• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.7
• /
• pp.191-198
• /
• 2021

Since the size of training dataset become large and the model is getting deeper to achieve high accuracy in deep learning, the deep neural network training requires a lot of computation and it takes too much time with a single node. Therefore, distributed deep learning is proposed to reduce the training time by distributing computation across multiple nodes. In this study, we propose hybrid allreduce strategy that considers the characteristics of each layer and communication and computational overlapping technique for synchronization of distributed deep learning. Since the convolution layer has fewer parameters than the fully-connected layer as well as it is located at the upper, only short overlapping time is allowed. Thus, butterfly allreduce is used to synchronize the convolution layer. On the other hand, fully-connecter layer is synchronized using ring all-reduce. The empirical experiment results on PyTorch with our proposed scheme shows that the proposed method reduced the training time by up to 33% compared to the baseline PyTorch.

• The Journal of the Korea Contents Association
• /
• v.21 no.7
• /
• pp.363-373
• /
• 2021

The purpose of this research is to analyze the Spill-over economic effect of the cultural and creative industries(CCI) in Henan Province, China. The research object is the CCI of Henan Province, which is mainly based on five sectors out of 42 industries in the industrial association table of the Statistical Bureau of Henan Province, China in 2017 (culture, sports; recreation and research sector; experimental development and integrated technical services sector; information transmission, computer services and software sector; education sector, etc), and is analyzed through secondary integration and redefinition of the CCI of Henan Province. Through the analysis of Henan Province Industry Association Table, this paper provides some enlightenment to the future direction of the cultural and creative industries. The main analysis results are as follows. The total production inducement of the CCI in Henan province is 48,848 billion yuan, and in particular, the production inducement coefficient of the industry in Henan province is 2.72809, 2.23909 (total of columns and rows), Index of the power of dispersion is 0.26325, and the index of the sensitivity of dispersion is 0.87535. Income induction coefficient is 0.55211, production tax induction coefficient is 0.09291. Because CCI of Henan Province has full development potential, the government needs to provide active support and policy support, in addition to the need for legal provisions and supervision of market management. In order to improve the innovative development of the CCI, it is necessary to develop a new model of "CCI+X".

• Journal of the Korea Convergence Society
• /
• v.10 no.3
• /
• pp.49-57
• /
• 2019

The core technology development of the Fourth Industrial Revolution is linked to the development of other core technologies, which will change the industrial structure in the future and create a new smart business model. In this paper, tried to analyze the technology maturity level and analyze the technology maturity. To do this, used technology trend information to investigate and integrate the market, policy, etc. Of core technology of the 4th Industrial Revolution to achieve a comprehensive maturity level. Because technology maturity measures are scored by technology developers, prejudices may be acted upon according to a person's tendency, which may be a subjective evaluation. It is also a measure of the maturity of individual technologies, and thus is not suitable for evaluating the overall system integration perspective. However, it is possible to evaluate the maturity before integrating the core element technologies constituting the whole system and to use it as a means to compare the effect of the whole system and its feasibility and play an important role in the planning of technology development.

• Journal of Energy Engineering
• /
• v.27 no.4
• /
• pp.103-110
• /
• 2018

In this paper, we introduce evaluation method of time series prediction model with new approach of Mean Absolute Percentage Error(hereafter MAPE) and Symmetric Mean Absolute Percentage Error(hereafter sMAPE). There are some problems using MAPE and sMAPE. First MAPE can't evaluate Zero observation of dataset. Moreover, when the observed value is very close to zero it evaluate heavier than other methods. Finally it evaluate different measure even same error between observations and predicted values. And sMAPE does different evaluations are made depending on whether the same error value is over-predicted or under-predicted. And it has different measurement according to the each sign, even if error is the same distance. These problems were solved by Maximum Mean Absolute Percentage Error(hereafter mMAPE). we used the absolute maximum of observed value as denominator instead of the observed value in MAPE, when the value is less than 1, removed denominator then solved the problem that the zero value is not defined. and were able to prevent heavier measurement problem. Also, if the absolute maximum of observed value is greater than 1, the evaluation values of mMAPE were compared with those of the other evaluations. With Beijing PM2.5 temperature data and our simulation data, we compared the evaluation values of mMAPE with other evaluations. And we proved that mMAPE can solve the problems that we mentioned.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.2
• /
• pp.211-219
• /
• 2019

Accomplishing information system, techniques, methodology have been studied continuously and give much help to auditors who are using them. Additionally audit report which is the conclusion of accomplishing ISA(information system audit), has law of a basis and phase with ITA/EA Law(Electronic Government Law). This paper is for better quality of ISA report. But it has more errors about sentence and Grammatical structures. In this paper, to achieve quality improvement objectives, it is necessary to recognize the importance of an audit report by investigating on objectives, functionality, structures and usability of a report firstly, and a legal basis, the presence of report next. Several types of audit reports were chosen and the reports errors were divided into several categories and analyzed. After grasping reasons of those errors, the methods for fixing those errors and check-lists model was provided. And based on that foundation, the effectiveness validation about real audit reports was performed. The necessity for efforts to improve the quality of audit reports was emphasized and further research subject(AI Automatic tool) of this paper conclusion. We also expect this paper to be useful for the organization to improve on ISA in the future.

• The Journal of Society for e-Business Studies
• /
• v.24 no.1
• /
• pp.121-137
• /
• 2019

De-identification is a method by which the remaining information can not be referred to a specific individual by removing the personal information from the data set. As a result, de-identification can lower the exposure risk of personal information that may occur in the process of collecting, processing, storing and distributing information. Although there have been many studies in de-identification algorithms, protection models, and etc., most of them are limited to structured data, and there are relatively few considerations on de-identification of unstructured data. Especially, in the medical field where the unstructured text is frequently used, many people simply remove all personally identifiable information in order to lower the exposure risk of personal information, while admitting the fact that the data utility is lowered accordingly. This study proposes a new method to perform de-identification by applying the k-anonymity protection model targeting unstructured text in the medical field in which de-identification is mandatory because privacy protection issues are more critical in comparison to other fields. Also, the goal of this study is to propose a new utility metric so that people can comprehend de-identified data set utility intuitively. Therefore, if the result of this research is applied to various industrial fields where unstructured text is used, we expect that we can increase the utility of the unstructured text which contains personal information.

• Journal of Digital Convergence
• /
• v.17 no.9
• /
• pp.23-30
• /
• 2019

The development of content for language education using AR/VR technology is a necessary task to be pursued in line with commercialization of 5G. Research on service platform for systematic management and service is currently being carried out by global companies competitively, The unique language education service model for unique areas of culture has the right to pursue R & D jointly with Korea and China. In this study, we applied the developed "Korean language education service platform for Chinese people based on e-learning" to improve the acceptance of AR/VR contents and applied AR/VR technology to video-based language education contents. And to present a new paradigm of language education. Contents development is to develop AR-based vocabulary learning services, develop experiential learning contents for VR-based step-by-step situations, and gradually develop contents to enable beginner / intermediate / advanced language education services. The service platform enables management of learning management and learning contents, and complies with metadata attributes to complete a platform capable of accommodating large capacity AR/VR contents. In the future, systematic research will be carried out in order to develop as a portal for educational services through development of various contents using mixed reality technology.

• KIPS Transactions on Computer and Communication Systems
• /
• v.8 no.2
• /
• pp.41-56
• /
• 2019

Cloud manufacturing is a new concept of manufacturing process that works like a single factory with connected multiple factories. The cloud manufacturing system is a kind of large-scale CPS that produces products through the collaboration of distributed manufacturing facilities based on technologies such as cloud computing, IoT, and virtualization. It utilizes diverse and distributed facilities based on centralized information systems, which allows flexible composition user-centric and service-oriented large-scale systems. However, the cloud manufacturing system is composed of a large number of highly heterogeneous subsystems. It has difficulties in interconnection, data exchange, information processing, and system verification for system construction. In this paper, we derive the user requirements of various aspects of the cloud manufacturing system, such as functional, human, trustworthiness, timing, data and composition, based on the CPS Framework, which is the analysis methodology for CPS. Next, by analyzing the user requirements we define the system requirements including scalability, composability, interactivity, dependability, timing, interoperability and intelligence. We map the defined CPS system requirements to the requirements of oneM2M, which is the platform standard for IoT, so that the support of the system requirements at the level of the IoT platform is verified through Mobius, which is the implementation of oneM2M standard. Analyzing the verification result, finally, we propose a large-scale cloud manufacturing platform based on oneM2M that can meet the cloud manufacturing requirements to support the overall features of the Cloud Manufacturing CPS with dependability.

• Journal of Internet Computing and Services
• /
• v.22 no.1
• /
• pp.23-30
• /
• 2021

This paper proposes a noise-tolerant image classification system using multiple autoencoders. The development of deep learning technology has dramatically improved the performance of image classifiers. However, if the images are contaminated by noise, the performance degrades rapidly. Noise added to the image is inevitably generated in the process of obtaining and transmitting the image. Therefore, in order to use the classifier in a real environment, we have to deal with the noise. On the other hand, the autoencoder is an artificial neural network model that is trained to have similar input and output values. If the input data is similar to the training data, the error between the input data and output data of the autoencoder will be small. However, if the input data is not similar to the training data, the error will be large. The proposed system uses the relationship between the input data and the output data of the autoencoder, and it has two phases to classify the images. In the first phase, the classes with the highest likelihood of classification are selected and subject to the procedure again in the second phase. For the performance analysis of the proposed system, classification accuracy was tested on a Gaussian noise-contaminated MNIST dataset. As a result of the experiment, it was confirmed that the proposed system in the noisy environment has higher accuracy than the CNN-based classification technique.

• Journal of Digital Convergence
• /
• v.19 no.8
• /
• pp.101-111
• /
• 2021

Irregular workers, which have recently caused various socio-economic issues and conflicts, generally have low loyalty to the organization and job satisfaction due to anxiety about employment. As a way to improve this, this study attempted to analyze the effect of organizational communication satisfaction of irregular workers on job satisfaction and organizational commitment. Among the 7th Human Capital Companies panel survey data, irregular workers survey data were collected and analyzed using the structural equation model analysis. The results were as follows: First, it was analyzed that organizational communication recognized by irregular workers had a positive(+) effect on job satisfaction and organizational commitment. Second, it was analyzed that job satisfaction had a positive(+) effect on organizational commitment. Third, it was analyzed that job satisfaction plays a mediating role in the relationship between communication satisfaction and organizational commitment. This study is significant in that it expanded the research subject to irregular workers from the existing service industry-oriented research, and that it included more diverse industries. The results of this study suggest that mission and vision sharing and communication activation system are needed to improve organizational effectiveness of irregular workers.