• 융합신호처리학회논문지
• /
• 제20권3호
• /
• pp.132-137
• /
• 2019

의료영상으로 생성된 데이터의 양은 전문적인 시각적 분석 한계를 점점 초과하여, 자동화된 의료영상 분석의 필요성이 증가되고 있는 실정이다. 이러한 이유 등으로 인하여 본 논문에서는 정상소견과 종양소견을 보이는 각각의 뇌 실질 MRI 의료영상을 이용하여 Inception V3 딥러닝 모델을 이용한 종양 유무에 따른 분류 및 정확도를 평가하였다. 연구 결과, 딥러닝 모델의 정확도 평가는 학습 데이터 세트의 경우 90%, 검증 데이터 세트의 경우 86%의 정확도를 나타내었다. 손실률 평가에서는 학습 데이터 세트의 경우 0.56, 검증 데이터 세트의 경우 1.28의 손실률을 나타내었다. 향 후 연구에서는 딥러닝 모델의 성능 향상 및 평가의 신뢰성 확보를 위하여 공개된 의료영상의 데이터를 충분히 확보하고, 라벨링 분류 작업을 통한 라벨링의 정확도를 개선하여 모델링을 구현해 볼 필요가 있다고 사료된다.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2019년도 학술발표회
• /
• pp.155-155
• /
• 2019

The Agricultural Policy/Environmental eXtender (APEX) models have been developed for assessing agricultural management efforts and their effects on soil and water at the field scale as well as more complex multi-subarea landscapes, whole farms, and watersheds. National Academy of Agricultural Sciences, Wanju, Korea, has modified a key component of APEX application, named APEX-Paddy for simulating water quality with considering appropriate paddy management practices, such as puddling and flood irrigation management. Calibration and validation are an anticipated step before any model application. Simple techniques are essential to assess whether or not a parameter should be adjusted for calibration. However, very few study has been done to evaluate the ability of APEX-Paddy to simulate the impact of multiple management scenarios on nutrients loss. In this study, the observation data from experimental fields at Iksan in South Kora was used in calibration and evaluation process during 2013-2015. The APEX auto- calibration tool (APEX-CUTE) was used for model calibration and sensitivity analysis. Four quantitative statistics, the coefficient of determination ($R^2$),Nash-Sutcliffe(NSE),percentbias(PBIAS)androotmeansquareerror(RMSE)were used in model evaluation. In this study, the hydrological process of the modified model, APEX-Paddy, is being calibrated and tested in predicting runoff discharge rate and nutrient yield. Field-scale calibration and validation processes are described with an emphasis on essential calibration parameters and direction regarding logical sequences of calibration steps. This study helps to understand the calibration and validation way is further provided for applications of APEX-Paddy at the field scales.

• Biomaterials and Biomechanics in Bioengineering
• /
• 제2권2호
• /
• pp.111-125
• /
• 2015

For engineers, generating a mesh in porous media (PMs) sometimes represents a smaller computational load than generating realistic stent geometries with computer fluid dynamics (CFD). For this reason, PMs have recently become attractive to mimic flow-diverter stents (FDs), which are used to treat intracranial aneurysms. PMs function by introducing a hydraulic resistance using Darcy's law; therefore, the pressure drop may be computed by test sections parallel and perpendicular to the main flow direction. However, in previous studies, the pressure drop parallel to the flow may have depended on the width of the gap between the stent and the wall of the test section. Furthermore, the influence of parameters such as the test section geometry and the distance over which the pressure drops was not clear. Given these problems, computing the pressure drop parallel to the flow becomes extremely difficult. The aim of the present study is to resolve this lack of information for stent modeling using PM and to compute the pressure drop using several methods to estimate the influence of the relevant parameters. To determine the pressure drop as a function of distance, an FD was placed parallel and perpendicular to the flow in test sections with rectangular geometries. The inclined angle method was employed to extrapolate the flow patterns in the parallel direction. A similar approach was applied with a cylindrical geometry to estimate loss due to pipe friction. Additionally, the pressure drops were computed by using CFD. To determine if the balance of pressure drops (parallel vs perpendicular) affects flow patterns, we calculated the flow patterns for an ideal aneurysm using PMs with various ratios of parallel pressure drop to perpendicular pressure drop. The results show that pressure drop in the parallel direction depends on test section. The PM thickness and the ratio of parallel permeability to perpendicular permeability affect the flow pattern in an ideal aneurysm. Based on the permeability ratio and the flow patterns, the pressure drop in the parallel direction can be determined.

• 대한조선학회논문집
• /
• 제43권1호
• /
• pp.119-127
• /
• 2006

Recently as a result of excessive development. pollution of the coast and occurrence of a typhoon year after year, fishermen suffer heavy losses in fish farming which is the one of the most important earnings ways. For solution of these problems, we need to go out into the open sea from an inland sea. In this study we suggested new fish cage which makes up for the structural weakness of existing wooden fish cages. It can farm fishes in the open sea of high wave and current with no damages from a typhoon. We substituted TPU(Thermoplastic Polyurethane) air chamber for existing styrofoam flotage which was harmful to the environment and impermanent. We used PE(Polyethylene) pipes for the maintenance of formation and the prevention of buoyancy loss caused by a breakdown of flotage. PE b rackets were designed to combine PE pipes with TPU air-chamber flotage. It has good strength and light weight. As a result of modeling test. it is great in buoyancy, strength and flexibility against wave. Because it can control buoyancy arbitrarily, moreover, we expect that it will reduce damages of a red water by applying it as semi-submerged fish cages.

• 한국군사과학기술학회지
• /
• 제24권5호
• /
• pp.482-491
• /
• 2021

It is obvious that understanding the effects of shallow water environment of Korea is very important to guarantee the optimal performance of active sonar such as monostatic and bistatic sonar. For this reason, in this paper, we analyzed the detection performance characteristics for various depth deployments of sonar in summer, winter and water temperature inversion environments, which environments are frequently observed in shallow water of Korea such as the Yellow sea. To analyze only effects of water temperature structures on target detection performance, we applied range independent conditions for bottom, sea surface and water temperature characteristics. To understand the characteristics of detection performance, we conducted transmission loss and signal excess modeling. From the results, we were able to confirm the characteristics of detection performance of active sonar. In addition, we verified that operation depth of transmitter and receiver affects the detection performance. Especially in the water temperature inversion environment, it was confirmed that the shadow zone could be minimized and the detection range could be increased through bistatic operation.

• 정보보호학회논문지
• /
• 제29권1호
• /
• pp.187-194
• /
• 2019

2014년 3월, 세계 최대의 비트코인 거래소였던 마운트곡스(Mt. Gox)가 해킹 공격으로 폐쇄된 사건 이래로 최근까지 국내 암호 화폐 거래소인 코인레일(Coinrail)이 해킹되는 등 사건이 잇달아 발생하고 있다. 이러한 거래소 해킹 사건은 단순한 시스템 해킹 수준을 넘어 사용자들의 자산이 탈취되는 자산 손실로까지 피해가 확산되고 있어, 암호 화폐 거래소에 대한 보안 이슈가 발생하였다. 위와 같은 문제를 해결하기 위해 탈중앙화 거래소(DEX, Decentralized Exchange)가 활발히 연구되고 있으나 이 또한 문제를 완화시킬 뿐 해결방안으로서는 부족한 실정이다. 따라서 본 논문에서는 기존의 암호 화폐 거래소들에 대한 보안위협을 분석하고 이에 대한 보안 요구사항을 도출한다. 또한 개인용 보안장치를 통한 안전한 분산형 암호 화폐 거래 모델을 제안하여 본 논문에서 제안하는 거래 모델이 앞선 보안위협에 대한 해결책임을 입증한다.

• Journal of Ecology and Environment
• /
• 제42권4호
• /
• pp.298-307
• /
• 2018

Background: Subalpine ecosystems at high altitudes and latitudes are particularly sensitive to climate change. In South Korea, the prediction of the species richness of subalpine plant species under future climate change is not well studied. Thus, this study aims to assess the potential impact of climate change on species richness of subalpine plant species (14 species) in the 17 mountain national parks (MNPs) of South Korea under climate change scenarios' representative concentration pathways (RCP) 4.5 and RCP 8.5 using maximum entropy (MaxEnt) and Migclim for the years 2050 and 2070. Results: Altogether, 723 species occurrence points of 14 species and six selected variables were used in modeling. The models developed for all species showed excellent performance (AUC > 0.89 and TSS > 0.70). The results predicted a significant loss of species richness in all MNPs. Under RCP 4.5, the range of reduction was predicted to be 15.38-94.02% by 2050 and 21.42-96.64% by 2070. Similarly, under RCP 8.5, it will decline 15.38-97.9% by 2050 and 23.07-100% by 2070. The reduction was relatively high in the MNPs located in the central regions (Songnisan and Gyeryongsan), eastern region (Juwangsan), and southern regions (Mudeungsan, Wolchulsan, Hallasan, and Jirisan) compared to the northern and northeastern regions (Odaesan, Seoraksan, Chiaksan, and Taebaeksan). Conclusions: This result indicates that the MNPs at low altitudes and latitudes have a large effect on the climate change in subalpine plant species. This study suggested that subalpine species are highly threatened due to climate change and that immediate actions are required to conserve subalpine species and to minimize the effect of climate change.

• 한국임상수의학회지
• /
• 제36권1호
• /
• pp.23-29
• /
• 2019

While research findings suggest that the highly pathogenic avian influenza (HPAI) is the leading cause of economic loss in Korean poultry industry with an estimated cumulative impact of $909 million since 2003, identifying the environmental and anthropogenic risk factors involved remains a challenge. The objective of this study was to identify areas at high risk for potential HPAI outbreaks according to the likelihood of HPAI virus detection in wild birds. This study integrates spatial information regarding HPAI surveillance with relevant demographic and environmental factors collected between 2003 and 2018. The Maximum Entropy (Maxent) species distribution modeling with presence-only data was used to model the spatial risk of HPAI virus. We used historical data on HPAI occurrence in wild birds during the period 2003-2018, collected by the National Quarantine Inspection Agency of Korea. The database contains a total of 1,065 HPAI cases (farms) tied to 168 unique locations for wild birds. Among the environmental variables, the most effective predictors of the potential distribution of HPAI in wild birds were (in order of importance) altitude, number of HPAI outbreaks at farm-level, daily amount of manure processed and number of wild birds migrated into Korea. The area under the receiver operating characteristic curve for the 10 Maxent replicate runs of the model with twelve variables was 0.855 with a standard deviation of 0.012 which indicates that the model performance was excellent. Results revealed that geographic area at risk of HPAI is heterogeneously distributed throughout the country with higher likelihood in the west and coastal areas. The results may help biosecurity authority to design risk-based surveillance and implementation of control interventions optimized for the areas at highest risk of HPAI outbreak potentials.

• 한국인터넷방송통신학회논문지
• /
• 제21권1호
• /
• pp.87-92
• /
• 2021

빅데이터 기반의 추천시스템 모델링에서 바이어스, 분산, 오류 및 학습은 성능에 중요한 요소이다. 이러한 시스템에서는 추천 모델이 설명도를 유지하면서 복잡도를 줄여야 한다. 또한 데이터의 희소성과 시스템의 예측은 서로 반비례의 속성을 가지기 마련이다. 따라서 희소성의 데이터를 인수분해 방법을 활용하여 상품간의 유사성을 학습을 통한 상품추천모델이 제안되어 왔다. 본 논문에서는 이 모델의 손실함수에 대한 최적화 방안으로 max-norm 규제를 적용하여 모델의 일반화 능력을 향상시키고자 한다. 해결방안은 기울기를 투영하는 확률적 투영 기울기 강하법을 적용하는 것이다. 많은 실험을 통하여 데이터가 희박해질수록 기존의 방법에 비해 제안된 규제 방법이 상대적으로 효과가 있다는 것을 확인하였다.

• 산업경영시스템학회지
• /
• 제44권4호
• /
• pp.12-22
• /
• 2021

This article suggests the machine learning model, i.e., classifier, for predicting the production quality of free-machining 303-series stainless steel(STS303) small rolling wire rods according to the operating condition of the manufacturing process. For the development of the classifier, manufacturing data for 37 operating variables were collected from the manufacturing execution system(MES) of Company S, and the 12 types of derived variables were generated based on literature review and interviews with field experts. This research was performed with data preprocessing, exploratory data analysis, feature selection, machine learning modeling, and the evaluation of alternative models. In the preprocessing stage, missing values and outliers are removed, and oversampling using SMOTE(Synthetic oversampling technique) to resolve data imbalance. Features are selected by variable importance of LASSO(Least absolute shrinkage and selection operator) regression, extreme gradient boosting(XGBoost), and random forest models. Finally, logistic regression, support vector machine(SVM), random forest, and XGBoost are developed as a classifier to predict the adequate or defective products with new operating conditions. The optimal hyper-parameters for each model are investigated by the grid search and random search methods based on k-fold cross-validation. As a result of the experiment, XGBoost showed relatively high predictive performance compared to other models with an accuracy of 0.9929, specificity of 0.9372, F₁-score of 0.9963, and logarithmic loss of 0.0209. The classifier developed in this study is expected to improve productivity by enabling effective management of the manufacturing process for the STS303 small rolling wire rods.