• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.27-65
• /
• 2020

Many companies on information and communication technology make public their own developed AI technology, for example, Google's TensorFlow, Facebook's PyTorch, Microsoft's CNTK. By releasing deep learning open source software to the public, the relationship with the developer community and the artificial intelligence (AI) ecosystem can be strengthened, and users can perform experiment, implementation and improvement of it. Accordingly, the field of machine learning is growing rapidly, and developers are using and reproducing various learning algorithms in each field. Although various analysis of open source software has been made, there is a lack of studies to help develop or use deep learning open source software in the industry. This study thus attempts to derive a strategy for adopting the framework through case studies of a deep learning open source framework. Based on the technology-organization-environment (TOE) framework and literature review related to the adoption of open source software, we employed the case study framework that includes technological factors as perceived relative advantage, perceived compatibility, perceived complexity, and perceived trialability, organizational factors as management support and knowledge & expertise, and environmental factors as availability of technology skills and services, and platform long term viability. We conducted a case study analysis of three companies' adoption cases (two cases of success and one case of failure) and revealed that seven out of eight TOE factors and several factors regarding company, team and resource are significant for the adoption of deep learning open source framework. By organizing the case study analysis results, we provided five important success factors for adopting deep learning framework: the knowledge and expertise of developers in the team, hardware (GPU) environment, data enterprise cooperation system, deep learning framework platform, deep learning framework work tool service. In order for an organization to successfully adopt a deep learning open source framework, at the stage of using the framework, first, the hardware (GPU) environment for AI R&D group must support the knowledge and expertise of the developers in the team. Second, it is necessary to support the use of deep learning frameworks by research developers through collecting and managing data inside and outside the company with a data enterprise cooperation system. Third, deep learning research expertise must be supplemented through cooperation with researchers from academic institutions such as universities and research institutes. Satisfying three procedures in the stage of using the deep learning framework, companies will increase the number of deep learning research developers, the ability to use the deep learning framework, and the support of GPU resource. In the proliferation stage of the deep learning framework, fourth, a company makes the deep learning framework platform that improves the research efficiency and effectiveness of the developers, for example, the optimization of the hardware (GPU) environment automatically. Fifth, the deep learning framework tool service team complements the developers' expertise through sharing the information of the external deep learning open source framework community to the in-house community and activating developer retraining and seminars. To implement the identified five success factors, a step-by-step enterprise procedure for adoption of the deep learning framework was proposed: defining the project problem, confirming whether the deep learning methodology is the right method, confirming whether the deep learning framework is the right tool, using the deep learning framework by the enterprise, spreading the framework of the enterprise. The first three steps (i.e. defining the project problem, confirming whether the deep learning methodology is the right method, and confirming whether the deep learning framework is the right tool) are pre-considerations to adopt a deep learning open source framework. After the three pre-considerations steps are clear, next two steps (i.e. using the deep learning framework by the enterprise and spreading the framework of the enterprise) can be processed. In the fourth step, the knowledge and expertise of developers in the team are important in addition to hardware (GPU) environment and data enterprise cooperation system. In final step, five important factors are realized for a successful adoption of the deep learning open source framework. This study provides strategic implications for companies adopting or using deep learning framework according to the needs of each industry and business.

• Food Science and Preservation
• /
• v.12 no.6
• /
• pp.617-623
• /
• 2005

Response surface methodology (RSM) was applied to microwave-assisted process (MAP) extraction for effective components from Mosla dianthera M. Microwave power (2,450 MHz, 0-160 W) and extraction time (1-5 min) were used as independent variables ($X_i$) for central composite design to yield 10 different extraction conditions. Optimum conditions were predicted for dependent variables of $75\%$ ethanol extracts, such as total yield($Y_1$), total phenolics($Y_2$), total flavonoids($Y_3$), and electron donation ability($Y_4$, EDA). Determination coefficients ($R^2$) of regression equations for dependent variables ranged from 0.8397 to 0.9801, and microwave power was observed to be more influential than extraction time in MAP. The maximal values of each dependent variable predicted at different extraction conditions of microwave power (W) and extraction time (min) were as follows; $6.76\%$ of total yield at 142.00 W and 4.36 min, 78.68 mg/g of total phenolics at 136.78 W and 4.40 min, 6.75 mg/g of total flavonoids at 159,69 W and 3.17 min, and $49.81\%$ of EDA at 133.87 W and 4.47 min, respectively. The superimposed contour maps for maximizing dependent variables illustrated the MAP conditions of 79 to 113 W in power and of 2.73 to 3.84 min in extraction time.

• Applied Biological Chemistry
• /
• v.48 no.2
• /
• pp.132-139
• /
• 2005

The major obstacle in the popularization of Chungkookjang is the short shelf-life of $2{\sim}3$ months and some problems concerning storage including the growth of molds even in the products even within shelf-life. To solve these problems we conducted a research to improve its storage by using the vacuumed packaging and sanitary method through seed culture, innoculation and sterilization. For the optimization of storage time, temperature and sterilization temperature, we measured viable cell numbers of bacteria and fungi, amount of gas outbreak and contents of amino type nitrogen and monitored these experimental results by response surface methodology of SAS program, so that we could observe the quality changes of Chungkookjang during shelf-life. Especially fungi, which are the biggest troublemaker in Chungkookjang shelf-life, couldn't be detected from the generally and vacuum-packed samples; also, viable cell numbers were highly influenced by sterilization temperature and in vacuum-packed samples. In the case of vacuum-packed samples, amount of gas outbreak was highly influenced by sterilization temperature of its storage conditions and it was higher in generally packed samples as compared to vacuum-packed samples even at any storage conditions. The changes of pH in generally and vacuum-packed samples were highly influenced by the storage temperature. As the temperatures of storage and sterilization were higher and the storage time was longer, so the amount of gas outbreak was accordingly lower. These results showed that amino type nitrogen contents in generally and vacuum-packed samples were systematically influenced by the temperature, storage time and sterilization temperature. Also the result showed that the change of amino type nitrogen contents during storage was less in vacuum-packed samples than in general ones. Based on the above results, we can produce Chungkookjang products with extended shelf-life of as far as 6 months without any quality change using sanitary manufacturing method, vacuumed packaging condition, sterilization in $70^{\circ}C$ for 60 minutes and storage under $10^{\circ}C$ during shelf-life. According to this research, we have the possibility to greatly increase the goods value of Chungkookjang by developing the manufacture processing and packaging.

• Journal of Life Science
• /
• v.18 no.10
• /
• pp.1377-1383
• /
• 2008

A bacterium producing non- or partially digestible dextran was isolated from kimchi broth by enrichment culture technique. The bacterium was identified tentatively as Leuconostoc sp. strain SKY. We established the response surface methodology (Box-Behnken design) to optimize the principle parameters such as culture pH, temperature, and yeast extract concentration for maximizing production of dextran. The ranges of parameters were determined based on prior screening works done at our laboratory and accordingly chosen as 5.5, 6.5, and 7.5 for pH, 25, 30, and $35^{\circ}C$ for temperature, and 1, 5, and 9 g/l yeast extract. Initial concentration of sucrose was 100 g/l. The mineral medium consisted of 3.0 g $KH_2PO_4$, 0.01 g $FeSO_4{\cdot}H_2O$, 0.01 g $MnSO_4{\cdot}4H_2O$, 0.2 g $MgSO_4{\cdot}7H_2O$, 0.01 g NaCl, and 0.05 g $CaCO_3$ per 1 liter deionized water. The optimum values of pH and temperature, and yeast extract concentration were obtained at pH (around 7.0), temperature (27 to $28^{\circ}C$), and yeast extract (6 to 7 g/l). The best dextran yield was 60% (dextran/g sucrose). The best dextran productivity was 0.8 g/h-l.

• Korean Journal of Food Science and Technology
• /
• v.30 no.2
• /
• pp.348-355
• /
• 1998

The developments of various processed foods and the high quality dried fruits, in particular, are urgently needed for the enhancement of fruit consumption and their competitive values. Therefore, in this study, three variables by three level factorial design and response surface methodology were used to determine optimum conditions for osmotic dehydration of kiwifruit. The relationships of moisture losses, solid gains, weight reductions, sugar contents, titratable acidities and vitamin C contents depending on changes with temperature, sugar concentration and immersion time were investigated. The moisture loss, solid gain, weight reduction and reduction of moisture content after osmotic dehydration were increased as temperature, sugar concentration and immersion time increased. The effect of concentration was more significant than those of temperature and time on mass transfer. Sugar content was increased by increasing sugar concentration, temperature, immersion time during osmotic dehydration. Titratable acidity and vitamin C content were increased by decreasing temperature, immersion time and increasing concentration during osmotic dehydration. The regression models showed a significant lack of fit (P>0.05) and were highly significant with satisfying values of $R^2$. At the given conditions such as $66{\sim}69%$ moisture content, above $24^{\circ}Brix$ sugar content and more than 23 mg% vitamin C, the optimum condition for osmotic dehydration was $37^{\circ}C,\;55^{\circ}Brix$ and 1.5 hour.

• Applied Biological Chemistry
• /
• v.48 no.3
• /
• pp.240-245
• /
• 2005

Response surface methodology was used to investigate clarification characteristics (turbidity, brown color, soluble solid, total sugar and reducing sugar) of enzyme in pomegranate extract. Enzyme was treated at 16 conditions including independent variables of temperature ($35{\sim}55^{\circ}C$), time ($30{\sim}70\;min$) and concentration ($0.02{\sim}0.10%$) based on central composition design. Turbidity was decreased with increase of enzyme concentration, and the minimum value of turbidity was 0.04 (OD) when 0.08% enzyme was treated at $37.99^{\circ}C$ for 60.90 min. Total sugar was affected by all treatment conditions and the maximum value was 8.37% when 0.03% enzyme was treated at $39.28^{\circ}C$ for 42.04 min. Reducing sugar and soluble solid were largely affected by enzyme concentration, and the maximum value of reducing sugar was 7.22% when 0.02% enzyme was treated at $42.96^{\circ}C$ for 46.21 min. The maximum value of soluble solid was 8.13% when 0.02% enzyme was treated at $46.91^{\circ}C$ for 42.13 min.

• Journal of Intelligence and Information Systems
• /
• v.21 no.2
• /
• pp.131-150
• /
• 2015

Omission of noun phrases for obligatory cases is a common phenomenon in sentences of Korean and Japanese, which is not observed in English. When an argument of a predicate can be filled with a noun phrase co-referential with the title, the argument is more easily omitted in Encyclopedia texts. The omitted noun phrase is called a zero anaphor or zero pronoun. Encyclopedias like Wikipedia are major source for information extraction by intelligent application systems such as information retrieval and question answering systems. However, omission of noun phrases makes the quality of information extraction poor. This paper deals with the problem of developing a system that can restore omitted noun phrases in encyclopedia documents. The problem that our system deals with is almost similar to zero anaphora resolution which is one of the important problems in natural language processing. A noun phrase existing in the text that can be used for restoration is called an antecedent. An antecedent must be co-referential with the zero anaphor. While the candidates for the antecedent are only noun phrases in the same text in case of zero anaphora resolution, the title is also a candidate in our problem. In our system, the first stage is in charge of detecting the zero anaphor. In the second stage, antecedent search is carried out by considering the candidates. If antecedent search fails, an attempt made, in the third stage, to use the title as the antecedent. The main characteristic of our system is to make use of a structural SVM for finding the antecedent. The noun phrases in the text that appear before the position of zero anaphor comprise the search space. The main technique used in the methods proposed in previous research works is to perform binary classification for all the noun phrases in the search space. The noun phrase classified to be an antecedent with highest confidence is selected as the antecedent. However, we propose in this paper that antecedent search is viewed as the problem of assigning the antecedent indicator labels to a sequence of noun phrases. In other words, sequence labeling is employed in antecedent search in the text. We are the first to suggest this idea. To perform sequence labeling, we suggest to use a structural SVM which receives a sequence of noun phrases as input and returns the sequence of labels as output. An output label takes one of two values: one indicating that the corresponding noun phrase is the antecedent and the other indicating that it is not. The structural SVM we used is based on the modified Pegasos algorithm which exploits a subgradient descent methodology used for optimization problems. To train and test our system we selected a set of Wikipedia texts and constructed the annotated corpus in which gold-standard answers are provided such as zero anaphors and their possible antecedents. Training examples are prepared using the annotated corpus and used to train the SVMs and test the system. For zero anaphor detection, sentences are parsed by a syntactic analyzer and subject or object cases omitted are identified. Thus performance of our system is dependent on that of the syntactic analyzer, which is a limitation of our system. When an antecedent is not found in the text, our system tries to use the title to restore the zero anaphor. This is based on binary classification using the regular SVM. The experiment showed that our system's performance is F1 = 68.58%. This means that state-of-the-art system can be developed with our technique. It is expected that future work that enables the system to utilize semantic information can lead to a significant performance improvement.

• The Korean Journal of Food And Nutrition
• /
• v.25 no.3
• /
• pp.570-578
• /
• 2012

Compound K(ginsenoside M1) is one of saponin metabolites and has many benefits for human health. This study was to investigate Compound K produced from ginseng crude saponin extract with commercial cellulolytic complex enzyme(cellulase, ${\beta}$-glucanase, and hemicellulase) obtained from Trichoderma reesei. The effect factors(temperature, pH, ginseng crude saponin extract and enzyme concentration, and reaction time) on production of Compound K from ginseng crude saponin extract were determined by one factor at a time method. The selected major factor variables were ginseng crude saponin extract of 2%(w/v), enzyme of 7%(v/v), reaction time of 48 hr. Based on the effect factors, response surface method was proceeded to optimize the enzymatic bioconversion conditions for the desirable Compound K production under the fixed condition of pH 5.0 and $50^{\circ}C$. The optimal reaction condition from RSM was ginseng crude saponin extract of 2.38%, enzyme of 6.06%, and reaction time of 64.04 hr. The expected concentration of Compound K produced from that reaction was 840.77 mg/100 g. Production of Compound K was 1,017.93 mg/100 g and 862.31 mg/100 g, by flask and bench-scale bioreactor($2.5{\ell}$) system, respectively.

• Journal of the Korean Society of Food Science and Nutrition
• /
• v.44 no.11
• /
• pp.1708-1714
• /
• 2015

Fermented red ginseng concentrate is known as a healthy food source, whereas it has off-flavor such as bitterness and sour flavor based on fermentation. ${\beta}$- and ${\gamma}$-cyclodextrin (CD) were used to encapsulate the off-flavor of fermented red ginseng concentrate by using response surface methodology design on ${\beta}$- and ${\gamma}-CD$ combination. The reducing effects were analyzed by sensory evaluation for bitter and sour tastes, ginsenoside Rb1, and total acidity. The optimized mixing ratio of ${\beta}$- and ${\gamma}-CD$ for reducing bitterness was the least expected value of 2.07 at ${\beta}-CD$ 3.74% versus the soluble solid content of fermented red ginseng concentrate and the ${\gamma}-CD$ 20.63% mixture. The encapsulation effects of ginsenoside Rb1 were the most expected value of 96.75% at ${\beta}-CD$ 3.47% and ${\gamma}-CD$ 19.89% mixture. The encapsulation effects of sour taste were the least expected value of 5.63 at ${\beta}-CD$ 9.34% and ${\gamma}-CD$ 9.96% mixture. The encapsulation effects of lactic acid were the most expected value of 67.73% at ${\beta}-CD$ 16.0% and ${\gamma}-CD$ 13.18% mixture. Based on encapsulation and each optimized combination, the most effective entrapping ${\beta}$-and ${\gamma}-CD$ combination ratio was ${\beta}-CD$ 10% and ${\gamma}-CD$ 13%.

• Culinary science and hospitality research
• /
• v.21 no.1
• /
• pp.77-89
• /
• 2015

The objective of this study was to optimize processing conditions for the production of an instant puffed rice product using response surface methodology (RSM) and contour analysis. Sensory and texture qualities, and physical properties of the puffed rice were analyzed with various processing conditions related to drying and puffing temperature, and moisture content. Preference, color intensity, cohesiveness, rehydration ratio, density and lightness of the puffed rice product significantly varied depending on the processing conditions. The responses showed high $R^2$ values (0.623, 0.852, 0.735, 0.688, and 0.790) and lack-of-fit. Rehydration ratio was found to have a negative correlation with density in the condition of drying and puffing temperature. Lightness and preference scores of the puffed rice increased as the moisture content increased. According to RSM, the preference scores were very highly related to the moisture content, and the optimum processing conditions of the puffed rice product were at $40^{\circ}C$ of drying temperature, with 11.0% of moisture content, and at $232.7^{\circ}C$ of puffing temperature.