• Radiation Oncology Journal
• /
• v.19 no.1
• /
• pp.53-65
• /
• 2001

Purpose : To improve the local control of patients with nasopharyngeal cancer, we have implemented 3-D conformal radiotherapy and forward intensity modulated radiation therapy (IMRT) to used of compensating filters. Three dimension conformal radiotherapy with intensity modulation is a new modality for cancer treatments. We designed 3-D treatment planning with 3-D RTP (radiation treatment planning system) and evaluation dose distribution with tumor control probability (TCP) and normal tissue complication probability (NTCP). Material and Methods : We have developed a treatment plan consisting four intensity modulated photon fields that are delivered through the compensating tilters and block transmission for critical organs. We get a full size CT imaging including head and neck as 3 mm slices, and delineating PTV (planning target volume) and surrounding critical organs, and reconstructed 3D imaging on the computer windows. In the planning stage, the planner specifies the number of beams and their directions including non-coplanar, and the prescribed doses for the target volume and the permissible dose of normal organs and the overlap regions. We designed compensating filter according to tissue deficit and PTV volume shape also dose weighting for each field to obtain adequate dose distribution, and shielding blocks weighting for transmission. Therapeutic gains were evaluated by numerical equation of tumor control probability and normal tissue complication probability. The TCP and NTCP by DVH (dose volume histogram) were compared with the 3-D conformal radiotherapy and forward intensity modulated conformal radiotherapy by compensator and blocks weighting. Optimization for the weight distribution was peformed iteration with initial guess weight or the even weight distribution. The TCP and NTCP by DVH were compared with the 3-D conformal radiotherapy and intensitiy modulated conformal radiotherapy by compensator and blocks weighting. Results : Using a four field IMRT plan, we have customized dose distribution to conform and deliver sufficient dose to the PTV. In addition, in the overlap regions between the PTV and the normal organs (spinal cord, salivary grand, pituitary, optic nerves), the dose is kept within the tolerance of the respective organs. We evaluated to obtain sufficient TCP value and acceptable NTCP using compensating filters. Quality assurance checks show acceptable agreement between the planned and the implemented MLC(multi-leaf collimator). Conclusion : IMRT provides a powerful and efficient solution for complex planning problems where the surrounding normal tissues place severe constraints on the prescription dose. The intensity modulated fields can be efficaciously and accurately delivered using compensating filters.

• Radiation Oncology Journal
• /
• v.23 no.4
• /
• pp.223-229
• /
• 2005

Purpose: In this work we designed and made MPBP(Multi Purpose Brachytherapy Phantom). The MPBP enables one to reproduce the same patient set-up in MPBP as the treatment of the patient and we tried to get an exact analysis of rectal doses in the phantom without need of in-vivo dosimetry. Materials and Methods: Dose measurements were tried at a point of rectum 1, the reference point of rectum, with a diode detector for 4 patients treated with tandem and ovoid for a brachytherapy of a cervix cancer. Total 20 times of rectal dose measurements were made with 5 times a patient. The set-up variation of the diode detector was analyzed. The same patient set-ups were reproduced in self-made MPBP and then rectal doses were measured with TLD. Results: The measurement results of the diode detector showed that the set-up variation of the diode detector was the maximum $11.25{\pm}0.95mm$ in the y-direction for Patient 1 and the maximum $9.90{\pm}4.50mm,\;20.85{\pm}4.50mm,\;and\;19.15{\pm}3.33mm$ in the z-direction for Patient 2, 3, and 4, respectively. Un analyzing the degree of variation in 3 directions the more variation was showed in the z-direction than x- and y-direction except Patient 1. The results of TLD measurements in MPBP showed the relative maximum error of 8.6% and 7.7% at a point of rectum 1 for Patient 1 and 4, respectively and 1.7% and 1.2% for Patient 2 and 3, respectively. The doses measured at R1 and R2 were higher than those calculated except R point of Patient 2. this can be thought to related to the algorithm of dose calculation, whcih corrects for air and water but is guessed not to consider the correction for the scattered rays, but by considering the self-error (${\pm}5%$) TLD has the relative error of values measured and calculated was analyzed to be in a good agreement within 15%. Conclusion: The reproducibility of dose measurements under the same condition as the treatment could be achieved owing to the self-made MPMP and the dose at the point of interest could be analyzed accurately. If a treatment is peformed after achieving dose optimization using the data obtained in the phantom, dose will be able to be minimized to important organs.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1993.06a
• /
• pp.1041-1043
• /
• 1993

Fuzzy logic based Control Theory has gained much interest in the industrial world, thanks to its ability to formalize and solve in a very natural way many problems that are very difficult to quantify at an analytical level. This paper shows a solution for treating membership function inside hardware circuits. The proposed hardware structure optimizes the memoried size by using particular form of the vectorial representation. The process of memorizing fuzzy sets, i.e. their membership function, has always been one of the more problematic issues for the hardware implementation, due to the quite large memory space that is needed. To simplify such an implementation, it is commonly [1,2,8,9,10,11] used to limit the membership functions either to those having triangular or trapezoidal shape, or pre-definite shape. These kinds of functions are able to cover a large spectrum of applications with a limited usage of memory, since they can be memorized by specifying very few parameters ( ight, base, critical points, etc.). This however results in a loss of computational power due to computation on the medium points. A solution to this problem is obtained by discretizing the universe of discourse U, i.e. by fixing a finite number of points and memorizing the value of the membership functions on such points [3,10,14,15]. Such a solution provides a satisfying computational speed, a very high precision of definitions and gives the users the opportunity to choose membership functions of any shape. However, a significant memory waste can as well be registered. It is indeed possible that for each of the given fuzzy sets many elements of the universe of discourse have a membership value equal to zero. It has also been noticed that almost in all cases common points among fuzzy sets, i.e. points with non null membership values are very few. More specifically, in many applications, for each element u of U, there exists at most three fuzzy sets for which the membership value is ot null [3,5,6,7,12,13]. Our proposal is based on such hypotheses. Moreover, we use a technique that even though it does not restrict the shapes of membership functions, it reduces strongly the computational time for the membership values and optimizes the function memorization. In figure 1 it is represented a term set whose characteristics are common for fuzzy controllers and to which we will refer in the following. The above term set has a universe of discourse with 128 elements (so to have a good resolution), 8 fuzzy sets that describe the term set, 32 levels of discretization for the membership values. Clearly, the number of bits necessary for the given specifications are 5 for 32 truth levels, 3 for 8 membership functions and 7 for 128 levels of resolution. The memory depth is given by the dimension of the universe of the discourse (128 in our case) and it will be represented by the memory rows. The length of a world of memory is defined by: Length = nem (dm(m)＋dm(fm) Where: fm is the maximum number of non null values in every element of the universe of the discourse, dm(m) is the dimension of the values of the membership function m, dm(fm) is the dimension of the word to represent the index of the highest membership function. In our case then Length=24. The memory dimension is therefore 128*24 bits. If we had chosen to memorize all values of the membership functions we would have needed to memorize on each memory row the membership value of each element. Fuzzy sets word dimension is 8*5 bits. Therefore, the dimension of the memory would have been 128*40 bits. Coherently with our hypothesis, in fig. 1 each element of universe of the discourse has a non null membership value on at most three fuzzy sets. Focusing on the elements 32,64,96 of the universe of discourse, they will be memorized as follows: The computation of the rule weights is done by comparing those bits that represent the index of the membership function, with the word of the program memor . The output bus of the Program Memory (μCOD), is given as input a comparator (Combinatory Net). If the index is equal to the bus value then one of the non null weight derives from the rule and it is produced as output, otherwise the output is zero (fig. 2). It is clear, that the memory dimension of the antecedent is in this way reduced since only non null values are memorized. Moreover, the time performance of the system is equivalent to the performance of a system using vectorial memorization of all weights. The dimensioning of the word is influenced by some parameters of the input variable. The most important parameter is the maximum number membership functions (nfm) having a non null value in each element of the universe of discourse. From our study in the field of fuzzy system, we see that typically nfm 3 and there are at most 16 membership function. At any rate, such a value can be increased up to the physical dimensional limit of the antecedent memory. A less important role n the optimization process of the word dimension is played by the number of membership functions defined for each linguistic term. The table below shows the request word dimension as a function of such parameters and compares our proposed method with the method of vectorial memorization[10]. Summing up, the characteristics of our method are: Users are not restricted to membership functions with specific shapes. The number of the fuzzy sets and the resolution of the vertical axis have a very small influence in increasing memory space. Weight computations are done by combinatorial network and therefore the time performance of the system is equivalent to the one of the vectorial method. The number of non null membership values on any element of the universe of discourse is limited. Such a constraint is usually non very restrictive since many controllers obtain a good precision with only three non null weights. The method here briefly described has been adopted by our group in the design of an optimized version of the coprocessor described in [10].

• Journal of Intelligence and Information Systems
• /
• v.20 no.1
• /
• pp.149-161
• /
• 2014

The export of domestic public services to overseas markets contains many potential obstacles, stemming from different export procedures, the target services, and socio-economic environments. In order to alleviate these problems, the business incubation platform as an open business ecosystem can be a powerful instrument to support the decisions taken by participants and stakeholders. In this paper, we propose an ontology model and its implementation processes for the business incubation platform with an open and pervasive architecture to support public service exports. For the conceptual model of platform ontology, export case studies are used for requirements analysis. The conceptual model shows the basic structure, with vocabulary and its meaning, the relationship between ontologies, and key attributes. For the implementation and test of the ontology model, the logical structure is edited using Prot$\acute{e}$g$\acute{e}$ editor. The core engine of the business incubation platform is the simulator module, where the various contexts of export businesses should be captured, defined, and shared with other modules through ontologies. It is well-known that an ontology, with which concepts and their relationships are represented using a shared vocabulary, is an efficient and effective tool for organizing meta-information to develop structural frameworks in a particular domain. The proposed model consists of five ontologies derived from a requirements survey of major stakeholders and their operational scenarios: service, requirements, environment, enterprise, and county. The service ontology contains several components that can find and categorize public services through a case analysis of the public service export. Key attributes of the service ontology are composed of categories including objective, requirements, activity, and service. The objective category, which has sub-attributes including operational body (organization) and user, acts as a reference to search and classify public services. The requirements category relates to the functional needs at a particular phase of system (service) design or operation. Sub-attributes of requirements are user, application, platform, architecture, and social overhead. The activity category represents business processes during the operation and maintenance phase. The activity category also has sub-attributes including facility, software, and project unit. The service category, with sub-attributes such as target, time, and place, acts as a reference to sort and classify the public services. The requirements ontology is derived from the basic and common components of public services and target countries. The key attributes of the requirements ontology are business, technology, and constraints. Business requirements represent the needs of processes and activities for public service export; technology represents the technological requirements for the operation of public services; and constraints represent the business law, regulations, or cultural characteristics of the target country. The environment ontology is derived from case studies of target countries for public service operation. Key attributes of the environment ontology are user, requirements, and activity. A user includes stakeholders in public services, from citizens to operators and managers; the requirements attribute represents the managerial and physical needs during operation; the activity attribute represents business processes in detail. The enterprise ontology is introduced from a previous study, and its attributes are activity, organization, strategy, marketing, and time. The country ontology is derived from the demographic and geopolitical analysis of the target country, and its key attributes are economy, social infrastructure, law, regulation, customs, population, location, and development strategies. The priority list for target services for a certain country and/or the priority list for target countries for a certain public services are generated by a matching algorithm. These lists are used as input seeds to simulate the consortium partners, and government's policies and programs. In the simulation, the environmental differences between Korea and the target country can be customized through a gap analysis and work-flow optimization process. When the process gap between Korea and the target country is too large for a single corporation to cover, a consortium is considered an alternative choice, and various alternatives are derived from the capability index of enterprises. For financial packages, a mix of various foreign aid funds can be simulated during this stage. It is expected that the proposed ontology model and the business incubation platform can be used by various participants in the public service export market. It could be especially beneficial to small and medium businesses that have relatively fewer resources and experience with public service export. We also expect that the open and pervasive service architecture in a digital business ecosystem will help stakeholders find new opportunities through information sharing and collaboration on business processes.

• Journal of Intelligence and Information Systems
• /
• v.18 no.2
• /
• pp.29-45
• /
• 2012

Bond rating is regarded as an important event for measuring financial risk of companies and for determining the investment returns of investors. As a result, it has been a popular research topic for researchers to predict companies' credit ratings by applying statistical and machine learning techniques. The statistical techniques, including multiple regression, multiple discriminant analysis (MDA), logistic models (LOGIT), and probit analysis, have been traditionally used in bond rating. However, one major drawback is that it should be based on strict assumptions. Such strict assumptions include linearity, normality, independence among predictor variables and pre-existing functional forms relating the criterion variablesand the predictor variables. Those strict assumptions of traditional statistics have limited their application to the real world. Machine learning techniques also used in bond rating prediction models include decision trees (DT), neural networks (NN), and Support Vector Machine (SVM). Especially, SVM is recognized as a new and promising classification and regression analysis method. SVM learns a separating hyperplane that can maximize the margin between two categories. SVM is simple enough to be analyzed mathematical, and leads to high performance in practical applications. SVM implements the structuralrisk minimization principle and searches to minimize an upper bound of the generalization error. In addition, the solution of SVM may be a global optimum and thus, overfitting is unlikely to occur with SVM. In addition, SVM does not require too many data sample for training since it builds prediction models by only using some representative sample near the boundaries called support vectors. A number of experimental researches have indicated that SVM has been successfully applied in a variety of pattern recognition fields. However, there are three major drawbacks that can be potential causes for degrading SVM's performance. First, SVM is originally proposed for solving binary-class classification problems. Methods for combining SVMs for multi-class classification such as One-Against-One, One-Against-All have been proposed, but they do not improve the performance in multi-class classification problem as much as SVM for binary-class classification. Second, approximation algorithms (e.g. decomposition methods, sequential minimal optimization algorithm) could be used for effective multi-class computation to reduce computation time, but it could deteriorate classification performance. Third, the difficulty in multi-class prediction problems is in data imbalance problem that can occur when the number of instances in one class greatly outnumbers the number of instances in the other class. Such data sets often cause a default classifier to be built due to skewed boundary and thus the reduction in the classification accuracy of such a classifier. SVM ensemble learning is one of machine learning methods to cope with the above drawbacks. Ensemble learning is a method for improving the performance of classification and prediction algorithms. AdaBoost is one of the widely used ensemble learning techniques. It constructs a composite classifier by sequentially training classifiers while increasing weight on the misclassified observations through iterations. The observations that are incorrectly predicted by previous classifiers are chosen more often than examples that are correctly predicted. Thus Boosting attempts to produce new classifiers that are better able to predict examples for which the current ensemble's performance is poor. In this way, it can reinforce the training of the misclassified observations of the minority class. This paper proposes a multiclass Geometric Mean-based Boosting (MGM-Boost) to resolve multiclass prediction problem. Since MGM-Boost introduces the notion of geometric mean into AdaBoost, it can perform learning process considering the geometric mean-based accuracy and errors of multiclass. This study applies MGM-Boost to the real-world bond rating case for Korean companies to examine the feasibility of MGM-Boost. 10-fold cross validations for threetimes with different random seeds are performed in order to ensure that the comparison among three different classifiers does not happen by chance. For each of 10-fold cross validation, the entire data set is first partitioned into tenequal-sized sets, and then each set is in turn used as the test set while the classifier trains on the other nine sets. That is, cross-validated folds have been tested independently of each algorithm. Through these steps, we have obtained the results for classifiers on each of the 30 experiments. In the comparison of arithmetic mean-based prediction accuracy between individual classifiers, MGM-Boost (52.95%) shows higher prediction accuracy than both AdaBoost (51.69%) and SVM (49.47%). MGM-Boost (28.12%) also shows the higher prediction accuracy than AdaBoost (24.65%) and SVM (15.42%)in terms of geometric mean-based prediction accuracy. T-test is used to examine whether the performance of each classifiers for 30 folds is significantly different. The results indicate that performance of MGM-Boost is significantly different from AdaBoost and SVM classifiers at 1% level. These results mean that MGM-Boost can provide robust and stable solutions to multi-classproblems such as bond rating.

• Journal of Intelligence and Information Systems
• /
• v.19 no.2
• /
• pp.73-85
• /
• 2013

Nowadays, in today's information society, the importance of the knowledge service using the information to creative value is getting higher day by day. In addition, depending on the development of IT technology, it is ease to collect and use information. Also, many companies actively use customer information to marketing in a variety of industries. Into the 21st century, companies have been actively using the culture arts to manage corporate image and marketing closely linked to their commercial interests. But, it is difficult that companies attract or maintain consumer's interest through their technology. For that reason, it is trend to perform cultural activities for tool of differentiation over many firms. Many firms used the customer's experience to new marketing strategy in order to effectively respond to competitive market. Accordingly, it is emerging rapidly that the necessity of personalized service to provide a new experience for people based on the personal profile information that contains the characteristics of the individual. Like this, personalized service using customer's individual profile information such as language, symbols, behavior, and emotions is very important today. Through this, we will be able to judge interaction between people and content and to maximize customer's experience and satisfaction. There are various relative works provide customer-centered service. Specially, emotion recognition research is emerging recently. Existing researches experienced emotion recognition using mostly bio-signal. Most of researches are voice and face studies that have great emotional changes. However, there are several difficulties to predict people's emotion caused by limitation of equipment and service environments. So, in this paper, we develop emotion prediction model based on vision-based interface to overcome existing limitations. Emotion recognition research based on people's gesture and posture has been processed by several researchers. This paper developed a model that recognizes people's emotional states through body gesture and posture using difference image method. And we found optimization validation model for four kinds of emotions' prediction. A proposed model purposed to automatically determine and predict 4 human emotions (Sadness, Surprise, Joy, and Disgust). To build up the model, event booth was installed in the KOCCA's lobby and we provided some proper stimulative movie to collect their body gesture and posture as the change of emotions. And then, we extracted body movements using difference image method. And we revised people data to build proposed model through neural network. The proposed model for emotion prediction used 3 type time-frame sets (20 frames, 30 frames, and 40 frames). And then, we adopted the model which has best performance compared with other models.' Before build three kinds of models, the entire 97 data set were divided into three data sets of learning, test, and validation set. The proposed model for emotion prediction was constructed using artificial neural network. In this paper, we used the back-propagation algorithm as a learning method, and set learning rate to 10%, momentum rate to 10%. The sigmoid function was used as the transform function. And we designed a three-layer perceptron neural network with one hidden layer and four output nodes. Based on the test data set, the learning for this research model was stopped when it reaches 50000 after reaching the minimum error in order to explore the point of learning. We finally processed each model's accuracy and found best model to predict each emotions. The result showed prediction accuracy 100% from sadness, and 96% from joy prediction in 20 frames set model. And 88% from surprise, and 98% from disgust in 30 frames set model. The findings of our research are expected to be useful to provide effective algorithm for personalized service in various industries such as advertisement, exhibition, performance, etc.

• The Korean Journal of Nuclear Medicine
• /
• v.39 no.3
• /
• pp.182-190
• /
• 2005

Purpose: There are differences between Standard Uptake Value (SUV) of CT attenuation corrected PET and that of $^{137}Cs$. Since various causes lead to difference of SUV, it is important to know what is the cause of these difference. Since only the X-ray CT and $^{137}Cs$ transmission data are used for the attenuation correction, in Philips GEMINI PET/CT scanner, proper transformation of these data into usable attenuation coefficients for 511 keV photon has to be ascertained. The aim of this study was to evaluate the accuracy in the CT measurement and compare the CT and $^{137}Cs$-based attenuation correction in this scanner. Methods: For all the experiments, CT was set to 40 keV (120 kVp) and 50 mAs. To evaluate the accuracy of the CT measurement, CT performance phantom was scanned and Hounsfield units (HU) for those regions were compared to the true values. For the comparison of CT and $^{137}Cs$-based attenuation corrections, transmission scans of the elliptical lung-spine-body phantom and electron density CT phantom composed of various components, such as water, bone, brain and adipose, were performed using CT and $^{137}Cs$. Transformed attenuation coefficients from these data were compared to each other and true 511 keV attenuation coefficient acquired using $^{68}Ge$ and ECAT EXACT 47 scanner. In addition, CT and $^{137}Cs$-derived attenuation coefficients and SUV values for $^{18}F$-FDG measured from the regions with normal and pathological uptake in patients' data were also compared. Results: HU of all the regions in CT performance phantom measured using GEMINI PET/CT were equivalent to the known true values. CT based attenuation coefficients were lower than those of $^{68}Ge$ about 10% in bony region of NEMA ECT phantom. Attenuation coefficients derived from $^{137}Cs$ data was slightly higher than those from CT data also in the images of electron density CT phantom and patients' body with electron density. However, the SUV values in attenuation corrected images using $^{137}Cs$ were lower than images corrected using CT. Percent difference between SUV values was about 15%. Conclusion: Although the HU measured using this scanner was accurate, accuracy in the conversion from CT data into the 511 keV attenuation coefficients was limited in the bony region. Discrepancy in the transformed attenuation coefficients and SUV values between CT and $^{137}Cs$-based data shown in this study suggests that further optimization of various parameters in data acquisition and processing would be necessary for this scanner.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.

• 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.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.34 no.1
• /
• pp.60-73
• /
• 1989

Two field experiments were carried out to assess the applicability of a central composite design (CCD) in determining optimum culture condition of an early rice cultivar, Unbongbyeo in southern Korea. A central composite design with two replicates was applied to five levels of five factors such as the number of hills per 3.3m2, the number of seedlings per hill, the levels of nitrogen, the transplanting date and the seedling age (Experiment 1). The levels of planting density were ranged from 30 hills to 150 hills per 3.3m2 ; the number of seedlings per hill from 1 seedling to 9 seedlings per hill; the levels of nitrogen application from 1 kg/l0a to 21 kg/l0a; the transplanting date from June 15 to July 5; the seedling age from 25 days to 45 days. A fractional factorial design was applied to three levels of five factors tested in CCD (Experiment 2). Yield per hill and per unit area were examined and the results obtained from both experiments were compared. The benefits from the central composite design were discussed. Maximum yield of brown rice per unit area was obtained at the combination of the central levels of one of five factors when the other four factors were fixed at central point. Furthermore, brown rice yield per unit area affected by interaction of two factors was maximized at the central point when the remain three factors being fixed at the central level. The responses of five factors to brown rice yield per hill and unit area were found to be a saddle point in both designs. Actual values of the stationary points were 107 hills per 3.3 m2, 4 seedlings per hill, 10 kg nitrogen per l0a, transplanting date of rice on June 26 and 33 days of seedling age in the central composite design. Brown rice yield per unit area at the stationary points were estimated 439 kg/l0a in the central composite design and 442 kg/l0a in the fractional factorial design. Considering the number of experimental treatment combinations, the central composite design was rather convenient in reducing the number of treatment combinations for similar information. It was more convenient for an experimenter to present the results from the central composite design than those from the fractional factorial design. Considering the optimum yields of brown rice per unit area at the stationary points being verified as saddle points in both designs. inter-heterogeneity of each of the factors should be avoided in setting up factors in pursuit of inducing unidirectional response of the factors to yield. Even though both the lower and higher levels in the central composite design being beyond the region of an experimenter's interest. they were considered highly valued in interpretation of the results. Conclusively. the central composite design was found to be more beneficial to optimize culture condition of paddy rice even with several levels of various factors were involved.