• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2008.10a
• /
• pp.195-199
• /
• 2008

본 논문은 택시들의 승차율을 높이고 승객들의 대기시간을 최소화하기 위하여 제주 택시들의 이동이력 데이터를 기반으로 하여 공차들을 승객을 만날 가능성이 많은 지역으로 배치하는 기법을 제시한다. 이동이력 데이터에 포함된 공차 보고와 승차 보고 수 사이의 스케일 차이를 극복하기 위하여 전체수와 영역내 합에 대한 비율로 정규화하는 방법을 설명하고 장단점을 분석한다. 또 시간대별, 요일별, 주간별 택시들의 승객 대기시간에 대한 통계 데이터에 기반하여 가장 수요와 공급이 적정하게 유지되는 시간구간을 발견하고 이 구간에 대한 택시 분포와 현재의 택시 분포의 차이에 의해 수요보다 공급이 많은 곳의 택시를 재배치한다.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.19 no.6
• /
• pp.606-611
• /
• 2013

In this study, the regional states of 7 items are analyzed, the regional risks are calculated by using normalized data & analysis hierarchy process to set regional oil recovery capacity. Area on-water oil recovery capacity, $7,500k{\ell}$, is separated and regional on-water oil recovery capacity is determined, based on calculated regional degree of risk. Excessive current oil recovery capacities, setting in Incheon, Gunsan, Mokpo, Busan region, are as a result distributed to the other regions in each area. In case of central region, Daesan is increased as much as $1,475k{\ell}$, Yeosu is increased as much as $375k{\ell}$, Ulsan is increased as much as $475k{\ell}$. The regional on-water oil recovery capacity, considering both the cause of accident aspects and marine environmental & economic aspects, is estimated as more balanced distribution model, compared to current standard of on-water oil recovery capacity.

• Journal of KIISE
• /
• v.44 no.7
• /
• pp.668-673
• /
• 2017

Deep neural networks have shown remarkable performance in various fields of pattern recognition such as voice recognition, image recognition and object detection. However, underlying mechanisms of the network have not been fully revealed. In this paper, we focused on empirical analysis of the network parameters. The Faster R-CNN(region-based convolutional neural network) was used as a baseline network of our work and three important parameters were analyzed: the dropout ratio which prevents the overfitting of the neural network, the size of the anchor boxes and the activation function. We also compared the performance of dropout and batch normalization. The network performed favorably when the dropout ratio was 0.3 and the size of the anchor box had not shown notable relation to the performance of the network. The result showed that batch normalization can't entirely substitute the dropout method. The used leaky ReLU(rectified linear unit) with a negative domain slope of 0.02 showed comparably good performance.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.1
• /
• pp.81-87
• /
• 2012

The current defacing method for keeping an anonymity of brain images damages the integrity of a precise brain analysis due to over removal, although it maintains the patients' privacy. A novel method has been developed to create an anonymous face model while keeping the voxel values of an image exactly the same as that of the original one. The method contains two steps: construction of a mockup brain template from ten normalized brain images and a substitution of the mockup brain to the brain image. A level set segmentation algorithm is applied to segment a scalp-skull apart from the whole brain volume. The segmented mockup brain is coregistered and normalized to the subject brain image to create an anonymous face model. The validity of this modification is tested through comparing the intensity of voxels inside a brain area from the mockup brain with the original brain image. The result shows that the intensity of voxels inside from the mockup brain is same as ones from an original brain image, while its anonymity is guaranteed.

• Knowledge Management Research
• /
• v.22 no.4
• /
• pp.27-40
• /
• 2021

Click-Through Rate (CTR) prediction is a key function that determines the ranking of candidate items in the recommendation system and recommends high-ranking items to reduce customer information overload and achieve profit maximization through sales promotion. The fields of natural language processing and image classification are achieving remarkable growth through the use of deep neural networks. Recently, a transformer model based on an attention mechanism, differentiated from the mainstream models in the fields of natural language processing and image classification, has been proposed to achieve state-of-the-art in this field. In this study, we present a method for improving the performance of a transformer model for CTR prediction. In order to analyze the effect of discrete and categorical CTR data characteristics different from natural language and image data on performance, experiments on embedding regularization and transformer normalization are performed. According to the experimental results, it was confirmed that the prediction performance of the transformer was significantly improved when the L2 generalization was applied in the embedding process for CTR data input processing and when batch normalization was applied instead of layer normalization, which is the default regularization method, to the transformer model.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.01a
• /
• pp.321-322
• /
• 2023

본 논문에서는 Scikit-learn 패키지의 LinearRegression 모델과 Keras 딥러닝 모델을 활용하여 수도권 도시 환경 요인에 따른 범죄 발생 건수를 예측 모델을 제안한다. 연구 방법으로 범죄 발생과 유의미한 관계가 있다고 파악되는 수도권의 각 자치구 별 데이터셋을 분석하여, CCTV, 파출소, 가로등의 수가 범죄 발생에 유의미한 영향을 끼치는 것을 확인하였다. 독립 변수들 간에 Scale을 줄이고자 정규화를 진행했고, 종속변수의 정규성 확보를 위해 로그변환을 취했다. 손실 함수는 회귀문제에서 사용되는 'relu'함수를 사용했고 모델의 성능을 확인할 수 있는 지표로 MSE(Mean Squared Error)를 사용해 모델을 구성하였다. 본 논문에서 설계한 이 프로그램은 범죄 발생율이 높은 지역구에 경찰 인력의 추가적 배치, 안전 시설 확충 등 실무적 조치를 취함에 있어 근거를 제공할 수 있을 것으로 사료된다.

• Korean Journal of Social Welfare
• /
• v.57 no.3
• /
• pp.73-103
• /
• 2005

This study takes of youth panel data(2002-2003), there is the purpose to know to youth employed's labor mobility conditions of employment status conversion and examine on factors affecting conversion of employment status. Main analysis result and policy imply, is as following. First, although youth employed's non-standard employment shows some decrease, employment youth hierarchy was construed that youth of our country is consisting very changefully because appear great turnover that is converted by unemployment or Not economically activity population same period. Specially, non-standard employment phenomenon of woman or low in scholarship person appeared notedly, and because phenomenon that is converted by unemployment or Not economically activity population is expose that is deepened, discriminating policy of government dimension is required for employment stabilization of these class. Second, show result that danger to escape to non-standard job risk trap which seeking employment activity of youth class is arranged case or company which is formed by official path is suitable becomes low, must formulate path of employment about youth class and improve qualitative level of employment through suitable job placement education of youth class or function (technology) level. Third, when was construed, but take into account that the although large enterprise have low risk in non-standard job, recently employment of youth class consists very limited, rather small scale business or smaller enterprise's competitive power preferably need to be plan. Finally, danger to non-standard job youth employed's company form is government connection wonder was expose that high, Such result can do that it is difficult by limited research period, but reflect actuality that youth unemployment policy of our country is enforced laying stress on public labor or unregular job employment such as internship system. Therefore, current youth unemployment policy may have to change by employment policy that can secure stable work record by youth class or act as bridge-building that promote conversion by full-time job.

• Journal of Ocean Engineering and Technology
• /
• v.11 no.4
• /
• pp.130-140
• /
• 1997

원형주상체 주위의 유동을 규명하기 위해 회류수조에서 원주방향으로 24개의 위치에 대하여 압력을 계측하였으며, laser sheet을 이용하여 유동을 가시화 하였다. Reynolds수가 4800에서 40000인 범위에 대하여 실험을 수행하였다. 또한, 원형단면체 주위의 비정상 층류유동에 대한 Navier-Stokes방정식의 해를 구하는 수치해석기법을 개발하였다. 효과적인 격자배치를 위하여 H와 O-type의 중첩격자를 사용하였고, 이산화 방법으로는 정규격자시스템에서 유한차분법을 적용하였다. 실험과 수치해석결과에서 뚜렷한 와류박리현상을 볼 수 있었으며, 압력계수 (C$_{p}$ ), 항력계수(C$_{D}$), 스트로얼수(St)를 정량적으로 비교하였을 때, 비교적 잘 일치하는 것을 확인하였다.다.

• Cartoon and Animation Studies
• /
• s.36
• /
• pp.393-416
• /
• 2014

Comics has three components: cuts, gap between cuts, speech balloons. Still, it is true that speech balloons are not commonly subject to the study for comics. A few preceding researches pinpoint exactly the morphological features and functions of speech balloons. In today when webtoon becomes generalized, these features and functions are continued as they are and are used in webtoon. We can catch that speech balloons are also affected since the environmental elements of web induce the change in the overall comics directing. There are two perspectives to sort out the features of speech balloons: first, the placement issue of speech balloons. The unlimited expansion of web space gives the environment for comicss to use the gap between cuts as wide as they can. It leads to turn out some of the ways to place the balloons, so we can sort them out general placement, exterior placement, the upper and lower placement, scroll-use type. Second, as the directing techniques for webtoon become digitalized by the morphological issue, speech balloon itself has been expanded its ways to express by various expression methods. Analyzing and classifying, recording the newly emerged conditions on the preceding study are worthy of trying and will become the cornerstone for the follow research.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.05a
• /
• pp.165-170
• /
• 2003

This paper presents an integrated approach to unconstrained face recognition in arbitrary scenes. The front end of the system comprises of a scale and pose tolerant face detector. Scale normalization is achieved through novel combination of a skin color segmentation and log-polar mapping procedure. Principal component analysis is used with the multi-view approach proposed in[10] to handle the pose variations. For a given color input image, the detector encloses a face in a complex scene within a circular boundary and indicates the position of the nose. Next, for recognition, a radial grid mapping centered on the nose yields a feature vector within the circular boundary. As the width of the color segmented region provides an estimated size for the face, the extracted feature vector is scale normalized by the estimated size. The feature vector is input to a trained neural network classifier for face identification. The system was evaluated using a database of 20 person's faces with varying scale and pose obtained on different complex backgrounds. The performance of the face recognizer was also quite good except for sensitivity to small scale face images. The integrated system achieved average recognition rates of 87% to 92%.