• Journal of KIISE:Software and Applications
• /
• v.30 no.11
• /
• pp.1054-1061
• /
• 2003

Q-Learning, a representative algorithm of reinforcement learning, experiences repeatedly until estimation values about all state-action pairs of state space converge and achieve optimal policies. When the state space is high dimensional or continuous, complex reinforcement learning tasks involve very large state space and suffer from storing all individual state values in a single table. We introduce Q-Map that is new function approximation method to get classified policies. As an agent learns on-line, Q-Map groups states of similar situations and adapts to new experiences repeatedly. State-action pairs necessary for fine control are treated in the form of rule. As a result of experiment in maze environment and mountain car problem, we can achieve classified knowledge and extract easily rules from Q-Map

• Journal of Broadcast Engineering
• /
• v.26 no.5
• /
• pp.621-642
• /
• 2021

In this paper, the subjective video quality assessment of the content quality was to examine by the assessors about the three content quality of HD, Up-scaling UHD, 4K UHD Native that is currently broadcasted. The comparative assessment was conducted using three types of TV sets, 55, 65, 75 inches, at a distance of 2.5 meters which is the distance of watching TV in general households. Among the participants who took part in the video quality evaluating experiment and the questionnaire survey, the answering data of the final 169 persons were adopted after removing the data of 4 persons who answered inadequately in the evaluation of the video quality. The effects of gender, preference of program genre, size of TV sets were analyzed statistically using SPSS 25.0 analysis package. In addition to these, the objective video quality assessment through the measuring instrument was performed, and compared with the results of subjective video quality assessment.

• Journal of Digital Convergence
• /
• v.17 no.2
• /
• pp.31-36
• /
• 2019

New technologies are being created and resulted as new types of fusion complex as the barrier between technology and industries are being broken and convergence is becoming more activated in the global economy of the era of fourth industrial revolution. Korea government is trying to foster innovative technologies for new technologies and new services to prepare for the fourth industrial revolution and gain global competitiveness, but many regulations make it difficult to verify and commercialize them. In response, the Korea government is pushing for the introduction of a regulation-free special zone system in which sandboxes are applied so that new technology and new service-based innovation projects can be freely commercialized through experiment and demonstration. This study aims to examine the limitations of the special zones for regional specialization development applied to the zones that are applied uniformly throughout the country and suggest ways for the deregulation special zone to be fostered as an empirical test bed based on new technologies and as a base for regional innovation.

• Journal of Digital Convergence
• /
• v.17 no.6
• /
• pp.19-25
• /
• 2019

Recently a number of car sharing studies have been conducted that share vehicles from a consumer perspective. Meanwhile, Japan's MLIT had empirically implemented social experiment and policy for the installation of sharing car in order to make better use of the business forms and rapidly increasing car sharing associated with express bus as a new type of transaction from a sharing economy perspective. Therefore, this study examined the case of connecting with the express bus in Japan and analyzed and discussed the contents at the practical level. In addition, data collected for 229 car sharing users were verified empirically. The multiple liner regression analysis showed that three types of perceived values effect on the usage intention in the order of economic value, time value and psychology value. These findings suggest that car sharing users' perceived values are very important for increasing the degree of satisfaction with usage intention.

• KIPS Transactions on Computer and Communication Systems
• /
• v.11 no.7
• /
• pp.205-216
• /
• 2022

One of the many tools used for distributed deep learning training is Kubeflow, which runs on Kubernetes, a container orchestration tool. TensorFlow jobs can be managed using the existing operator provided by Kubeflow. However, when considering the distributed deep learning training jobs based on the parameter server architecture, the scheduling policy used by the existing operator does not consider the task affinity of the distributed training job and does not provide the ability to dynamically allocate or release resources. This can lead to long job completion time and low resource utilization rate. Therefore, in this paper we proposes a new operator that efficiently schedules distributed deep learning training jobs to minimize the job completion time and increase resource utilization rate. We implemented the new operator by modifying the existing operator and conducted experiments to evaluate its performance. The experiment results showed that our scheduling policy improved the average job completion time reduction rate of up to 84% and average CPU utilization increase rate of up to 92%.

• Journal of Digital Policy
• /
• v.2 no.1
• /
• pp.31-35
• /
• 2023

This study was conducted to examine the effect of Broccoli Extract on the proliferation inhibition of human-derived cancer cells and the degree of inhibition. The three cell lines used in the experiment were respiratory system lung cancer cells A549, digestive system liver cancer cells SNU-182 and biliary tract cancer SNU-1196. All cancer cells were derived from the human body, and the CCK-8 method was used to measure the degree of inhibition of cancer cell proliferation. As a result of examining the effect on Broccoli Extract 10ug/mL, 100ug/mL, 1000ug/mL, Broccoli Extract inhibited proliferation in a concentration-dependent manner in most cancer cells, In particular, lung cancer cell A549 and liver cancer cell SNU-182 showed significant proliferation inhibition at 1000ug/mL.As a result, it can be seen that broccoli extract provides potential as a cancer preventive and therapeutic agent for tumor suppression mechanisms proven through cell experiments.

• Journal of Energy Engineering
• /
• v.25 no.4
• /
• pp.93-102
• /
• 2016

Vehicle-to-Grid (V2G) system, bi-direction power trading technology, enables drivers possessing electric vehicle to sell the spare electricity charged in the vehicle to power distribution company. The drivers gain profit by charging electricity in the day time of high electricity rate. In this regard, the government is preparing the policies of building and supporting V2G infrastructure and demanding the potential consumers' preference for the V2G system. This paper attempts to analyze the consumers' preference using the data from obtained a survey of randomly selected 1,000 individuals. To this end, choice experiment, an economic technique, is employed here. The attributes considered in the study are residual amount of electricity, electricity trading hours, required plug-in time, and price measured as an amount additional to current gasoline vehicle price. The multinomial logit model, which requires the assumption of 'independence of irrelevant alternatives', is applied but the assumption could not be satisfied in our data. Thus, we finally utilized nested logit model which does not require the assumption. All the parameter estimates in the utility function are statistically significant at the 10% level. The estimation results show that the marginal willingness to pay (MWTP) for one hour increase in electricity trading hours is estimated to be KRW 1,601,057. On the other hand, a one percent reduction in residual amount of electricity and one hour reduction in required plug-in time in V2G system are computed to be KRW -91,911 and -470,619, respectively. The findings can provide policy makers with useful information for decision-making about introducing and managing V2G system.

• Journal of Digital Convergence
• /
• v.11 no.2
• /
• pp.127-135
• /
• 2013

For the successful implementation of IT projects, individual consultant's competency in the project is very important. Especially, 3 key factors which are 1) Learning-by-Doing, 2) Learning-from-Others, and 3) Learning-by-Investment with individual consultant's competency, are required for solving various critical issues which can be occurred during implementing IT project. The objective of this research is to examine the effects of these learning processes on decision performance of consultants. Prior to setup the research model, we conducted 3 times in-depth interviews with IT consultants who have over 20 years IT project experiences. Through interviews with IT project expert, we tried to validate our research model and develop survey questionnaires. Over 100 consultants, who are working at SI companies those of Samsung SDS, LG CNS, SK C&C and other small SI companies, were participated to survey. In the contrary of our thoughts before conducted experiment, we got the interesting result from pilot experiment. Most influenced learning process was Learning-by-Doing and less influenced learning process was Learning-from-Others.

• Journal of Digital Convergence
• /
• v.18 no.6
• /
• pp.389-399
• /
• 2020

This study was conducted to identify the effects of fundamental nursing practice education using flipped learning on self confidence performance, academic achievement and critical thinking. The subjects of this study were 38 experimental groups and 36 control groups, who were in the second year of the department of nursing at D university, and Flip learning was applied to the experimental group for four weeks, while the existing teaching methods were applied to the control group. As a result, the academic achievement of vital signs in the experimental group after the experiment with flip learning was significantly higher than the control group (t=2.921, p=.005), but other variables were not significant. In addition, the critical thinking of the experimental group was significantly increased after the experiment than before (t=2.277 p=.029). However, it was not significant in the control group. In the future, various studies using flip learning in nursing are needed.

• Journal of Digital Convergence
• /
• v.13 no.6
• /
• pp.29-37
• /
• 2015

Entering the convergence education era, the emergence of smart devices removed the constraint of time and space for study, so if we use smart devices appropriately for education, it will strengthen students' abilities and cultivate creative human resource. Therefore the current study analyzed the general application condition of the smart devices through surveys targeted to students and proposed a measure in applying the smart device as an educational information. In relation to information application, a test was proceeded after carrying out education targeted to experiment group students by naming the group information search, communication, cooperation, sharing, report generation, data storage, online assessment and project management activity. Through the test and survey analysis, it was discovered that the experiment group students displayed higher self-efficacy and ability in applying the smart device as information compared to the control group.