• Korean Journal of Remote Sensing
• /
• v.34 no.5
• /
• pp.811-827
• /
• 2018

The purpose of this study is to compare machine learning algorithm and deep learning algorithm in crop classification using multi-temporal remote sensing data. For this, impacts of machine learning and deep learning algorithms on (a) hyper-parameter and (2) training sample size were compared and analyzed for Haenam-gun, Korea and Illinois State, USA. In the comparison experiment, support vector machine (SVM) was applied as machine learning algorithm and convolutional neural network (CNN) was applied as deep learning algorithm. In particular, 2D-CNN considering 2-dimensional spatial information and 3D-CNN with extended time dimension from 2D-CNN were applied as CNN. As a result of the experiment, it was found that the hyper-parameter values of CNN, considering various hyper-parameter, defined in the two study areas were similar compared with SVM. Based on this result, although it takes much time to optimize the model in CNN, it is considered that it is possible to apply transfer learning that can extend optimized CNN model to other regions. Then, in the experiment results with various training sample size, the impact of that on CNN was larger than SVM. In particular, this impact was exaggerated in Illinois State with heterogeneous spatial patterns. In addition, the lowest classification performance of 3D-CNN was presented in Illinois State, which is considered to be due to over-fitting as complexity of the model. That is, the classification performance was relatively degraded due to heterogeneous patterns and noise effect of input data, although the training accuracy of 3D-CNN model was high. This result simply that a proper classification algorithms should be selected considering spatial characteristics of study areas. Also, a large amount of training samples is necessary to guarantee higher classification performance in CNN, particularly in 3D-CNN.

• Journal of Korean Society of Environmental Engineers
• /
• v.35 no.11
• /
• pp.835-860
• /
• 2013

Recently, the consumption of nanomaterials has been significantly increased in both industrial and commercial sectors, as a result of steady advancement in the nano-technologies. This ubiquitous use of nanomaterials has brought up the concern that their exposure to environments may cause detrimental effects on human health as well as natural ecosystems, and it is required to characterize their behavior in various environmental media and to evaluate their ecotoxicity. For the sake of accomplishing those assessments, the development of methods to effectively separate them from diverse media and to quantify their properties should be requisitely accompanied. Among a number of separation techniques developed so far, this study focuses on Field-Flow Fractionation (FFF) because of its strengths, such as relatively less disturbance of samples and simple pretreatment, and we review overseas and domestic literatures on the separation of nanomaterials using the FFF technique. In particular, researches with Flow Field-Flow Fractionation (FlFFF) are highlighted due to its most frequent application among FFF techniques. The basic principle of the FlFFF is briefly introduced and the studies conducted so far are classified and scrutinized based on the sort of target nanomaterials for the purpose of furnishing practical data and information for the researchers struggling in this field. The literature review suggests that the operational conditions, such as pretreatment, selection of membrane and carrier solution, and rate (velocity) of each flow, should be optimized in order to effectively separate them from various matrices using the FFF technique. Moreover, it seems to be a prerequisite to couple or hyphenate with several detectors and analyzers for quantification of their properties after their separation using the FFF. However, its application has been restricted regarding the types of target nanomaterials and environmental media. Furthermore, domestic literature data on both separation and characterization of nanomaterials are extremely limited. Taking into account the overwhelmingly increasing consumption of nanomaterials, the efforts for the area seem to be greatly urgent.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.8
• /
• pp.187-194
• /
• 2020

The T&E (Test and Evaluation) results were applied for a judgment basis to decide the developmental process of system engineering for efficient weapon system R&D (Research and Development). During the OT&E (Operational Test and Evaluation) and DT&E (Development Test and Evaluation), an environmental test is essential for weapon system development owing to their highly exposed operational conditions. Based on the MIL-STD-810, MIL-HDBK-310, and AECTP 200, the ROK armed forces recommended operating temperatures for the ROK weapon system and applied this to the DT&E and OT&E. This study examined the compatibility of Korean temperature guidelines for stockpile material considering recent climate change. Moreover, this study analyzed the data from hourly measured temperatures on 101 observatories during 60 years, from 1960 to 2020, and percentage (0.5%, 1%, 5%, and 10%) and the 𝜎 (3𝜎, 2𝜎, and 1𝜎) frequency of occurrence on rigorous hot (August) and cold (January) periods, respectively. The results indicate that the highest temperature was 41℃, and the 0.5% frequency of occurrence was 37.0℃. In the case of the cold period, the lowest temperature was -32.6℃ and the 0.5% frequency of occurrence was -21.1℃. By considering the previously recommended operating temperature range for a general ground system, -30 ~ 40℃, regional operation probability is recognized 99.999%. Despite the recent abnormal climate change from global warming, the Korean temperature guidelines are compatible with the stockpile material environmental test.

• Journal of the Mineralogical Society of Korea
• /
• v.30 no.3
• /
• pp.93-102
• /
• 2017

Heavy mineral provide an important information for sediment provenance as well as a potential submarine mineral resources. We compared the heavy mineral characteristics between Southeastern Yellow Sea Mud (SEYSM) and Southwestern Cheju Island Mud (SWCIM) surface sediments. We separated heavy minerals from 28 surface sediments in each mudbelt, and then carried out stereo-microscopic, field-emission scanning electron microscopic, energy dispersive spectroscopic and electron probe microanalysis to characterize the type, abundance, mineralogical properties and distribution pattern of heavy mineral. Amphibole and epidote, which are two major heavy minerals, account for more than 70% of total heavy minerals. Zircon and sphene contents are more abundant in SEYSM, whereas apatite and rutile contents are more abundant in SWCIM. Monazite only occurs in some area of SEYSM. Sphene and monazite content decrease to the south in SEYSM. Both garnet-zircon index (GZi) and rutile-zircon index (RuZi) are low in SEYSM but high in SWCIM. Amphiboles in SEYSM primarily correspond to hornblende, however those in SWCIM represent variable composition from pargasite, tshermakite, hornblende to tremolite. Garnets in SEYSM have high Mg and low Ca, but those in SWCIM have low Mg with variable Ca. Different heavy mineral characteristics between SEYSM and SWCIM suggests that sediments in each mudbelt have different provenances. Although this study implies that SEYSM sediment may mostly come from nearby Korean western rivers such as the Keum and Han rivers, this study does not suggest any idea of the source area of SWCIM sediment. Further study is needed to interpret the provenance and transportation mechanism of mudbelt sediments through the heavy mineral research for the river sediments flowing into the Yellow Sea and much more marine sediments.

• Korean Journal of Agricultural Science
• /
• v.5 no.2
• /
• pp.43-48
• /
• 1978

To information on the inheriance of heading date of rice under the short-day condition, the $F_2$ seeds obtained from all possible combinations of the diallel crosses between five leading rice varieties: Nongbaek, Tongil, Palgueng, Mangyeong and Gimmaze were grown under the short-day condition and natural condition. The days to heading were investigated and analysed for genetic components. The results obtained are summarized as follows. 1. Under the natural condition, the durations to heading of all combinations excluding Nong-baek x Palgueng and Nongbaek x Mangyeong were longer than those of their parents. mean. Whereas, under the short-day condition, the duration to heading of Mangyeong x Gimmaze combination was equal to that of their parents mean, and those of all other combinations were significantly shorter than those of their parents mean. 2. Under the natural condition, the non-allelic gene interactions were significant in days to heading, but under the short-day condition, the non-allelic gene interactions were not significant, and inheritance of days to heading revealed partial dominace in which additive effects were greater than dominant effects. 3. Tongil and Gimmaze were located on dominant zone under the natural day-length, and Nongbaek and Mangyeong were under the short-day condition. 4. Under the natual condition, dominant effects were greater than additive effects, and both heritabilities of the broad and the narrow senses were significantly low. Under the short-day condition, additive effects were significantly higher than dominant effects, and both heritabilities of the broad and the narrow senses were high.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.41 no.1
• /
• pp.21-33
• /
• 2013

The purpose of this study is to find applicabilities of human bioclimatic maps, using human thermal sensation(comfort) in summer, with microclimatic in situ data and computer simulation results at the study site of downtown Daegu. This includes the central business district(CBD) area and two urban parks, the Debt Redemption Movement Memorial Park and the 2.28 Park, for urban and landscape planning and design. Climatic data and urban setting information for the analysis of human thermal sensation were obtained from in situ measurement and the geographic information system data. As a result, the CBD had higher air temperature than the parks when the wind speed was low. Relative humidities were opposite to the air temperature. Especially, same directional streets with local wind direction had lower air temperature than streets perpendicular to the wind direction. The most important climatic variable of human thermal sensation in summer was direct beam solar radiation. Also, creating shadow areas would be the most relevant method for modifying hot thermal environments in urban areas. The most effective method of creating shadow patterns was making a tree shadow over a pergola, and the second best one was making a tree shadow on the front of north directional building walls. Moreover, how to plant trees for creating shadow patterns was important as well as what kind of trees should be planted. The results of human thermal sensation were warm to very hot at sunny areas and neutral to warm at shaded ones. At the sunny areas, wide, squared shape areas had a little bit higher thermal sensation than those of narrow streets. The albedo change of building walls 0.15 and ground surface 0.1 could change 1/6 of a sensation level at the shaded areas and 1/3 at the sunny ones. These microclimatic approaches will be useful to find appropriate methods for modifying thermal environments in urban areas.

• Journal of The Korean Association For Science Education
• /
• v.40 no.4
• /
• pp.385-398
• /
• 2020

This study aims to examine genetics problem-solving processes of high school students with different learning approaches. Two second graders in high school participated in a task that required solving the complicated pedigree problem. The participants had similar academic achievements in life science but one had a deep learning approach while the other had a surface learning approach. In order to analyze in depth the students' problem-solving processes, each student's problem-solving process was video-recorded, and each student conducted a think-aloud interview after solving the problem. Although students showed similar errors at the first trial in solving the problem, they showed different problem-solving process at the last trial. Student A who had a deep learning approach voluntarily solved the problem three times and demonstrated correct conceptual framing to the three constraints using rule-based reasoning in the last trial. Student A monitored the consistency between the data and her own pedigree, and reflected the problem-solving process in the check phase of the last trial in solving the problem. Student A's problem-solving process in the third trial resembled a successful problem-solving algorithm. However, student B who had a surface learning approach, involuntarily repeated solving the problem twice, and focused and used only part of the data due to her goal-oriented attitude to solve the problem in seeking for answers. Student B showed incorrect conceptual framing by memory-bank or arbitrary reasoning, and maintained her incorrect conceptual framing to the constraints in two problem-solving processes. These findings can help in understanding the problem-solving processes of students who have different learning approaches, allowing teachers to better support students with difficulties in accessing genetics problems.

• Progress in Medical Physics
• /
• v.22 no.3
• /
• pp.131-139
• /
• 2011

A method to get a size of the radiation isocenter of linear accelerators using star-shot images was presented and a computer program was developed to automate the method. Accuracy of the method was verified. The developed program was used to measure sizes of the radiation isocenters for a Clinac 21EX (Varian, USA) using data of quality assurance (QA) performed from June 2008 to December 2010. To calculated the size of radiation isocenter, positions of two points on each central ray of the star-shot image were found and the equation of the central ray was determined using the positions of two points. Using the equations of central rays the radius of the minimum circle intersecting all the central rays, which is one half of the size of radiation isocenter, was calculated. The program measured x-intercepts and y-intercepts of the central rays within errors of 0.084 mm and sizes of radiation isocenters within 0.053 mm. All the errors were less than the spatial resolution of star-shot images 0.085 mm. The radiation isocenter sizes of Clinac 21EX were $0.33{\pm}0.27mm$, $0.71{\pm}0.36mm$, $0.50{\pm}0.16mm$ for collimator, gantry and couch respectively. During the measurement period all the measured sizes were less than 2.0 mm and within tolerance. The developed program could calculate the size of radiation isocenters and it would be helpful to routine QA.

• Pediatric Gastroenterology, Hepatology & Nutrition
• /
• v.7 no.1
• /
• pp.8-15
• /
• 2004

Purpose: This study was conducted to identify potential dangers involved in procedure and evaluate complications of percutaneous endoscopic gastrostomy (PEG) comparing to surgical gastrostomy (SG). Methods: A retrospective study of 66 children with feeding gastrostomy between 1994 and 2002 was done. Results: Of 66 children, 23 (mean age 29 months) had PEG and 43 (mean age 49 months) had SG. 31 cases of SG group had fundoplication for gastroesophageal reflux disease. PEG groups were followed up with an average 13 months and SG groups with 21 months. Major complications occurred in 33% of PEG group (8/23) and 55% of SG group (24/43). Major complications were significantly lower in PEG group than SG group and minor complications, too (p<0.05). Of major complications, aspiration pneumonia was the most common but paralytic ileus was significantly higher in SG group than PEG group. 8 patients died of underlying disease but not related to gastrostomy. Removals of stomata were done in 5 of PEG group and 3 of SG group. GER recurred in 25% of SG group with fundoplication and newly developed in 17% of SG group. GER persisted in 17% and newly developed in 5% of PEG group. Conclusion: The gastrostomy was a significant procedure with the potential to produce complications. PEG is recommended as an initial procedure in children requiring a feeding gastrostomy but should be considered a major undertaking.

• Journal of Life Science
• /
• v.28 no.4
• /
• pp.389-397
• /
• 2018

The structure of glycan residues attached to glycoproteins can influence the biological activity, stability, and safety of pharmaceutical proteins delivered from transgenic pig milk. The production of therapeutic glycoprotein in transgenic livestock animals is limited, as the glycosylation of mammary gland cells and the production of glycoproteins with the desired homogeneous glycoform remain a challenge. The ${\beta}$-1,3-N-acetylglucosaminylatransferase1 (B3GNT1) gene is an important enzyme that attaches N-acetylglucosamine (GlcNAc) to galactose (Gal) residues for protein glycosylation; however, there is limited information about pig glycosyltransferases. Therefore, we cloned the pig B3GNT1 (pB3GNT1) and investigated its functional properties that could attach N-acetylglucosamine to galactose residue. Using several different primers, a partial pB3GNT1 mRNA sequence containing the full open reading frame (ORF) was isolated from liver tissue. The ORF of pB3GNT1 contained 1,248 nucleotides and encoded 415 amino acid residues. Organ-dependent expression of the pB3GNT1 gene was confirmed in various organs from adult and juvenile pigs. The pB3GNT1 mRNA expression level was high in the muscles of the heart and small intestine but was lower in the lungs. For functional characterization of pB3GNT1, we established a stable expression of the pB3GNT1 gene in the porcine kidney cell line (PK-15). As a result, it was suggested that the glycosylation pattern of pB3GNT1 expression in PK-15 cells did not affect the total sialic acid level but increased the poly N-acetyllactosamine level. The results of this study can be used to produce glycoproteins with improved properties and therapeutic potential for the generation of desired glycosylation using transgenic pigs as bioreactors.