• ETRI Journal
• /
• v.46 no.2
• /
• pp.205-217
• /
• 2024

We propose a bandwidth prediction approach based on deep learning. The approach is intended to accurately predict the bandwidth of various types of mobile networks. We first use a machine learning technique, namely, the gradient boosting algorithm, to recognize the connected mobile network. Second, we apply a handover detection algorithm based on network recognition to account for vertical handover that causes the bandwidth variance. Third, as the communication performance offered by 3G, 4G, and 5G networks varies, we suggest a bidirectional long short-term memory model with time distribution for bandwidth prediction per network. To increase the prediction accuracy, pretraining and fine-tuning are applied for each type of network. We use a dataset collected at University College Cork for network recognition, handover detection, and bandwidth prediction. The performance evaluation indicates that the handover detection algorithm achieves 88.5% accuracy, and the bandwidth prediction model achieves a high accuracy, with a root-mean-square error of only 2.12%.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.941-946
• /
• 2005

In this paper an active random noise control using adaptive learning rate neural networks is presented. The adaptive learning rate strategy increases the learning rate by a small constant if the current partial derivative of the objective function with respect to the weight and the exponential average of the previous derivatives have the same sign, otherwise the learning rate is decreased by a proportion of its value. The use of an adaptive learning rate attempts to keep the learning step size as large as possible without leading to oscillation. It is expected that a cost function minimize rapidly and training time is decreased. Numerical simulations and experiments of active random noise control with the transfer function of the error path will be performed, to validate the convergence properties of the adaptive learning rate Neural Networks. Control results show that adaptive learning rate Neural Networks control structure can outperform linear controllers and conventional neural network controller for the active random noise control.

• Educational Technology International
• /
• v.21 no.1
• /
• pp.97-123
• /
• 2020

The purpose of this study is to investigate the emotions of students toward using digital textbooks, and to examine the factors affecting the emotions. This study examined the relationship between individual characteristics and computer usage, students' emotions, and the perceived learning effects. For this study, 2,950 1st grade middle school students participated in a survey which measured individual characteristics, computer usage behavior, emotions toward using digital textbooks, and perceived learning effects of digital textbooks. The results showed that positive emotions toward using digital textbooks were higher than negative emotions. The students' positive emotions were most affected by intrinsic motivation, self-regulated learning, and student's use of computers for learning and entertainment. Similarly, perceived learning effects were positively correlated to intrinsic motivation and self-regulated learning, but the students' positive emotions towards using digital textbooks was the strongest predictor. Digital textbook efficacy was the most influential factor that affected the students' negative emotions, while computer addiction was associated with negative emotions.

• Journal of Korea Multimedia Society
• /
• v.25 no.1
• /
• pp.52-60
• /
• 2022

Few-shot learning is a sub-area of machine learning problems, which aims to classify target images that only contain a few labeled samples for training. As a representative few-shot learning method, the Prototypical network has been received much attention due to its simplicity and promising results. However, the Prototypical network uses the sample mean of samples from the same class as the prototypes of that class, which easily results in learning uncharacteristic features in the low-data scenery. In this study, we propose to use local descriptors (i.e., patches along the channel within feature maps) from the same class to explicitly obtain more representative prototypes for Prototypical Network so that significant intra-class feature information can be maintained and thus improving the classification performance on few-shot learning tasks. Experimental results on various benchmark datasets including mini-ImageNet, CUB-200-2011, and tiered-ImageNet show that the proposed method can learn more discriminative intra-class features by the local descriptors and obtain more generic prototype representations under the few-shot setting.

• Journal of The Korean Association of Information Education
• /
• v.15 no.3
• /
• pp.355-363
• /
• 2011

Level-Based Education is necessary for the multi-cultural learners because they tend to have the academic underachievement and learning deficiency that cause the huge educational gap. However, it is very hard to make the best of competence for the multi-cultural learners in the classroom. So, it is needed to suggest how we can use the educational contents that are appropriate for the Level-Based Learning and Individual Learning to make good use of teaching the learners from multi-cultural families. However, developing the new educational contents takes much time and cost, we have to improve existing contents for the student from multi-cultural families to use it. Hence, the purpose of this thesis is to develop the educational appropriateness evaluation scale to verify the educational contents that are for the multi-cultural students based on the educational content's evaluation tool, so by developing the scale, I intend to evaluate the 4~6 grades' Korean contents of the E-learning service and provide the ways of improvement.

• The Journal of the Korea Contents Association
• /
• v.16 no.12
• /
• pp.695-706
• /
• 2016

The present study sought to shed light on specific designs and strategies by investigating the impact that team project-based learning approach using Naver Band had on learning outcomes specifically, on satisfaction and achievement. Variables of perceived usefulness and perceived ease of use were chosen as critical predictors of the learning outcomes. 70 undergraduate students were divided into 18 groups and instructed to work on a team project using Naver Band for 6 weeks. Data analysis was completed using descriptive statistic analysis, correlation analysis, hierarchical regression analysis. Also, descriptions of Band from team reflection journal were analyzed. The major results of the study are as follows: first, perceived usefulness was the significant predictor of satisfaction and achievement. Second, 5 functions were revealed as strengths about Band.

• Journal of Digital Convergence
• /
• v.17 no.5
• /
• pp.99-108
• /
• 2019

This study aims to explore key factors which influence user's decision-making on the adoption of e-learning. We integrated UTAUT and Information Success Models to test that four independent factors affect student satisfaction to use e-learning in Rwanda's higher education. Data was collected by surveying students of University of Rwanda and Protestant Institute of Social Sciences (n=206). The analysis results showed that performance expectancy, facilitating conditions and effort expectancy except for social influence have a significant effect on students' satisfaction. This can help university administrators understand the factors that influence students' adoption of e-learning and incorporate these results into Rwanda's e-learning design and implementation. In final, Rwanda's government can contribute to establishing the e-learning policy and allocating its relevant resources centered on student needs.

• The KIPS Transactions:PartA
• /
• v.18A no.1
• /
• pp.33-38
• /
• 2011

The purpose of this study is to develop and use the e-learning contents accessibility guidelines to improve contents accessibility for the non-technical developers and designers. The accessibility guidelines used for web or digital contents are usually technical, field dependent, or specific that are not friendly for many developers or designers. In this study, the e-Learning Contents Accessibility Guidelines was developed based on the principles of Universal Design for Learning. The guidelines could be used to map the necessary skills for the developers and the instructional designers. In this study, 5 users with different disabilities tested 6 e-learning contents, and surveyed e-learning experts to identify core elements for accessibility guidelines. Due to the limited accessibility of the contents, we need to offer manuals and training for developers and designers, need collaborative efforts between different stake holders, include accessibility in quality assurance guidelines, and further research to improve accessibility for many existing Flash contents.

• Journal of Science Education
• /
• v.39 no.1
• /
• pp.28-43
• /
• 2015

The purpose of this study was to measure middle and high school students' stage of concern(SoC) and their level of use(LoU) on the self-directed learning in science learning based on the CBAM(Concern-Based Adoption Model). Additionally, this research was designed to analyze the difference between the degree of students' SoC and their LoU according to the their background variables. For this, 440 middle and high school students participated in the research. The results of this study were as follow: Firstly, since the students' SoC and LoU about the self-directed learning in science learning are low(Stage0 : awareness and Level II : preparation), we should draw students' immediate concern by developing training programs that would enable them to actually participate in the process of implementing the self-directed learning. Secondly, the students' SoC and LoU on self-directed learning in science learning vary depending on their residence, gender, and grade. This is the reason why we have to develop customized training programs on self-directed learning that suits their background. Thirdly, it shows that students, who have higher concern on self-directed learning in science learning, implement it better than those who are not concerned with it at all. It implies that we need a training program that considers both the students' concern and implementation on self-directed learning in science learning.

• Journal of IKEEE
• /
• v.28 no.3
• /
• pp.474-478
• /
• 2024

In this paper, we proposes a deep learning network for quality inspection in a multi-camera inline inspection system for pharmaceutical containers. The proposed deep learning network is specifically designed for pharmaceutical containers by using data produced in real manufacturing environments, leading to more accurate quality inspection. Additionally, the use of an inline-capable deep learning network allows for an increase in inspection speed. The development of the deep learning network for quality inspection in the multi-camera inline inspection system consists of three steps. First, a dataset of approximately 10,000 images is constructed from the production site using one line camera for foreign substance inspection and three area cameras for dimensional inspection. Second, the pharmaceutical container data is preprocessed by designating regions of interest (ROI) in areas where defects are likely to occur, tailored for foreign substance and dimensional inspections. Third, the preprocessed data is used to train the deep learning network. The network improves inference speed by reducing the number of channels and eliminating the use of linear layers, while accuracy is enhanced by applying PReLU and residual learning. This results in the creation of four deep learning modules tailored to the dataset built from the four cameras. The performance of the proposed deep learning network for quality inspection in the multi-camera inline inspection system for pharmaceutical containers was evaluated through experiments conducted by a certified testing agency. The results show that the deep learning modules achieved a classification accuracy of 99.4%, exceeding the world-class level of 95%, and an average classification speed of 0.947 seconds, which is superior to the world-class level of 1 second. Therefore, the effectiveness of the proposed deep learning network for quality inspection in a multi-camera inline inspection system for pharmaceutical containers has been demonstrated.