• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.1-17
• /
• 2017

The deep learning framework is software designed to help develop deep learning models. Some of its important functions include "automatic differentiation" and "utilization of GPU". The list of popular deep learning framework includes Caffe (BVLC) and Theano (University of Montreal). And recently, Microsoft's deep learning framework, Microsoft Cognitive Toolkit, was released as open-source license, following Google's Tensorflow a year earlier. The early deep learning frameworks have been developed mainly for research at universities. Beginning with the inception of Tensorflow, however, it seems that companies such as Microsoft and Facebook have started to join the competition of framework development. Given the trend, Google and other companies are expected to continue investing in the deep learning framework to bring forward the initiative in the artificial intelligence business. From this point of view, we think it is a good time to compare some of deep learning frameworks. So we compare three deep learning frameworks which can be used as a Python library. Those are Google's Tensorflow, Microsoft's CNTK, and Theano which is sort of a predecessor of the preceding two. The most common and important function of deep learning frameworks is the ability to perform automatic differentiation. Basically all the mathematical expressions of deep learning models can be represented as computational graphs, which consist of nodes and edges. Partial derivatives on each edge of a computational graph can then be obtained. With the partial derivatives, we can let software compute differentiation of any node with respect to any variable by utilizing chain rule of Calculus. First of all, the convenience of coding is in the order of CNTK, Tensorflow, and Theano. The criterion is simply based on the lengths of the codes and the learning curve and the ease of coding are not the main concern. According to the criteria, Theano was the most difficult to implement with, and CNTK and Tensorflow were somewhat easier. With Tensorflow, we need to define weight variables and biases explicitly. The reason that CNTK and Tensorflow are easier to implement with is that those frameworks provide us with more abstraction than Theano. We, however, need to mention that low-level coding is not always bad. It gives us flexibility of coding. With the low-level coding such as in Theano, we can implement and test any new deep learning models or any new search methods that we can think of. The assessment of the execution speed of each framework is that there is not meaningful difference. According to the experiment, execution speeds of Theano and Tensorflow are very similar, although the experiment was limited to a CNN model. In the case of CNTK, the experimental environment was not maintained as the same. The code written in CNTK has to be run in PC environment without GPU where codes execute as much as 50 times slower than with GPU. But we concluded that the difference of execution speed was within the range of variation caused by the different hardware setup. In this study, we compared three types of deep learning framework: Theano, Tensorflow, and CNTK. According to Wikipedia, there are 12 available deep learning frameworks. And 15 different attributes differentiate each framework. Some of the important attributes would include interface language (Python, C ++, Java, etc.) and the availability of libraries on various deep learning models such as CNN, RNN, DBN, and etc. And if a user implements a large scale deep learning model, it will also be important to support multiple GPU or multiple servers. Also, if you are learning the deep learning model, it would also be important if there are enough examples and references.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.17 no.2
• /
• pp.92-100
• /
• 1984

The seasonal variation of catch and the fishing ground formation of anchovy caught by gill net are studied by using the data for 14 years, 1969 to 1982, published by the Fisheries Research and Development Agency of Korea. The main fishing season of anchovy by gill net can be devised into two seasons: spring and autumn. The former begins early in spring, marks peak in May with the monthly mean catch of 3,000 $\frac{M}{T}$ and ends in summer. The latter begins early in autumn, marks peak in October with the monthly mean catch of 1,500$\frac{M}{T}$ and ends in winter. The fishing ground begins to be formed in the southern waters of Korea with the begining of spring fishing season, and it is extended all over the south-eastern waters from spring to summer and it is converged to the coastal areas from autumn to winter. From the calculation of correlationship between adjacent fishing sections, the fishing ground can be devided into three areas; the northern area of $37^{\circ}N$, the southern area of $35^{\circ}N$ and the area between $35^{\circ}N\;and\;37^{\circ}N$. In the northern area of $37^{\circ}N$, monthly centers of the fishing ground are located in the adjacent aea area of Sockcho-Jumunjin district in the whole year, and its annual mean variance shows about 8 miles in the latitudinal direction and 10 miles in the longitudinal direction. In the area between $35^{\circ}N\;and\;37^{\circ}N$, monthly centers are located in the adjacent sea area of Kijang-Kuryongpo district, and the variance shows about 10 miles in the longitudinal direction and 20 miles in the latitudinal direction. In the southern area of $35^{\circ}N$, monthly centers are located in the open sea in spring and summer, and are conversed to the coastal area in autumn and winter, and the variance shows 8 miles in the latitudinal direction and 35 miles in the longitudinal direction. Water temperature and salinity at the fishing ground where the anchovy gill net was effectively operated are estimated from 14 to $20^{\circ}C$ and from 33.0 to $34.0\%0$ respectively.

• Journal of Preventive Medicine and Public Health
• /
• v.7 no.2
• /
• pp.383-393
• /
• 1974

The importance of the health care for the miners not only concerns the productivity of the enterprises but also as a fundamental human right, it has to be secured and protected by the society and the government. The Korean coal miners began to find their works at overseas since 1963 when the Korean government selected and dispatched its miners to West Germany. As years go by, the demand and the supply of the coal miners, the program of the coal miners health care became the important tasks to be solved and the further studies and research were necessary to support these tasks achievement. The writers made a statistical analysis on health status of the 1,559 Korean applicants applying for the coal miners in West Germany during the year of 1972 and 1973 by the using of the data of physical examination which were directed by the Korea Overseas Development Cooperation. The standard of physical examination which was applied for coal miners applying the jobs in West Germany was authorized by the Government Office of Labor in 1967. The results are as following; 1. The applicants were from the various provinces throughout Korea; Gang Weon 50.2%, Jeon-Nam 16.8%, Chung-Nam 13.7%, Gyeong-Nam 10.5%, City of Seoul 5.4%, and others 3.7%. 2. The ages of the applicants were from 20 to 44. The age group of 25-29(36.3%) and of 30-34(55.2%) together occupied the 91.5 per cent of the total applicants. 3. Among the 1,559 applicants 52.1 per cent passed the physical examinations. The tendency were shown that as the age increased, the passing rate decreased, and the married applicnts had lower passing rates compared to the unmarried applicants. 4. The height of the applicants averaged $168.4{\pm}4.74(cm)$ and the weight averaged $61.3{\pm}5.65(kg)$. 5. The average vision of the left eye of the applicants was $0.92{\pm}0.23$ and the right eye was $0.91{\pm}0.30$. 6. The mean value of the systolic blood pressure of the applicants was $145{\pm}13(mmHg)$ and the diastolic blood pressure was $85{\pm}8(mmHg)$. 7. Fifteen hundred and fifty-nine applicants were taken chest X-ray and 17.2 per cent were found abonrmal. For each 1,000 applicants, 46 Pulmonary calcification and fibrosis, 45 chronic bronchitis, 27 pleural thickening and adhesion, 20 pulmonary tuburculosis, 11 anomalies of the rib, 10 pulmonary infiltration and 21 others diseases were found. 8. Lumber X-ray examination showed that 23.9 per cent (372 applicants) were found abnormal. For each 1,000 applicants, 77 osteoarthritis, 56 lumbarization, 15 fracture of vertebrae, 15 spondyloarthritis, 14 deformity of vertebrae, 13 spina bifida, 12 sacralization, 12 spondylolysis, and 5 others diseases were found. 9. In total, 47.9 per cent of 1,559 applicants failed the physical examinations, and the main causes for the failure were shown to be 7.1% of anomalies of spine, 6.3% of osteoarthritis, 4.7% of pulmonary calcification and fibrosis, 4.6% of chronic brochitis, 3.6% of hypertension (only), 3.4% of fracture of vertebrae. 3.1% of underweight(under 57.0kg), 2.0% of spondyloarthritis, 2.0% oe pulmonary tuberculosis, 1.7% of pleurisy, 1.0% of spina bifida, 1.5% of syphilis, 1.5% of color blindess, and 1.0% of underheight(under 160.0cm).

• Korean Journal of Oriental Medicine
• /
• v.2 no.1
• /
• pp.71-83
• /
• 1996

Hyang-Yack-Ku-Keup-Bang(鄕藥救急方) is our own, medical work written about the middle of the time of Korea Dynasty. I restored and researched this book because it needed to be illuminated about its medico-historic value and then I came to some conclusions as follows. 1. Hyang-Yack-Ku-Keup-Bang was published in Dae-jang-do-kam(大藏都監) of Kanghaw island(江華島) about the middle of Korea Dynasty. Choi Ja-ha(崔自河) republished it on original publication ground in Euiheung(義興) of Kyungsang-Province(慶尙道) in July, Taejong's(太宗) 17th year of Chosen Dynasty (A.D.1417) and this book was published again in Chungcheng Province(忠淸道) in Sejong's(世宗) 9th year(A.D.1427). The book published in Taejong's days was in the possession of books department of Kung-nae-cheng(宮內廳) in Japan and was the oldest medical book of existing ones. 2. Bang-Jung-Hyang-Yack-Mock-Cho-Bu(方中鄕藥目草部) of this book was originally intended to be adjusted in each division with the title of Bang-Jung-Hyang-Yack-Mock(方中鄕藥目). But Herb part(草部) only followed editing progress of Jeung-Lew-Bon-Cho(證類本草), the rest is not divided into each part and is together arranged at the below of Herb part with the title of Bang-Jung-Hyang-Yack-Mock-Cho-Bu. The Korean inscriptions on some drugstuffs in this book are different between Native Name(鄕名) of three volumes of provisions and general-spoken(俗云) of Bang-Jung-Hyang-Yack-Mock-Cho-Bu. In this, it is estimated that the publishing time and editor of tile volume of provisions and Bang-Jung-Hyang-Yack-Mock-Cho-Bu are different. I think Choi Ja-ha compiled this behind three volumes of provisions when he published. 3. This book picked some prescriptions which consisted of obtainable drugs with ease in Korea in the books of Chell-Keum-Yo-Bang(千金要方), Oi-Dae-Bi-Yo(外臺秘要), Tae-Peong-Sung-Hye-Bang(太平聖惠方), Ju-Hu-Bang(?後方), Kyung-Hum-Yang- Bang(經驗良方) Bo-Je-Bon-Sa-Bang(普濟本事方) Bi-Ye-Baik-Yo-Bang(備預百要方) and so on and got together our own prescriptions. On the whole Bi-Ye-Baik-Yo-Bang was a chief referrence book, On this, other books referred to and corrected. 4. In provisions quoted from Hyang-Yack-Jip-Sung-Bang(鄕藥集成方), there are seven provisions; leg-paralysis part, coughing part, headache part, obstetrics part, etc. don't show in this book. This is why Choi Ja-ha published only certain texts on Dae-jang-do-kam edition his own posession. So we can think the existing edition has a little misses compared with original edition. 5. This book recorded only names of drugstuffs in animal drug department like fowls, crab, goldbug, earthworm, etc. and didn't tell us ways of taking those. This is effect of Buddhist culture on medicine. This is efforts to practice 'Don't murder';one of Five Prohibition of Buddhism. 6. Beacause this book was published at the time, when our originative medicine would be set forth. This followed the Chinese ways in Theory, Treatment, Prescription and used 'Hyang Yack' in Medication out of theory of Korean medicine, which was a transitional form. So this is all important material which tell us aspects of development of 'Hyang Yack' the middle of Korea Dynasty.and this is also the beginning of originative, medical works like Dong-Eui-Bo-Kam(東醫寶鑑), Dong-Eui-Su-Bo-Won(東醫壽世保元). 7. There are few contents based on 'Byen-Jeung-Lon-Chi(辨證論治)'in this book. So we can see this book is not for doctors who study medical thoughts but for general public who suffer from diseases resulted from war. Because this book was written for a first-aid treatmeant, this is an index of medical service for the people those days. And this is also an useful datum for first-aid medicine or military medicine in these modern days. 8. Nowadays, parts of learned world of Korean medicine disregard essential theories and want to explain Korean medicine only by the theories or the methods of Western medicine. Moreover they don't adopt Chinese and Japanese theorys & thoughts about Oriental medicine in our own style and just view in there level. What was worse, there is a growing tendency for them to indulge in a trimming policy of scholarship and to take others' ideas. I think these trends to ignore our own medical thoughts involving growth of 'Hyang Yack' in the middle of Korea Dynasty, Dong-Eui-Bo-Kam and Dong-Eui-Su-Se-Bo-Won. So we, as researchers of Korean medicine, must get out of this tendency, and take over brilliant tradition and try to develop originative Korean medicine.

• Journal of Distribution Research
• /
• v.17 no.5
• /
• pp.129-146
• /
• 2012

Recently, Korea became the 7th country in the world which has got into the 20-50 clubs that means the population is 50 million and per capital income is $20,000. From the view point of the retail industry, it suggests that Korea should have its own self-sufficient market by itself. With abundant labor and increased disposable income, it made the consumer's needs change. Responding to the change, retail and F&B oriented retail malls are now starting to add culture, art and entertainment facilities in the configurations. Such complexity and variety of the shopping malls became trendy and many shopping complexes are scheduled to open in the near future across the country. Due to above reasons, it became the common trend to develop shopping complex all over the cities now. However, the history of the shopping mall in Korea is much shorter than developed countries such as America and Japan. Thereby, a lot of problems, trial and error have occurred in the process of developing and operating them. If development of shopping complex failed, it would return lots of damages to the stakeholder. Therefore, the corporations should develop the mall properly and government might support positively. In this study, we would like to propose on how all the mall should be developed and well managed and what are the ways for the vitalizing factors of the shopping complex after benchmarking other shopping mall cases. Through the case study, we realized that the most crucial factors for vitalizing shopping complex were interior design, merchandising and how well they operate the mall. In case of the failed shopping mall, developers sell each store to individual proprietors and never take care of them after they get the profit, which make it hard to have integrated marketing strategies. That causes the overall slump of the mall. Corporations developing the mall should operate it as well, so that it could be possible to make the mall consistently well managed and promoted. There is a certificate for the shopping mall expert in The States and Japan. However, we do not have this kind of certificate. In fact, if we judge the capability of a person who is involved in the shopping mall industry, we usually measure how many years they stick to the same industries and that is not equally the same as their competence. Therefore it is necessary to organize "Shopping Mall Associate" and introduce shopping mall license for the mall expert. Due to retail trends, we can easily see a lot of shopping facilities all over the cities but not every mall is able to be successful. We think it's essential that the government should certify the malls which are qualified for the design, merchandising and proficiency of the operation. For the qualified mall, the government could confer a benefit on the company such as reducing corporation taxes. In multi-complex shopping mall it is possible to make convenient for customers. However, if a mall failed to succeed, it would be disaster. To build a shopping complex, developers should invest huge money even take out loan so that many people would be connected to the project, which may affect their whole financial conditions. In addition, only qualified corporation should develop and operate shopping complex and the government must support and aid the developers in order to make a better shopping environment in which customers might be happy during their shopping experience.

• Journal of Life Science
• /
• v.18 no.8
• /
• pp.1132-1139
• /
• 2008

A CpG island is a short stretch of DNA in which the frequency of the CG dinucleotide is higher than other regions. CpG islands are present in the promoters and exonic regions of approximately $30{\sim}60$% of mammalian genes so they are useful markers for genes in organisms containing 5-methylcytosine in their genomes. Recent evidence supports the notion that the hypermethylation of CpG island, by silencing tumor suppressor genes, plays a major causal role in cancer, which has been described in almost every tumor types. In this respect, CpG island search by computational methods is very helpful for cancer research and computational promoter and gene predictions. I therefore developed a window program (called CpGi) on the basis of CpG island criteria defined by D. Takai and P. A. Jones. The program 'CpGi' was implemented in Visual C++ 6.0 and can determine the locations of CpG islands using diverse parameters (%GC, Obs (CpG)/Exp (CpG), window size, step size, gap value, # of CpG, length) specified by user. The analysis result of CpGi provides a graphical map of CpG islands and G+C% plot, where more detailed information on CpG island can be obtained through pop-up window. Two human contigs, i.e. AP00524 (from chromosome 22) and NT_029490.3 (from chromosome 21), were used to compare the performance of CpGi and two other public programs for the accuracy of search results. The two other programs used in the performance comparison are Emboss-CpGPlot and CpG Island Searcher that are web-based public CpG island search programs. The comparison result showed that CpGi is on a level with or outperforms Emboss-CpGPlot and CpG Island Searcher. Having a simple and easy-to-use user interface, CpGi would be a very useful tool for genome analysis and CpG island research. To obtain a copy of CpGi for academic use only, contact corresponding author.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.16 no.2
• /
• pp.97-105
• /
• 2011

This study investigated the community structure and spatio-temporal variation of macrobenthic assemblages around Gijang Province, East Sea of Korea. Macrobenthos collected seasonally using a modified van Veen grab sampler from March to November 2006. A total of 157 macrobenthic fauna were collected. The overall average macrobenthos density were $552 \;ind/m^2$. The species number of macrobenthos was in the range from 62 in winter and spring to 122 in autumn. On the other hand, abundance fluctuated between 6,540 (in spring) and 17,920 (in autumn) inds./$18m^2$. Cluster analysis and non-metric multi-dimensional scaling (nMDS) were applied to assess the spatio-temporal fluctuation in the macrobenthic assemblages. Cluster analysis and nMDS ordination analysis based on the Bray-Curtis similarity identified 3 station groups. The group 1 (station 8~10, 12, 13, 17 and 18) was characterized by high abundance of the polychaete Lumbrineris longifolia, the bivalve Ennucula tenuis and the Amphipoda spp., with mean phi range from $6.2{\Phi}$ to $7.1{\Phi}$ (above 50m water depth). The group 2 (station 5~7, 11, 14~16) was numerically dominated by the po1ychaete Ampharete arctica and the bivalve Theorafragilis (mean phi: $6.0{\sim}7.0{\Phi}$; within 40 m water depth). Finally group 3 (station 1~4) was characterized by high density of the polychaetes Magelona japonica and Sternaspis scutata, with mean phi range from $3.5{\Phi}$ to $6.9{\Phi}$ (below 30 m water depth). In conclusion, the Macrobenthic community structure showed a distinct spatial and temporal trend, which seemed to be related to the water depth and sediment composition.

• Journal of Aquaculture
• /
• v.18 no.2
• /
• pp.122-129
• /
• 2005

Ammonia is the major limiting factor in intensive aquaculture production systems. Therefore, quantification of ammonia excretion is important for the water quality management in aquaculture systems. Ammonia excretion is known to be affected by many factors such as body weight and dietary protein level (DPL). In this study, experiments were carried out to investigate the effects of body weight and DPLs on the rates of ammonia excretion of Nile tilapia Oreochromis niloticus. Three sizes of fishes (mean initial weight; 4.8 g,42.7 g and 176.8 g) were fed each of two dietary protein levels (30.5% and 35.5%). Daily feeding levels for the three fish sizes of 4.8 g, 42.7 g and 176.8 g were 6%, 3%, and 1.5% body weight per day, respectively. Each group of fish was stocked in a 17.1-L aquarium and all treatments were triplicated. Following feeding, the weight-specific ammonia excretion rate of O. niloticus increased, peaked at 4 to 8 h, and returned to pre-feeding levels within 24 h. Total ammonia nitrogen (TAN) excretion.ate per unit weight decreased with the increase of fish weight for each diet (P<0.05). The TAN excretion rate increased with increasing dietary protein content for each fish size (P<0.05). TAN excretion rates (Y) for each diet with different fish weights were described by the following equations: low DPL diet (30.5%): $Y\;(mg\;kg^{-1}\;d^{-1})=955.69-147.12\;lnX\;(r^2=0.95)$, high DPL diet (35.5%): $Y\;(mg\;kg^{-1}\;d^{-1})=1362.41-209.79\;lnX\;(r^2=0.99)$. Where: X=body weight (g wet wt.). The TAN excretion rates ranged 28.5%-37.1% of the total nitrogen ingested for the low DPL diet (30.5%) and 37.4-38.5% for the high DPL diet (35.5%). Total nitrogen losses of fish fed the high DPL diet $(35.5%;\;0.26\sim0.91g\;kg^{-1}\;d^{-1})$ were higher than those fed the low DPL diet $(30.5%;\;0.22\sim0.68g\;kg^{-1}\;d^{-1})$. The losses decreased per kg of fish as fish size increased. Results will provide valuable information fer water quality management and culture of Nile tilapia in recirculating aquaculture systems.

• Korean Journal of Remote Sensing
• /
• v.39 no.2
• /
• pp.193-205
• /
• 2023

A large amount of floating debris from land-based sources during heavy rainfall has negative social, economic, and environmental impacts, but there is a lack of monitoring systems for floating debris accumulation areas and amounts. With the recent development of artificial intelligence technology, there is a need to quickly and efficiently study large areas of water systems using drone imagery and deep learning-based object detection models. In this study, we acquired various images as well as drone images and trained with You Only Look Once (YOLO)v5s and the recently developed YOLO7 and YOLOv8s to compare the performance of each model to propose an efficient detection technique for land-based floating debris. The qualitative performance evaluation of each model showed that all three models are good at detecting floating debris under normal circumstances, but the YOLOv8s model missed or duplicated objects when the image was overexposed or the water surface was highly reflective of sunlight. The quantitative performance evaluation showed that YOLOv7 had the best performance with a mean Average Precision (intersection over union, IoU 0.5) of 0.940, which was better than YOLOv5s (0.922) and YOLOv8s (0.922). As a result of generating distortion in the color and high-frequency components to compare the performance of models according to data quality, the performance degradation of the YOLOv8s model was the most obvious, and the YOLOv7 model showed the lowest performance degradation. This study confirms that the YOLOv7 model is more robust than the YOLOv5s and YOLOv8s models in detecting land-based floating debris. The deep learning-based floating debris detection technique proposed in this study can identify the spatial distribution of floating debris by category, which can contribute to the planning of future cleanup work.

• Journal of the Korean Society of Radiology
• /
• v.16 no.1
• /
• pp.67-75
• /
• 2022

The aim of this study is to establish a new QC method that can simultaneously evaluate the resolution of the x/y plane and the z-axis by producing a phantom that can reflect exposure and reconstruction parameter of MDCT system. It was used with Aquilion ONE(Cannon Medical System, Otawara, Japan), and the examination was scanned using of 120 kV, 260 mA, and the D-FOV of 300 mm². It produced new SSP phantom modules in which two aluminum plates inclined at 45° to a vertical axis and a transverse axis to evaluate high contrast resolution of x/y plane and z axis. And it changed factors such as the algorithm, distance from gantry iso-center. All images were reconstructed in five steps from 0.6 mm to 10.0 mm slice thickness to measure resolution of x/y plane and z-axis. The image data measured FWHM and FWTM using Profile tool of Aquarius iNtusion Edition ver. 4.4.13 P6 software(Terarecon, California, USA), and analysed SPQI and signal intensity by ImageJ program(v1.53n, National Institutes of Health, USA). It decreased by 4.09~11.99%, 4.12~35.52%, and 4.70~37.64% in slice thickness of 2.5 mm, 5.0 mm, and 10.0 mm for evaluating the high contrast resolution of x/y plane according to distance from gantry iso-center. Therefore, the high contrast resolution of the x/y plane decreased when the distance from the iso-center increased or the slice thickness increased. Additionally, the slice thicknesses of 2.5 mm, 5.0 mm, and 10.0 mm with a high algorithm increased 74.83, 15.18 and 81.25%. The FWHM was almost constant on the measured SSP graph for evaluating the accuracy of slice thickness which represents the resolution of x/y plane and z-axis, but it was measured to be higher than the nominal slice thickness set by user. The FWHM and FWTM of z-axis with axial scan mode tended to increase significantly as the distance increased from gantry iso-center than the helical mode. Particularly, the thinner slice thickness that increased error range compare with the nominal slice thickness. The SPQI increased with thick slice thickness, and that was closer to 90% in the helical scan than the axial scan. In conclusion, by producing a phantom suitable for MDCT detectors and capable of quantitative resolution evaluation, it can be used as a specific method in the management of research quality and management of outdated equipment. Thus, it is expected to contribute greatly to the discrimination of lesions in the field of CT imaging.