반 클러스터링의 근사적 접근 방법

황재준(Jae-Joon Hwang),황규영(Yang-Sae Moon),문양세(Kyu-Young Whang),장주현(Joo-Hyun Jang),김진호(Jin-Ho Kim) 한국정보과학회 2005 한국정보과학회 학술발표논문집 Vol.32 No.1

본 논문에서는 기존의 밀도 기반 전지 클러스터링 알고리즘의 성능을 개선한 밀도기 반 클러스터링의 근사적 접근법을 제안한다. 기존의 밀도 기반 전지 알고리즘은 다차원 색인의 많은 검색 공간을 빠르게 전지하면서도 원하는 클러스터를 정확히 찾아내는 특징을 가지고 있다. 그러나 기존 알고리즘은 전지를 위한 한계 값 설정을 위하여 단말 영역들의 밀도 값을 사용함으로써, 내부 영역에 속한 단말 영역들 간의 밀도 편차가 큰 경우 전지 여부에 대한 판별이 빨리 이루어지지 않는다. 또한, 최악의 경우에는 모든 단말 페이지를 검색하여야 하고, 이에 따라 성능이 저하될 수 있다. 반면에, 제안하는 근사적 접근법에서는 한계 값 설정을 위해 단말 영역이 아닌 내부 영역의 밀도 값을 사용한다. 일반적으로, 내부 영역들 간의 밀도 편차는 단말 영역들 간의 밀도 편차보다 크지 않으므로, 근사 밀도 기반 전지 알고리즘에서는 더욱 많은 검색 공간의 전지 여부의 빨리 판별할 수 있게 된다. 성능 평가 실험을 수행한 결과, 제안한 알고리즘은 기존의 알고리즘과 비교하여 정확성 측면에서는 큰 차이가 없는 반면 수행 시간 측면에서는 최대 17%의 성능 향상 효과가 있는 것으로 나타났다.

다차원 색인을 이용한 하향식 계층 클러스터링

황재준(Jae-Joon Hwang),문양세(Yang-Sae Moon),황규영(Kyu-Young Whang) 한국정보과학회 2002 정보과학회논문지 : 데이타베이스 Vol.29 No.5

최근 공간 데이타 분석, 영상 분석 등과 같은 대용량 데이타를 관리하는 다양한 응용 업무들이 증가함에 따라, 대용량의 데이타베이스를 위한 클러스터링 기법이 많이 연구되고 있다. 그 중에서도 계층클러스터링 기법은 데이타베이스의 계층 분할을 표현하는 계층 트리를 생성하고 이를 이용하여 효율적인 클러스터링을 수행하는 방법으로서, 지금까지는 주로 트리를 하위 계층으로부터 상위 계층으로 생성해 가는 상향식(bottom-up) 계층 클러스터링 기법들이 연구되었다. 이러한 상향식 클러스터링 방법은 트리를 생성하기 위하여 전체 데이타베이스를 한 번 이상 액세스하여야 할 뿐만 아니라, 하위 계층에서부터 검색을 시작하기 때문에 트리의 많은 부분을 검색하여야 하는 문제점이 있다. 본 논문에서는 대부분의 데이타베이스 응용에서 이미 유지하고 있는 다차원 색인을 이용하여 클러스터링을 수행하는 새로운 하향식(top-down) 계층 클러스터링 기법을 제안한다. 일반적으로 다차원 색인에서는 가까운 객체들이 동일한 (혹은 인접한) 페이지에 저장될 가능성이 큰 클러스터링 성질을 가진다. 이러한 다차원 색인의 클러스터링 성질을 사용하면 각 객체들간의 거리를 일일이 계산하지 않고도 이웃한 객체들을 식별할 수 있다. 우선 객체들의 밀도에 기반하여 클러스터를 정형적으로 정의한다. 이를 위하여, 객체를 포함하는 영역의 밀도를 이용한 영역 대조 분할(region contrast partition) 개념을 사용한다. 또, 클러스터링 알고리즘에서의 빠른 검색을 위하여 분기 한정(branch-and-bound) 알고리즘을 사용하며, 여기서의 한계값(bound)을 제안하고 이의 정확성을 이론적으로 증명한다. 실험 결과, 제안한 방법은 상향식 계층 클러스터링 방법인 BIRCH와 비교하여, 정확성 측면에서 우수하거나 유사한 것으로 나타났으며, 데이타 페이지 액세스 횟수를 데이타베이스 크기에 따라 최고 26~187배까지 감소시킨 것으로 나타났다. 이 같은 결과로 볼 때, 제안한 방법은 대용량 데이타베이스에서의 클러스터링 성능을 크게 향상시키는 기법으로서, 일반 데이타베이스 응용에 실용적으로 적용 가능하다고 판단된다. 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 pages. 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.

시뮬레이터의 복수 실체계 연동을 위한 리모트인터럽트 방식 게이트웨이 모델 구현 방안

권구형(Koo-Hyung Kwon),황재준(Jae-Joon Hwang),정현숙(Hyun-Sook Jeong),임원기(Won-Gi Lim),윤영득(Young-Deuk Yoon) 한국통신학회 2013 韓國通信學會論文誌 Vol.38 No.1(융합기술)

무선통신망 설계를 위한 네트워크 모델링에 관한 연구

권구형(Koo-Hyung Kwon),황재준(Jae-Joon Hwang),김주명(Ju-Myoung Kim) 한국통신학회 2010 한국통신학회 학술대회논문집 Vol.2010 No.2

식도암수술시 식도-위연결부 수기봉합과 기계봉합의 비교

신재승,최영호,김광택,황재준,Shin, Jae-Seung,Choi, Young-Ho,Kim, Kwang-Taik,Hwang, Jae-Joon 대한기관식도과학회 1999 大韓氣管食道科學會誌 Vol.5 No.2

Background : the technical modality of esophagogastric anastomosis plays an important part in the factors leading to anastomotic leakage and stricture. Objective : The objective of this study was to compare the leakage and stricture rates between the different anastomosis techniques. Materials and Methods : A retrospective study was undertaken in the patients with esophageal cancer who underwent a esophagectomy. Hand-sewn anastomosis was performed in one layer with interrupted sutures(Manual group). The stapler group was divided into two subgroups, such as PCEEA group and Endo-GIA group. Results : The differences of preoperative patient profiles between the groups were not significant. Overall mortality fates were 5% in manual group, 5% in PCEEA group and 11.5% in Endo-GIA group. Fewer anastomotic leakage occurred in manual group(5%) than in PCEEA group(10%) or in Endo-GIA group(15.4%). The postoperative stricture rate was higher in PCEEA group(35%) than in manual group(5%) or in Endo-GIA group(0%) Conclusion : Statistically, there were no significant differences in the anastomotic leakage and stricture rates between the hand-sewn and mechanical anastomosis. But the stricture rate was lower in the group using the linear stapling device and the leakage rate was lower in the hand-sewn group than the other groups.

VGG-16 딥러닝 알고리즘을 활용한 우식치아와 건전치아 분류

변민지 ( Min-ji Byon ),전은주 ( Eun-joo Jun ),김지수 ( Ji-soo Kim ),황재준 ( Jae-joon Hwang ),정승화 ( Seung-hwa Jeong ) 대한예방치과·구강보건학회 2021 大韓口腔保健學會誌 Vol.45 No.4

Objectives: Diagnosis of dental caries is based on the dentist’s observation and subjective judgment; therefore, a reliable and objective approach for diagnosing caries is required. Intraoral camera images combined with deep learning technology can be a useful tool to diagnose caries. This study aimed to evaluate the accuracy of the VGG-16 convolutional neural network (CNN) model in detecting dental caries in intraoral camera images. Methods: Images were obtained from the Internet and websites using keywords linked to teeth and dental caries. The 670 images that were obtained were categorized by an investigator as either sound (404 sound teeth) or dental caries (266 dental caries), and used in this study. The training and test datasets were divided in the ratio of 7:3 and a four-fold cross validation was performed. The Tensorflow-based Python package Keras was used to train and validate the CNN model. Accuracy, Kappa value, sensitivity, specificity, positive predictive value, negative predictive value, ROC (receiver operating characteristic) curve and AUC (area under curve) values were calculated for the test datasets. Results: The accuracy of the VGG-16 deep learning model for the four datasets, through random sampling, was between 0.77 and 0.81, with 0.81 being the highest. The Kappa value was 0.51- 0.60, indicating moderate agreement. The resulting positive predictive values were 0.77-0.82 and negative predictive values were 0.80-0.85. Sensitivity, specificity, and AUC values were 0.66-0.74, 0.81-0.88, and 0.88-0.91, respectively. Conclusions: The VGG-16 CNN model showed good discriminatory performance in detecting dental caries in intraoral camera images. The deep learning model can be beneficial in monitoring dental caries in the population.

고립성 폐 결절로 나타난 경화성 혈관종의 폐엽 절제술에 의한 치험 1예 : A case report

황은구,이송암,황재준,김상윤,권혁중,김요한 건국대학교 의과학연구소 2004 건국의과학학술지 Vol.13 No.-

Patient was sixty-eight years old female. She was transferred to our department to resect the solitary pulmonary nodule on right lower lobe. Intraoperative frozen-section revealed bronchioloalveolar carcinoma. So, we performed right lower lobectomy and mediastinal lymph node dissection. Permanent pathologic diagnosis of this patient was sclerosing hemangioma. We report experience of treatment of sclerosing hemagioma presenting solitary pulmonary nodule by lobectomy with the review of literatures.

대동맥 차단 및 우회에 따른 혈역학 변화 : 실험연구 An Experimental Study

이혜원,박성민,선경,김광택,김형묵,황재준,백광제,이송암 대한외상학회 2000 大韓外傷學會誌 Vol.13 No.1

Background: Cross-clamping the thoracic aorta is accompanied by dramatic changes in both proximal and distal hemodynamics. Although many studies focusing on the hypoperfusion effects distal to a clamp have been reported, only a few have noticed derangement of the proximal hemodynamics caused by the abrupt elevation of preload and afterload. The purpose of this study was to determine the effects of simple aortic cross-clamping on the hemodynamics of the proximal cardiovascular system and to compare them with those of a shunt bypass. Material and Method: Hemodynamic changes were examined in a canine model of descending thoracic aortic surgery. Sixteen mongrel dogs (19~25kg) were randomly divided into two groups: (1) a simple clamping group (CLAMP, n=7) and (2)a group of clamping with heparincoated shunt group(SHUNT, n=9). Parameters observed were the proximal and distal blood pressures, the heart rate, the left atrial pressure, the carotid blood flow, and the cardiac output. Each was measured serially and compared before clamping (baseline), immediately after clamping: 30 minutes after clamping, and after declamping. Result: In the CLAMP group, proximal pressures increased abruptly after clamping; from 117±15 to 177±17mmHg in systole, from 82 ±13 to 129±16mmHg in diastole, and from 96±11 to 147±14mmHg in mean (p$lt;0.001). The distal aorta lost its pulsation with a mean pressure of 20±1.5mmHg. The heart rate decreased from 125±23 to 103±20 (p$lt;0.05). The left atrial pressure increased from 8±2 to 14±2mmHg (p$lt;0.05), and the carotid blood flow increased from 120±58 to 291±131ml/min (p$lt;0.05). In SHUNT group, no significant changes were observed in any of the parameters, either after clamping or declamping (p=NS), and the distal aorta maintained its pulsatility with a mean pressure of above 70mmHg during the entire clamping period. In the both groups, the carotid blood flow showed a positive correlation to the proximal pressures and the left atrial pressure, and was inversely correlated to the heart rate. Conclusion: Shunt bypassing may reduce the deleterious effects of proximal hypertension and distal hypoperfusion during aortic cross-clamping. We conclude that a shunt bypass is a safe, mechanical adjuvant in descending aortic surgery.