• KIISE Transactions on Computing Practices
• /
• v.22 no.12
• /
• pp.646-653
• /
• 2016

Outbound market is a rapidly growing global industry, and has evolved into a 11 trillion won trade. A lot of recommender systems, which are based on collaborative and content filtering, target the existing purchase log or rely on studies based on similarity of products. These researches are not highly efficient as data was not obtained in advance, and acquiring the overwhelming amount of data has been relatively slow. The characteristics of an outbound product are that it should be purchased at least twice in a year, and its pricing should be in the higher category. Since the repetitive purchase of a product is rare for the outbound market, the old recommender system which profiles the existing customers is lacking, and has some limitations. Therefore, due to the scarcity of data, we suggest an improved customer-profiling method using web usage mining, algorithm of association rule, and rule-based algorithm, for faster recommender system of outbound product.

• Science of Emotion and Sensibility
• /
• v.18 no.2
• /
• pp.3-18
• /
• 2015

Numerosity perception is considered as an innate ability of human being where its sensivitiy may widely vary across each individual person. The present study explored the relationship between visual working memory (VWM), visual search efficiency, and numerosity sensitivity. To accomplish this, we calculated each participant's K-value from change detection performance representing one's storage capacity in VWM, slopes of search RTs representing the search efficiency, and discrimination sensitivity for a quantity difference across two sets of dot arrays representing the numerosity sensitivity. The correlational analysis across the measurements revealed that participants with a high VWM capacity better discriminated the numerosity difference in the arrays when the spatial information in the two dot arrays was preserved. In contrast, the participants with high search efficiency discriminated better the difference in the arrays when the spatial information in the arrays was not preserved. The results indicate high VWM-capacity individuals were presumably able to use a strategy of storing the dot arrays by grouping them into a smaller pattern of dot arrays while high search-efficiency individuals were able to use a strategy of rapidly switching their focused attention across the dots in the arrays to count each individual dot. These in sum suggest that individual differences in numerosity sensitivity rely on one's working memory capacity as well as their efficient use of switching focused attention.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.31 no.4
• /
• pp.259-267
• /
• 2013

GIS-based web map service is all the more accessible to the public. Among others, location query services are most frequently utilized, which are currently restricted to only one keyword search. Although there increases the demand for the service for querying multiple keywords corresponding to sequential activities(banking, having lunch, watching movie, and other activities) in various locations POI, such service is yet to be provided. The objective of the paper is to develop the k-IPS algorithm for quickly and accurately querying multiple POIs that internet users input and locating the search outcomes on a web map. The algorithm is developed by utilizing hierarchical tree structure of $R^*$-tree indexing technique to produce overlapped geometric regions. By using recursive $R^*$-tree index based spatial join process, the performance of the current spatial join operation was improved. The performance of the algorithm is tested by applying 2, 3, and 4 multiple POIs for spatial query selected from 159 keyword set. About 90% of the test outcomes are produced within 0.1 second. The algorithm proposed in this paper is expected to be utilized for providing a variety of location-based query services, of which demand increases to conveniently support for citizens' daily activities.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.8
• /
• pp.64-71
• /
• 2017

A damage estimation method is presented for jacket-type support structure of offshore wind turbine using a change of modal properties due to damage and committee of neural networks for effective structural health monitoring. For more practical monitoring, it is necessary to monitor the critical and prospective damaged members with a limited number of measurement locations. That is, many data channels and sensors are needed to identify all the members appropriately because the jacket-type support structure has many members. This is inappropriate considering economical and practical health monitoring. Therefore, intensive damage estimation for the critical members using a limited number of the measurement locations is carried out in this study. An analytical model for a jacket-type support structure which can be applied for a 5 MW offshore wind turbine is established, and a training pattern is generated using the numerical simulations. Twenty damage cases are estimated using the proposed method. The identified damage locations and severities agree reasonably well with the exact values and the accuracy of the estimation can be improved by applying the committee of neural networks. A verification experiment is carried out, and the damage arising in 3 damage cases is reasonably identified.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.22 no.3
• /
• pp.239-248
• /
• 2020

Exact rock classification helps suitable support patterns to be installed. Face mapping is usually conducted to classify the rock mass using RMR (Rock Mass Ration) or Q values. There have been several attempts to predict the grade of rock mass using mechanical data of jumbo drills or probe drills and photographs of excavation surfaces by using deep learning. However, they took long time, or had a limitation that it is impossible to grasp the rock grade in ahead of the tunnel surface. In this study, a method to predict the Q value ahead of excavation surface is developed using recurrent neural network (RNN) technique and it is compared with the Q values from face mapping for verification. Among Q values from over 4,600 tunnel faces, 70% of data was used for learning, and the rests were used for verification. Repeated learnings were performed in different number of learning and number of previous excavation surfaces utilized for learning. The coincidence between the predicted and actual Q values was compared with the root mean square error (RMSE). RMSE value from 600 times repeated learning with 2 prior excavation faces gives a lowest values. The results from this study can vary with the input data sets, the results can help to understand how the past ground conditions affect the future ground conditions and to predict the Q value ahead of the tunnel excavation face.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2011.05a
• /
• pp.456-456
• /
• 2011

국내에서는 1999년부터 국가수자원관리종합정보시스템(WAMIS) 구축 및 운영을 통해 물관리정보화사업을 추진하고 있으며 기초정보분석 중심에서 수자원 계획 수립과 정책 결정을 지원할 수 있는 시스템으로 확대하여 대국민 수자원 정보 제공과 홍보를 활성화하려고 노력하고 있다. 하지만 지금까지 국내에서는 수자원 현황을 평가하기 위해 상수도보급율, 하천개수율 등을 이용하여 수자원의 단편적인 분야만을 평가하여 왔으며 이들 개별지표들만으로 국내 수자원 상황에 대해 국민들이 체감하는데 한계가 있어 이들 사업의 성과를 지수화하여 수자원 정책과 사업의 효과를 국민들에게 적극적으로 홍보할 필요가 있다. 이를 지원하기 위해 2007년부터 2009년까지 수자원 각 분야별 수자원계획수립 업무지원체계를 구축하였으며 수자원 현황 평가를 위해 분야별 평가지수를 개발하고 수자원 통합지수를 선정하여 중권역별로 산정한 바 있다(건설교통부, 2007; 국토해양부, 2008, 2009). 보다 합리적인 수자원 평가를 위해서는 분야별(물이용, 치수, 하천환경) 평가 지수의 공간적 범위 확대 및 세부지표를 추가 고려함으로써 과거에서부터 현재까지의 분야별 변화 패턴을 파악해야 한다. 이를 통해 수자원 관련 정책 및 사업의 성과를 평가하고 구축된 기초자료 및 분석정보를 제공해 줄 수 있는 도구의 개발이 무엇보다 중요하다고 하겠다. 따라서 본 연구에서는 기 개발된 분야별 평가지수 산정 결과와 분야별 현황과의 비교 분석을 통해 지수의 현장 적용성을 검토함으로써 기 개발된 지표의 취약점 및 한계점을 제시하였다. 보다 합리적이고 타당한 분야별 평가를 위해 세부지표를 추가로 선정하였으며 선정된 분야별 세부지표를 PSR 구성체계에 맞추어 구성하였다. 또한 분야별 현황 및 특성을 평가할 수 있는 분야별 평가지수를 개발하였으며 물이용 특성을 평가할 수 있는 지수를 '물이용안전성지수', 치수 특성을 평가할 수 있는 지수를 '홍수안전성지수', 하천환경 특성을 평가할 수 있는 지수를 '하천환경건강성지수'라 명명하였다. 또한 분야별 평가지수를 통합하여 수자원 현황을 평가하고 관리할 수 있는 수자원 통합지수인 '물만족지수'를 개발하여 제시하였다. 분야별 평가지수를 구성하는 각 세부지표의 특성을 파악하여 지표 산정 범위를 점, 선, 면으로 확대하여 제시하였으며 세분화된 공간단위별로 기초자료를 조사, 수집하여 시계열 DB를 구축하였다. 개발된 분야별 평가지수 및 물만족지수를 연도별 표준유역별로 구축된 DB를 이용하여 산정하고 비교 분석하였으며 상대적인 분야별 안전성 및 건강성을 평가하여 지수의 적용성을 검토하였다. 지표 관련 기초자료 및 분석을 통해 생성된 정보자료는 수자원정책 수립과정에서 매우 유용한 정보를 제공해 줌으로써 정책결정을 지원할 수 있고, 일반인과 관련 전문가들에게 수자원 관련 다양한 정보를 제공할 수 있다.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2017.05a
• /
• pp.211-211
• /
• 2017

장기유출해석 모델은 수자원의 안정적인 확보와 이용, 유역단위 기초자료 조사관리 등을 위하여 수자원 장기종합계획 및 전국유역조사사업 등에 활용되고 있다. 주로 국외에서 개발된 모형이 활용되고 있어, 국내의 여건에 맞추어 편의성이 개선된 모형을 찾는 것은 매우 어려운 일이다. 또한, 유출해석을 수행하기에 앞서 지속적으로 업데이트된 모델에 대한 객관적인 평가를 수행한 사례는 드물다. 따라서, 본 연구에서는 국내에서 주로 활용되고 있는 장기유출해석모델(TANK, SWAT, SSARR, PRMS 등)에 대한 비교검토를 토대로 각종 사업과의 연계성, 계산의 효율성, 정확도 등을 고려하여 USGS에서 개발한 PRMS v.4.0.2를 기반으로 국내유역에 활용이 가능하도록 개선한 $K-BASIN^{RR}$ 및 입력자료 전처리기를 개발하였다. PRMS 모형은 융설 및 지하수 흐름 등 다양한 기능을 포함하여 강우유출 분석에 활용성 높은 모형으로 평가받고 있으나, 국내 OS환경 및 활용 단위계에서 활용성이 떨어지는 단점이 있다. 본 연구에서는 소스코드 개선 및 GUI구축을 통하여 PC 환경에서 구동이 쉽도록 재구성하였고, 사용자 편의성 확보를 위한 입력자료 전처리기를 개발함으로써 수자원단위지도 3.0, 임상도 재분류 테이블, 토양도 재분류 테이블의 DB화 및 모형의 구동을 위한 HRU분할, 입력자료 생성이 가능하도록 하였다. 매개변수 최적화를 위하여 하천 유량뿐만 아니라 기저유출량을 대상으로 Monte-Carlo 시뮬레이션 기반의 매개변수를 최적화 기능을 탑재하였다. 개발된 모형의 적용성 평가를 위하여 용담댐 시험유역을 대상으로 11년 간(2005-2015)의 강우 및 온도자료를 입력자료로 활용하여 모의한 결과 샘플의 개수에 따라 NSE(Nash-Sutcliffe Efficiency)를 0.9까지 추정이 가능함을 파악하였다. 또한, 유출량과 기저유출에 대하여 동시에 최적화를 수행하는 경우 NSE를 유출량에 대하여 0.8, 기저유출량에 대하여 0.6까지 추정이 가능하였다. 최적화된 모의 결과에 대한 검토를 위하여 계산증발산량을 측정증발산량과 비교한 결과, 유사한 패턴을 나타내는 것을 확인할 수 있었다. 본 연구에서 개발한 $K-BASIN^{RR}$을 활용하는 경우 장기유출해석 업무에 효율성 및 정확도를 향상할 수 있을 것으로 판단된다.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.2
• /
• pp.651-658
• /
• 2021

Because bolts with anti-loosening coatings are used mainly for joining safety-related components in automobiles, accurate automatic screening of these coatings is essential to detect defects efficiently. The performance of the convolutional neural network (CNN) used in a previous study [Identification of bolt coating defects using CNN and Grad-CAM] increased with increasing number of data for the analysis of image patterns and characteristics. On the other hand, obtaining the necessary amount of data for coated bolts is difficult, making training time-consuming. In this paper, resorting to the same VGG16 model as in a previous study, transfer learning was applied to decrease the training time and achieve the same or better accuracy with fewer data. The classifier was trained, considering the number of training data for this study and its similarity with ImageNet data. In conjunction with the fully connected layer, the highest accuracy was achieved (95%). To enhance the performance further, the last convolution layer and the classifier were fine-tuned, which resulted in a 2% increase in accuracy (97%). This shows that the learning time can be reduced by transfer learning and fine-tuning while maintaining a high screening accuracy.

• Journal of Nutrition and Health
• /
• v.48 no.5
• /
• pp.381-389
• /
• 2015

Purpose: The objective of this study was to investigate the effects of a high fructose and fat diet on bone growth and maturation in growing female rats. Methods: Three-week-old female SD rats were randomly assigned to four experimental groups; the control group (CON: fed control diet based on AIN-93G, n = 8); the high-fructose diet group (HFrc: fed control diet with 30% fructose, n = 8); the high-fat diet group (Hfat: fed control diet with 45 kcal% fat, n = 8); and the high-fat diet plus high fructose group (HFrc + HFat: fed diets 45 kcal% fat with 30% fructose, n = 8). Each group was assigned their respective diets for the remaining eight weeks. Bone-related parameters (bone mineral density (BMD) and structural parameters, osteocalcin (OC), deoxypyridinoline (DPD)) and morphologic changes of kidney were analyzed at the end of the experiment. Results: Final body weights and weight gain were higher in the HFat and HFrc + HFat groups and showed higher tendency in the HFrc group compared with those of the CON group (p < 0.05); however, no significant difference in caloric intake was observed among the four experimental groups. The serum OC levels of the HFrc and HFrc + HFat groups were lower than those of the CON and HFat groups (p < 0.05). Urinary levels of DPD did not differ among the experimental groups. BV/TV and Tb.N of trabecular bone were higher in the HFrc + HFat group and showed a higher tendency in the HFrc group than those of the CON and HFat groups (p < 0.05). Tb.Pf of trabecular bone were lower in the HFrc + HFat group than those in the CON and HFat groups (p < 0.05). However, no difference in trabecular BMD was observed among the experimental groups. Cortical bone volume was higher in the HFat and HFrc + HFat groups than in the CON and HFrc groups (p < 0.05). No morphology change in kidney was observed among the experimental groups. Conclusion: Our study suggests that 8 weeks of high-fructose and high fat intake could improve the bone quality (Structural parameters) of trabecular and cortical bone of tibia in growing female rats.

• Journal of Biomedical Engineering Research
• /
• v.19 no.5
• /
• pp.455-468
• /
• 1998

For the objective interpretation of cerebral metabolic patterns in epilepsy patients, we developed computer-aided classifier using artificial neural network. We studied interictal brain FDG PET scans of 257 epilepsy patients who were diagnosed as normal(n=64), L TLE (n=112), or R TLE (n=81) by visual interpretation. Automatically segmented volume of interest (VOI) was used to reliably extract the features representing patterns of cerebral metabolism. All images were spatially normalized to MNI standard PET template and smoothed with 16mm FWHM Gaussian kernel using SPM96. Mean count in cerebral region was normalized. The VOls for 34 cerebral regions were previously defined on the standard template and 17 different counts of mirrored regions to hemispheric midline were extracted from spatially normalized images. A three-layer feed-forward error back-propagation neural network classifier with 7 input nodes and 3 output nodes was used. The network was trained to interpret metabolic patterns and produce identical diagnoses with those of expert viewers. The performance of the neural network was optimized by testing with 5~40 nodes in hidden layer. Randomly selected 40 images from each group were used to train the network and the remainders were used to test the learned network. The optimized neural network gave a maximum agreement rate of 80.3% with expert viewers. It used 20 hidden nodes and was trained for 1508 epochs. Also, neural network gave agreement rates of 75~80% with 10 or 30 nodes in hidden layer. We conclude that artificial neural network performed as well as human experts and could be potentially useful as clinical decision support tool for the localization of epileptogenic zones.