• 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.

• Journal of the Mineralogical Society of Korea
• /
• v.22 no.3
• /
• pp.177-189
• /
• 2009

Iron (oxyhydr)oxides commonly form as secondary minerals of high reactivity and large surface area resulting from alteration and weathering of primary minerals, and they are efficient sorbents for inorganic and organic contaminants. Accordingly, they have a great potential in industrial applications and are also of substantial interest in environmental sciences. Goethite (${\alpha}$-FeOOH) is one of the most ubiquitous and stable forms of iron (oxyhydr)oxides in terrestrial soils, sediments, and ore deposits, as well as a common weathering product in rocks of all types. This study focused on adsorption reaction as a main mechanism in scavenging arsenic using goethite. Goethite was synthesized in the laboratory to get high purity, and a variety of mineralogical and physicochemical features of goethite were measured and related to adsorption characteristics of arsenic. To compare differences in adsorption reactions between arsenic species, in addition, a variety of experiments to acquire adsorption isotherm, adsorption edges, and adsorption kinetics were accomplished. The point of zero charge (PZC) of the laboratory-synthesized goethite was measured to be 7.6, which value seems to be relatively higher, compared to those of other iron (oxyhydr)oxides. Its specific surface area appeared to be $29.2\;m^2/g$ and it is relatively smaller than those of other (oxyhydr)oxides. As a result, it was speculated that goethite shows a smaller adsorption capacity. It is likely that the affinity of goethite is much more larger for As(III) (arsenite) than for As(V) (arsenate), because As(III) was observed to be much more adsorbed on goethite than As(V) in equivalent pH conditions. When the adsorption of each arsenic species onto goethite was characterized in various of pH, the adsorption of As(III) was largest in neutral pH range (7.0~9.0) and decreased in both acidic and alkaline pH conditions. In the case of As(V), the adsorption appeared to be highest in the lowest pH condition, and then decreased with an increase of pH. This peculiarity of arsenic adsorption onto goethite might be caused by macroscopic electrostatic interactions due to variation in chemical speciation of arsenic and surface charge of goethite, and also it is significantly affected by change in pH. Parabolic diffusion model was adequate to effectively evaluate arsenic adsorption on goethite, and the regression results show that the kinetic constant of As(V) is larger than that of As(III).

• Cross-Cultural Studies
• /
• v.28
• /
• pp.115-139
• /
• 2012

Interval or duration of segments, syllables, words and phrases is an important acoustic feature which influences the naturalness of speech. A number of cross-sectional studies regarding acoustic characteristics of children's speech development found that intervals of segments, syllables, words and phrases tend to change with the growing age. One hypothesis assumed that decreases in intervals would be greater when children were younger and smaller decreases in intervals when older (Thelen,1991), it has been supported by quite a number of researches on the basis of cross-sectional studies (Tingley & Allen,1975; Kent & Forner,1980; Chermak & Schneiderman, 1986), but the other hypothesis predicted that decreases in intervals would be smaller when children were younger and greater decreases in intervals when older (Smith, Kenney & Hussain, 1996). Researchers seem to come up with conflicting postulations and inconsistent results about the change trends concerning intervals of segments, syllables, words and phrases, leaving it as an issue unresolved. Most acoustic investigations of children's speech production have been conducted via cross-sectional designs, which involves studying several groups of children. So far, there are only a few longitudinal studies. This issue needs more longitudinal investigations; moreover, the acoustic measures of the intervals of child speech are hardly available. All former studies focus on word stages excluding the babbling stages especially the canonical babbling stage, but we need to find out when concrete changes of intervals begin to occur and what causes the changes. Therefore, we conducted an acoustic study of interval characteristics of segments and words concerning Canonical Babble ( CB) and early speech in an infant aged from 0;9 to 2;4 acquiring Mandarin Chinese. The current research addresses the following two questions: 1. Whether decreases in interval would be greater when children were younger and smaller when they were older or vice versa? 2. Whether the child speech concerning the acoustic features of interval drifts in the direction of the language they are exposed to? The female infant whose L1 was Southern Mandarin living in Changsha was audio- and video-taped at her home for about one hour almost on a weekly basis during her age range from 0;9 to 2;4 under natural observation by us investigators. The recordings were digitized. Parts of the digitized material were labeled. All the repetitions were excluded. The utterances were extracted from 44 sessions ranging from 30 minutes to one hour. The utterances were divided into segments as well as syllable-sized units. Age stages are 0;9-1;0,1;1-1;5, 1;6-2;0, 2;1-2;4. The subject was a monolingual normal child from parents with a good education. The infant was audio-and video-taped in her home almost every week. The data were digitized, segments and syllables from 44 sessions spanning the transition from babble to speech were transcribed in narrow IPA and coded for analysis. Babble was coded from age 0;9-1;0, and words were coded from 1;0 to 2;4, the data has been checked by two professionally trained persons who majored in phonetics. The present investigation is a longitudinal analysis of some temporal characteristics of the child speech during the age periods of 0;9-1;0, 1;1-1;5, 1;6-2;0, 2;1-2;4. The answer to Research Question 1 is that our results are in agreement with neither of the hypotheses. One hypothesis assumed that decreases in intervals would be greater when children were younger and smaller decreases in intervals when older (Thelen,1991); but the other hypothesis predicted that decreases in intervals would be smaller when children were younger and greater decreases in intervals when older (Smith, Kenney & Hussain, 1996). On the whole, there is a tendency of decrease in segmental and syllabic duration with the growing age, but the changes are not drastic and abrupt. For example, /a/ after /k/ in Table 1 has greater decrease during 1;1-1;5, while /a/ after /p/, /t/ and /w/ has greater decrease during 2;1-2;4. /ka/ has greater decrease during 1;1-1;5, while /ta/ and /na/ has greater decrease during 2;1-2;4.Across the age periods, interval change experiences lots of fluctuation all the time. The answer to Research Question 2 is yes. Babbling stage is a period in which the children's acoustic features of intervals of segments, syllables, words and phrases is shifted in the direction of the language to be learned, babbling and children's speech emergence is greatly influenced by ambient language. The phonetic changes in terms of duration would go on until as late as 10-12 years of age before reaching adult-like levels. Definitely, with the increase of exposure to ambient language, the variation would be less and less until they attain the adult-like competence. Via the analysis of the SPSS 15.0, the decrease of segmental and syllabic intervals across the four age periods proves to be of no significant difference (p>0.05). It means that the change of segmental and syllabic intervals is continuous. It reveals that the process of child speech development is gradual and cumulative.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.65 no.3
• /
• pp.222-230
• /
• 2020

Maize grain quality can be improved by raising lysine content, which is an essential amino acid present in insufficient quantities in normal maize. Maize varieties with such modifications are known as quality protein maize (QPM). To date, no Korean maize cultivars contain high amounts of lysine. To introduce quality protein maize to Korean cultivars, we crossed QPM CIMMYT maize lines (CML) with KS140, an elite inbred line used as a parent of several cultivars such as 'Gangdaok' and 'Pyeonganok'. We analyzed the phenotypic characteristics of F₁ plants as well as the protein contents, amino acids, and fatty acids profiles of the self-pollinated seeds of the F₁ hybrids, and evaluated the feasibility of CML as a source of QPM. Days to anthesis of the CML ranged from 78 to 90 days after planting (DAP), whereas a range of 81~87 was recorded for F₁ hybrids. The average days to anthesis was 85 for KS140, CML, and the F₁ hybrids. The protein content of the CML was measured to be between 9.1 and 12.1%, with the highest and lowest values being recorded in CML153 and 191, respectively. The F₁ hybrids had protein contents of 9.1~11.1%, and the highest content was observed in KS141/CML188. The fatty acids profiles were very similar across all analyzed maize samples, and linoleic acid (C18:2) composed the greatest proportion. Glutamic acid made up the largest proportion of amino acids in all maize samples. Lysine composition was highest in CML155 (6.92% of all amino acids), with an average composition of 4.83% across the CML. In contrast, KS140 showed a lysine content of 2.51%. In F₁ hybrids, the average lysine composition was 3.46%, and KS140/CML164 (4.18%) and KS140/CML163 (4.99%) contained more lysine than either parent. Taken together, these results indicate that CML could become promising QPM sources to improve grain quality in Korean maize cultivars.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.39 no.1
• /
• pp.33-43
• /
• 2003

The midwater pair trawl which is being used at present in Korea have several problems. Firstly, it is difficult to control the net height on high towing speed. Secondly, net breaking often occurs owing to floats and thirdly, the volume of net on the net drum is so large. This study is aiming for examining the possibility of application for the Korean midwater pair trawl through the model experiment of non-float midwater pair trawl. The model of non-float midwater pair trawl was manufactured as 1/100 of the full scale net which is being used in bottom pair trawl for 850ps class considering the Tauti's Similarity law. The model experiment was carried out to analyze the opening efficiency according to the variation of lower warp length and the opening efficiency was investigated between th proto type and non-float type. The results obtained can be summarized as follows ; 1. The hydrodynamic resistance of non-float type was about 10~20% smaller than that of the proto type and it increased about 1ton according to the increase of dL at the condition of the same flow speed. The resistance acting on the lower warp decreased about 5% but that of the upper warp increased according to the increase of lower warp length (dL) at the condition of the same flow speed. 2. The net height of the non-float type decreased almost linearly according as the increased of flow speed and it increased in a logarithmic functional form with the increase of the lower warp length at the condition of the same flow speed. On the decreasing rate of the net height, the non-float type was lower than the proto type and the difference of the decreasing rate was about 12% at 3.0 knot, 25% at 4.0 knot, 25% at 4.0 knot respectively when dL was 30m. 3. The net width of non-float type was not varied so much as only 2m range and was larger than that of proto type. 4. The mouth area of non-float type decreased in a exponential functional form. On the decreasing rate of the mouth area, the non-float type was lower than the proto type. The filtering volume increased in a logarithmic functional form with increasing flow speed and the filtering volume of proto type decreased steeply over 3.0knot, but that of non-float type increased until 4.0knot. 5. The optimal length of lower warp was when the value of dL was about 30m and the optimal position of front weight was at the connection point of four net pendants.

• Journal of Korean Society of Archives and Records Management
• /
• v.5 no.1
• /
• pp.79-105
• /
• 2005

This study is that the name from Tagjijunjeol is compared with the actual papers used in historical documents. I have selected the families having had influencing power in Gyungi-Do(京畿道), Gyungsang-Do(慶尙道), Jeonra-Do(全羅道), Chungcheung-Do(忠淸道), and analyzed 1504 articles of historical documents owned by those families. I checked up each papers used in historical documents in it's shape of screens, density, raw material. I cannot find the feature of it's region, but prominent feature according to the types and period of historical documents. At first, I have surveyed the impressions of screen marks in papers. The number of strips can be divided into two cases, around 10 and around 17. The case around 17 is similar to the number seen in the book paper of the early Joseon Dynasty, but the case around 10 is rare. Dard Hunter said in his book, published in 1933, that screens were made of Korean grass(Miscanthus sp.) at that time in Korea. I guess the impressions of screen marks of the case around 10 was made of Korean grass, Miscanthus sp.. The case around 10 and the case around 17 were not different in period, but the contemporary appearance. However, if we look into the change of strips' number, we can learn that it becomes more dense as time goes by. The intervals of hair stitch, contrary to book paper, have appeared irregularly and a short hair stitch was frequently found between long hair stitches. This was the result of fastening extra stitch to increase the firmness of strips, as the part of screens which was frequently touched by hands would wear out easily. Then, have look into the density. Each kinds of historical documents had a wide range of density. But the average have had relationship with the kinds of paper written in Tagjijunjeol. The historical documents having the same item of paper have kept similar densities. From the density variation of Gosingyoji(告身敎旨), the letter of appointment the official rank over 5 by king, based on time periods, we can see a great disparity between before and after Imjin-waeran war. Before Imjin-waeran war paper's quality was high, right after Imjin-waeran war went abruptly down, and went gradually up just from 1650's, but since 1700's quality has not risen more, kept the value. On the contrary, the density of Gyocheop(敎牒), the letter of appointment the official rank under 5 by the Ministry of Personnel and the Ministry of Military, have not had a great disparity according to time periods, but had a lower density than Gosingyoji. Joseon Dynasty has strictly divided Gosingyoji appointing the official rank over 5 from Gyocheop appointing the official rank under 5, it seems to distinguish the appointment by king from the appointment by Ijo(吏曹), the Ministry of Personnel and Byeongjo(兵曹), the Ministry of Military.

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.25 no.3
• /
• pp.366-379
• /
• 2015

Objectives: The objective of this study is to assess airborne particulate matter(PM) pollution and its effect on health of residents living near Ansim Briquette Fuel Complex in Daegu metropolitan region. Methods: The California Puff(CALPUFF) dispersion model, version 5.8, which can estimate the dispersion direction and range of airborn $PM_{10}$ was used to determine the possible areas affected by $PM_{10}$ pollutants emitted from Ansim briquette fuel complex. The CALPUFF modeling with 200 m grid-cell resolution was performed based on $PM_{10}$ emissions estimated from the amount of coal consumption in the fuel complex for four months in 2012. The Weather Research and Forecasting(WRF) fields were processed using CALMET to produce CALPUFF-ready meteorological inputs. Also, the distance from Ansim Briquette Fuel Complex to the residence of each environmental pneumoconiosis patient was analyzed. In addition, the affecting region of the pollutants emitted from briquette factories in Ansim Briquette Fuel Complex was determined. Results: CALPUFF modeling results showed that the highest concentrations of $PM_{10}$ were found near around the fuel complex. The modeled $PM_{10}$ distributions were characterized by significant decreases in concentration with distance from the complex. Seasonally, the highest concentration of $45{\mu}g/m^3$ was calculated in October which was mostly due to the distinct variation of amount of emission. Additional modeling with the maximum $PM_{10}$ emission of about 88 tons per year in 1986 showed that the highest concentration in October was nearly increased by 8 times than the concentration modeled with emission of 2010. As a result of medical examination and interviews for the residents in Ansim Briquette Fuel Complex and its surroundings, 8 environmental pneumoconiosis patients were found. These patients do not have occupational exposure and history. These patients have lived 0.3~1.1 km area in Ansim Briquette Fuel Complex and its surroundings. Conclusions: Airborne particles emitted from Ansim Briquette Fuel Complex can contribute to significant increase in $PM_{10}$ concentration in residential areas near around the complex. Especially, the residents near fuel complex may exposed to the pollutants emitted from the factories in Ansim Briquette Fuel Complex.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.13 no.3
• /
• pp.31-42
• /
• 2009

Purpose: As the number of domestic medical institutions installing PET/CT is increasing rapidly, the transfer of PET/CT images among medical institutions is also increasing. Thus, it is necessary to collect the comparative SUV data from several medical institutions' PET/CT systems through a phantom study which semi-quantitatively compares the SUV on one bed, the change scale of the SUV on the slices, and the time of measuring. The phantom study to find differences among the SUVs from various PET/CT offers the opportunity to obtain the reliability of the SUV in PET/CT images. Materials and Methods: Ten PET/CT systems from medical institutions in Korea were used. To obtain the accurate data, the study has been using the radiation detector of Korea Research Institute of Standards and Science to verify. The internal structures of NEMA $phantom^{TM}$ were removed and Six thousand milliliters of distilled water which has 1mCi of $^{18}F$-FDG put into the phantom. The water was properly integrated with $^{18}F$-FDG using magnetic stirrer. The images were acquired at 60, 70, 80, 90, 100, 110 and 120-minutes for 3 minute each. Two hundred square centimeters of region of interests were placed and analyzed. To confirm the usefulness, the correction-table came out from patients' data. Results: The coefficient of variability of the SUV from -11.0 to 9.90 % fell into the range of international standards(${\pm}10%$) along with the SUV on a bed, the change scale of the SUV on the slices, and the time of measuring, except one PET/CT system. Using the data of the differences among the SUVs, we came to withdraw the correction-table ranging from 0.803 to 1.246. The correction-table was confirmed its usefulness through Linear Regression Analysis which was applied to normal cases. Conclusions: Although studies have been made on the variation of the SUV, there is little attention on the standardization of the SUV. Based on this study of the quantitatively comparable data about the SUV accommodating the correction-table, it would help to have more corrective diagnosis.

• Korean Journal of Food Science and Technology
• /
• v.45 no.5
• /
• pp.531-536
• /
• 2013

We developed a simple, sensitive, and specific analytical method for prometryn using gas chromatography-mass spectrometry (GC-MS). Prometryn is a selective herbicide used for the control of annual grasses and broadleaf weeds in cotton and celery crops. On the basis of high specificity, sensitivity, and reproducibility, combined with simple analytical operation, we propose that our newly developed method is suitable for use as a Ministry of Food and Drug Safety (MFDS, Korea) official method in the routine analysis of individual pesticide residues. Further, the method is applicable in clams. The separation condition for GC-MS was optimized by using a DB-5MS capillary column ($30m{\times}0.25mm$, 0.25 ${\mu}m$) with helium as the carrier gas, at a flow rate of 0.9 mL/min. We achieved high linearity over the concentration range 0.02-0.5 mg/L (correlation coefficient, $r^2$ >0.998). Our method is specific and sensitive, and has a quantitation limit of 0.04 mg/kg. The average recovery in clams ranged from 84.0% to 98.0%. The reproducibility of measurements expressed as the coefficient of variation (CV%) ranged from 3.0% to 7.1%. Our analytical procedure showed high accuracy and acceptable sensitivity regarding the analytical requirements for prometryn in fishery products. Finally, we successfully applied our method to the determination of residue levels in fishery products, and showed that none of the analyzed samples contained detectable amounts of residues.

• Korean Journal of Environmental Agriculture
• /
• v.40 no.2
• /
• pp.108-117
• /
• 2021

BACKGROUND: Dichlobentiazox is a newly registered pesticide in Korea as a triazole fungicide and requires establishment of an official analysis method for the safety management. Therefore, the aim of this study was to determine the residual analysis method of dichlobentiazox for the five representative agricultural products. METHODS AND RESULTS: Three QuEChERS methods were applied to establish the extraction method, and the EN method was finally selected through the recovery test. In addition, various adsorbent agents were applied to establish the clean-up method. As a result, it was found that the recovery of the tested pesticide was reduced when using the d-SPE method with PSA and GCB, but C₁₈ showed an excellent recovery. Therefore this method was established as the final analysis method. For the analysis, LC-MS/MS was used with consideration of the selectivity and sensitivity of the target pesticide and was operated in MRM mode. The results of the recovery test using the established analysis method and inter laboratory validation showed a valid range of 70-120%, with standard deviation and coefficient of variation of less than 3.0% and 11.6%, respectively. CONCLUSION: Dichlobentiazox could be analyzed with a modified QuEChERS method, and the method determined would be widely available to ensure the safety of residual pesticides in Korea.