• 정보보호학회논문지
• /
• 제31권6호
• /
• pp.1193-1204
• /
• 2021

정보보안에서 AI 기술은 알려지지 않은 악성코드를 탐지하기 위해 사용한다. AI 기술은 높은 정확도를 보장하지만, 오탐을 필연적으로 수반하므로 AI가 예측한 결과를 해석하기 위해 XAI 도입을 고려하고 있다. 그러나, XAI는 단순한 해석결과만 제공할 뿐 그 해석을 평가하거나 검증하는 XAI 평가 연구는 부족하다. XAI 평가는 어떤 기술이 더 정확한지 안전성 확보를 위해 필수적이다. 본 논문에서는 악성코드 분야에서 AI 예측에 크게 기여한 feature로 AI 결과를 해석하고, 이러한 AI 결과해석에 대한 평가방안을 제시한다. 약 94%의 정확도를 보이는 tree 기반의 AI 모델에 두 가지 XAI 기술을 사용하여 결과해석을 진행하고, 기술 정확도 및 희소성을 분석하여 AI 결과해석을 평가한다. 실험 결과 AI 결과해석이 적절하게 산출되었음을 확인하였다. 향후, XAI 평가로 인해 XAI 도입 및 활용은 점차 증가하고, AI 신뢰성 및 투명성이 크게 향상될 것으로 예상한다.

• 보건교육건강증진학회지
• /
• 제23권4호
• /
• pp.13-28
• /
• 2006

Objectives: This study used the characteristics of the knowledge discovery and data mining algorithms to develop hypertension predictive model for hypertension management using the Korea National Health Insurance Corporation database(the insureds' screening and health care benefit data). Methods: This study validated the predictive power of data mining algorithms by comparing the performance of logistic regression, decision tree, and ensemble technique. On the basis of internal and external validation, it was found that the model performance of logistic regression method was the best among the above three techniques. Results: Major results of logistic regression analysis suggested that the probability of hypertension was: - lower for the female(compared with the male)(OR=0.834) - higher for the persons whose ages were 60 or above(compared with below 40)(OR=4.628) - higher for obese persons(compared with normal persons)(OR= 2.103) - higher for the persons with high level of glucose(compared with normal persons)(OR=1.086) - higher for the persons who had family history of hypertension(compared with the persons who had not)(OR=1.512) - higher for the persons who periodically drank alcohol(compared with the persons who did not)$(OR=1.037{\sim}1.291)$ Conclusions: This study produced several factors affecting the outbreak of hypertension using screening. It is considered to be a contributing factor towards the nation's building of a Hypertension Management System in the near future by bringing forth representative results on the rise and care of hypertension.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2020년도 학술발표회
• /
• pp.85-85
• /
• 2020

South Korea experiences few large scale erosion and sedimentation problems, however, there are numerous local sedimentation problems. A reliable and consistent approach to modelling and management for sediment processes are desirable in the country. In this study, field measurements of sediment concentration from 34 alluvial river basins in South Korea were used with the Modified Einstein Procedure (MEP) to determine the total sediment load at the sampling locations. And then the Flow Duration-Sediment Rating Curve (FD-SRC) method was used to estimate the specific degradation for all gauging stations. The specific degradation of most rivers were found to be typically 50-300 tons/㎢·yr. A model tree data mining technique was applied to develop a model for the specific degradation based on various watershed characteristics of each watershed from GIS analysis. The meaningful parameters are: 1) elevation at the middle relative area of the hypsometric curve [m], 2) percentage of wetland and water [%], 3) percentage of urbanized area [%], and 4) Main stream length [km]. The Root Mean Square Error (RMSE) of existing models is in excess of 1,250 tons/㎢·yr and the RMSE of the proposed model with 6 additional validations decreased to 65 tons/㎢·yr. Erosion loss maps from the Revised Universal Soil Loss Equation (RUSLE), satellite images, and aerial photographs were used to delineate the geospatial features affecting erosion and sedimentation. The results of the geospatial analysis clearly shows that the high risk erosion area (hill slopes and construction sites at urbanized area) and sedimentation features (wetlands and agricultural reservoirs). The result of physiographical analysis also indicates that the watershed morphometric characteristic well explain the sediment transport. Sustainable management with the data mining methodologies and geospatial analysis could be helpful to solve various erosion and sedimentation problems under different conditions.

• Archives of Plastic Surgery
• /
• 제50권6호
• /
• pp.627-634
• /
• 2023

In recent decades, a number of simulation models for microsurgical training have been published. The human placenta has received extensive validation in microneurosurgery and is a useful instrument to facilitate learning in microvascular repair techniques as an alternative to using live animals. This study uses a straightforward, step-by-step procedure for instructing the creation of simulators with dynamic flow to characterize the placental vascular tree and assess its relevance for plastic surgery departments. Measurements of the placental vasculature and morphological characterization of 18 placentas were made. After the model was used in a basic microsurgery training laboratory session, a survey was given to nine plastic surgery residents, two microsurgeons, and one hand surgeon. In all divisions, venous diameters were larger than arterial diameters, with minimum diameters of 0.8 and 0.6 mm, respectively. The majority of the participants considered that the model faithfully reproduces a real microsurgical scenario; the consistency of the vessels and their dissection are similar in in vivo tissue. Furthermore, all the participants considered that this model could improve their surgical technique and would propose it for microsurgical training. As some of the model's disadvantages, an abundantly thick adventitia, a thin tunica media, and higher adherence to the underlying tissue were identified. The color-perfused placenta is an excellent tool for microsurgical training in plastic surgery. It can faithfully reproduce a microsurgical scenario, offering an abundance of vasculature with varying sizes similar to tissue in vivo, enhancing technical proficiency, and lowering patient error.

• 한국조경학회지
• /
• 제43권6호
• /
• pp.16-24
• /
• 2015

본 연구는 국내 4대강 유역에 조성된 수변녹지를 대상으로 탄소의 저장 및 연간 흡수를 계량화하고, 수변녹지의 탄소저감 효과를 증진하기 위한 조성방안을 모색하였다. 표본 선정한 40개소 연구 대상지의 녹지구조 및 식재기법은 흉고직경이 평균 $6.9{\pm}0.2cm$이고 식재밀도가 $10.4{\pm}0.8$주/$100m^2$로서, 소형 수목의 저밀 단층식재로 대표된다. 식재수목에 의한 단위면적당 탄소의 저장량과 연간 흡수량은 각각 평균 $8.2{\pm}0.5t/ha$, $1.7{\pm}0.1t/ha$/년이고, 식재밀도가 높을수록 증가하는 경향을 보였다. 토양의 유기물함량과 단위면적당 탄소저장량은 각각 $1.4{\pm}0.1%$, $26.4{\pm}1.5t/ha$이었다. 대상지의 수목과 토양은 1ha당 약 61kL의 휘발유 소비에 상당하는 탄소량을 저장하고, 수목은 해마다 1ha당 3kL의 휘발유 소비에 기인한 탄소배출량을 상쇄하는 효과를 나타냈다. 이 탄소저감은 식재 후 5년 이상~10년 미만 생장한 효과로서 식재수목의 생장과 더불어 훨씬 더 증가할 것으로 예측된다. 연구 대상지와 상이한 식재기법의 조성모델들을 선정하여 향후 30년 동안 수목생장에 따른 연간 탄소흡수량의 변화를 비교 시뮬레이션하였다. 그 결과, 경과년도별 누적 탄소흡수량은 식재규격이 더 크고 식재밀도가 더 높은 다층 군식의 생태식재모델에서 저밀 단층식재인 대상지보다 10년 및 30년 경과시 각각 약 1.9배, 1.5배 더 많았다. 수변녹지의 탄소저감 효과를 증진하기 위해서는 규격이 상대적으로 큰 수목을 혼식하는 다층 군식, 속성수를 포함하여 연간 생장률이 양호한 자생수종의 중 고밀 식재, 식재수종의 정상적 생장에 적합한 토양조건 구비 등이 요구된다. 본 연구결과는 조성 초기단계인 수변녹지 사업에서 수질보전 및 생물서식에 부가하여 탄소흡수원의 역할을 제고하기 위한 실용적 지침이 될 것으로 기대한다.

• 한국산학기술학회논문지
• /
• 제19권11호
• /
• pp.126-136
• /
• 2018

본 연구는 머신러닝을 활용하여 급성 뇌졸중 퇴원 환자의 중증도 보정 사망 예측 모형 개발을 목적으로 시행하였다. 전국 단위의 퇴원손상심층조사 2006~2015년 자료 중 한국표준질병사인분류(Korean standard classification of disease-KCD 7)에 따라 뇌졸중 코드 I60-I63에 해당하는 대상자를 추출하여 분석하였다. 동반질환 중증도 보정 도구로는 Charlson comorbidity index(CCI), Elixhauser comorbidity index(ECI), Clinical classification software(CCS)의 3가지 도구를 사용하였고 중증도 보정 모형 예측 개발은 로지스틱회귀분석, 의사결정나무, 신경망, 서포트 벡터 머신 기법을 활용하여 비교해 보았다. 뇌졸중 환자의 동반질환으로는 ECI에서는 합병증을 동반하지 않은 고혈압(hypertension, uncomplicated)이 43.8%로, CCS에서는 본태성고혈압(essential hypertension)이 43.9%로 다른 질환에 비해 가장 월등하게 높은 것으로 나타났다. 동반질환 중중도 보정 도구를 비교해 본 결과 CCI, ECI, CCS 중 CCS가 가장 높은 AUC값으로 분석되어 가장 우수한 중증도 보정 도구인 것으로 확인되었다. 또한 CCS, 주진단, 성, 연령, 입원경로, 수술유무 변수를 포함한 중증도 보정 모형 개발 AUC값은 로지스틱 회귀분석의 경우 0.808, 의사결정나무 0.785, 신경망 0.809, 서포트 벡터 머신 0.830로 분석되어 가장 우수한 예측력을 보인 것은 서포트 벡터머신 기법인 것으로 최종 확인되었고 이러한 결과는 추후 보건의료정책 수립에 활용될 수 있을 것이다.

• 품질경영학회지
• /
• 제39권3호
• /
• pp.391-399
• /
• 2011

Our study was carried out to analyze the variation factors of severity-adjusted length of stay(LOS) in coronary artery bypass graft(CABG). The subjects were 932 CABG inpatients of the Korean National Hospital Discharge In-depth Injury Survey from 2004 through 2008. The data were analyzed using $x^2$ test and the severity-adjusted model was developed using data mining technique. The results of the study were as follows: male(71.1%), older than 61 years of age(61.6%), more than 500 beds(92.8%) and admitting via ambulatory care(70.0%) appeared to have higher rate than otherwise. In-hospital mortality of CABG inpatients was 2.8%. In addition, 46.4% of the patients received their care in other residence. The angina pectoris(45.6%) was found to be the highest in principle diagnosis, followed by chronic ischemic heart disease(36.9%) and acute myocardial infarction(12.0%). We developed severity-adjusted LOS model using the variables such as gender, age and comorbidity. Comparison of adjusted values in predicted LOS revealed that there were significant variations in LOS by location of hospital, bed size, and whether patients received the care in their residences. The variations of LOS can be explained as the indirect indicator for quality variation of medical process. It is suggested that the severity-adjusted LOS model developed in this study should be utilized as a useful method for benchmarking in hospital and it is necessary that national standard clinical practice guideline should be developed.

• Computers and Concrete
• /
• 제19권6호
• /
• pp.651-658
• /
• 2017

High-performance concrete, besides aggregate, cement, and water, incorporates supplementary cementitious materials, such as fly ash and blast furnace slag, and chemical admixture, such as superplasticizer. Hence, it is a highly complex material and modeling its behavior represents a difficult task. This paper presents an evolutionary system for the prediction of high performance concrete strength. The proposed framework blends a recently developed version of genetic programming with a local search method. The resulting system enables us to build a model that produces an accurate estimation of the considered parameter. Experimental results show the suitability of the proposed system for the prediction of concrete strength. The proposed method produces a lower error with respect to the state-of-the art technique. The paper provides two contributions: from the point of view of the high performance concrete strength prediction, a system able to outperform existing state-of-the-art techniques is defined; from the machine learning perspective, this case study shows that including a local searcher in the geometric semantic genetic programming system can speed up the convergence of the search process.

• 한국복합재료학회:학술대회논문집
• /
• 한국복합재료학회 2005년도 춘계학술발표대회 논문집
• /
• pp.172-175
• /
• 2005

The numerical simulation is performed for the acoustic emission and the wave propagation due to fiber breakage in single fiber composite plates by the finite element transient analysis. The acoustic emission and the following wave motions from a fiber breakage under a static loading is simulated to investigate the applicability of the explicit finite element method and the equivalent volume force model as a simulation tool of wave propagation and a modeling technique of an acoustic emission. For such a simple case of the damage event under static loading, various parameters affecting the wave motion are investigated for reliable simulations of the impact damage event. The high velocity and the small wave length of the acoustic emission require a refined analysis with dense distribution of the finite element and a small time step. In order to fulfill the requirement for capturing the exact wave propagation and to cover the 3-D simulation, we utilize the parallel FE transient analysis code and the parallel computing technology.

• 말소리와 음성과학
• /
• 제9권1호
• /
• pp.19-26
• /
• 2017

The main objective of this study is to investigate the influence of utterance internal pause in English speaking evaluation. To avoid possible confusion with other errors caused by segmental and prosodic inaccuracy, stem utterances with two different length obtained from a native speaker were manipulated to make a set of stimuli tokens through insertion of pauses whose length and position vary. After a total of 90 participants classified into three proficiency groups rated the stimuli, the scored data set was statistically analyzed in terms of the mixed effects model. It was confirmed that predictors such as pause length, pause position and utterance length significantly influence raters' evaluation scores. Especially, a dominating effect was found in such a way that raters gradually deducted scores in accordance with the increase of pause duration. In another experiment, a tree-based statistical learning technique was utilized to check which of the significant predictors played a more influential role than others. The findings in this paper are expected to be practically informative for both the test takers who are preparing for an English speaking test and the raters who desire to develop more objective rubric of speaking evaluation.