• Journal of Korea Multimedia Society
• /
• v.25 no.6
• /
• pp.819-832
• /
• 2022

Cross-domain recommender system transfers knowledge across different domains to improve the recommendation performance in a target domain that has a relatively sparse model. However, they suffer from the "negative transfer" in which transferred knowledge operates as noise. This paper proposes a novel Multiple Fusion-based Deep Cross-Domain Recommendation named MFDCR. We exploit Doc2Vec, one of the famous word embedding techniques, to fuse data user-wise and transfer knowledge across multi-domains. It alleviates the "negative transfer" problem. Additionally, we introduce a simple multi-layer perception to learn the user-item interactions and predict the possibility of preferring items by users. Extensive experiments with three domain datasets from one of the most famous services Amazon demonstrate that MFDCR outperforms recent single and cross-domain recommendation algorithms. Furthermore, experimental results show that MFDCR can address the problem of "negative transfer" and improve recommendation performance for multiple domains simultaneously. In addition, we show that our approach is efficient in extending toward more domains.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.165-165
• /
• 2023

Streamflow prediction is a critical task in water resources management and essential for planning and decision-making purposes. However, the streamflow prediction is challenging due to the complexity and non-linear nature of hydrological processes. The transfer learning is a powerful technique that enables a model to transfer knowledge from a source domain to a target domain, improving model performance with limited data in the target domain. In this study, we apply the transfer learning using the Informer model, which is a state-of-the-art deep learning model for streamflow prediction. The model was trained on a large-scale hydrological dataset in the source basin and then fine-tuned using a smaller dataset available in the target basin to predict the streamflow in the target basin. The results demonstrate that transfer learning using the Informer model significantly outperforms the traditional machine learning models and even other deep learning models for streamflow prediction, especially when the target domain has limited data. Moreover, the results indicate the effectiveness of streamflow prediction when knowledge transfer is used to improve the generalizability of hydrologic models in data-sparse regions.

• Journal of the Institute of Convergence Signal Processing
• /
• v.15 no.4
• /
• pp.119-127
• /
• 2014

Recently, various methods derived from the experiences and the practices to manage knowledge have been proposed. However, the effects of the most of those methods have not been demonstrated quantitatively, but only the successful cases reported. In this paper, a method for knowledge transfer in a group is proposed to evaluate the effect of knowledge management and to express a shape that knowledge is transferred in the group by the multi-agent simulation. Simulation results show that the performance of the entire group is improved by introducing a knowledge database and a knowledge manager. In addition, the experience knowledge of the entire group is shown to be lost by reviewing results of the social issues by the proposed method. As a countermeasure for that, it was confirmed that the introduction of the knowledge database and the knowledge manager is more effective than those of an extension of the retirement age and of an increase of the intermediate recruitment.

• Journal of the Korean Society for Aviation and Aeronautics
• /
• v.14 no.4
• /
• pp.48-59
• /
• 2006

This Study focused on the process of technology transfer for the aircraft development program by the strategic alliance. Especially, this study showed how the learning firms absorb new technology from the foreign leading company. This case study concludes that teachability, asset specificity, relation capital with partner, information sharing system in organization and knowledge base are key factors for absorbing the new technology from the technology leading partner.

• Asian Pacific Journal of Cancer Prevention
• /
• v.15 no.18
• /
• pp.7707-7711
• /
• 2014

Background: Currently poster presentations offer a common visual medium for knowledge transfer by a wide range of health professionals. Our study aimed to determine the scientific importance of poster presentations for Medical and Radiation Oncologists. Methods: A survey form including 40 questions was distributed to a total of 131 oncologists experienced in poster presentations. One hundred completed survey forms were included in the study. Descriptive statistics and modified thematic analyses were performed on the responses. Results: Overall 64% of the participants agreed that posters were a good medium for knowledge transfer. Some 88% agreed that concise and clear styled presentations would increase appealing interests for poster contents. Visual appearance was cited more influential than content of the subject; 70% of participants agreed that appearances of posters could help to draw more viewer attention. Of respondents, 63% believed that posters accompanied by their author were more attractive for congress attendees, and 33% of them declared that the halo effect of the poster presenter was also important. Conclusion: The present study indicated that intelligibility, appearance and visuality of posters are most important factors from the aspect of oncologist participants. Presenters must take into account these important points when preparing their academic posters.

• ETRI Journal
• /
• v.41 no.2
• /
• pp.242-253
• /
• 2019

We devise a layer-wise hint training method to improve the existing hint-based knowledge distillation (KD) training approach, which is employed for knowledge transfer in a teacher-student framework using a residual network (ResNet). To achieve this objective, the proposed method first iteratively trains the student ResNet and incrementally employs hint-based information extracted from the pretrained teacher ResNet containing several hint and guided layers. Next, typical softening factor-based KD training is performed using the previously estimated hint-based information. We compare the recognition accuracy of the proposed approach with that of KD training without hints, hint-based KD training, and ResNet-based layer-wise pretraining using reliable datasets, including CIFAR-10, CIFAR-100, and MNIST. When using the selected multiple hint-based information items and their layer-wise transfer in the proposed method, the trained student ResNet more accurately reflects the pretrained teacher ResNet's rich information than the baseline training methods, for all the benchmark datasets we consider in this study.

• Quality Improvement in Health Care
• /
• v.9 no.2
• /
• pp.116-133
• /
• 2002

Background : It is very common in Korea to take care of non-acute patients in an acute setting, due to the lack of long-term facilities. Long term hospitalization increase medical expenses and decreases the bed utilization, which can affect the urgent and emergent admissions, and eventually jeopardize the hospital financially. In this study, strategies for effective transfers to the lower levels of care, and to decrease the length of stay were presented by surveying and analyzing the patient's knowledge of the transfer needs, and the willingness to transfer those whose hospital length of stay was more than 30days. Method : The survey is subject to a group of 251 patients who have been hospitalized over 30 days in a general hospital in Seoul. Excluding those that were in the Intensive Care Unit and psychiatric ward, 214 in-patients were used as participants. They were surveyed from April 9, 2002 to April 17, 2002. One hundred and thirty seven out of 214 were responded which made the response rate 64%. Data were analyzed by SAS and SPSS. Result : Multi-variable Logistic Regression Analysis showed a significant effect in medical expenses, knowledge of referral system and the information of the receiving hospital. The financial burden in medical expenses made the patient 10.7 times more willing to be transferred, knowledge of the referral system made them 5 times more willing to be transferred, and the information of receiving hospital makes 6.5 times more willing to be transferred. Reasons for willing to be transferred to a lower level of care were the phase of physical therapy, the distance from home, the attending physician's advice and being unable to be treated as an out patient. Reasons for refusing to be transferred were the following. The attending physician's competency, not being ready to be discharged, not trusting the receiving hospital's competency due to the lack of information, or never hearing about the referring system by the attending physician. Conclusion : Based on this, strategies for the effective transfer to the lower levels of care were suggested. It is desirable for the attending physician to be actively involved by making an effort to explain the transfer need, and referring to the Healthcare Coordinating Center, which can help the patient make the right decision. Nationwide networking for the referral system is the another key factor that may need to be suggested as an alternative to decrease the medical expenses. Collaborating with the Home Health Agency for the early discharge planning and the Social Service Department for financial aid are also needed. It is recommended that the hospital should expedite the transfer process by prioritizing the cost and the information as medical expenses, knowledge of referring system and the information of the receiving hospital, are the most important factors to the willingness to transfer to a lower level of care.

• Knowledge Management Research
• /
• v.11 no.1
• /
• pp.1-17
• /
• 2010

As information technology had shown tremendous development in late 20th century, various service opportunities appeared in many industries. Also, new types of service are becoming available such as, reservation, teleconsultation, telemedicine. In health care industry, in which, many hospitals are faced operational difficulties and competing impetuously, a web site has become a effective tool to attract patients and transfer tremendous health information to the patients. This study is based on many previous researches on online service quality, try to figure out e-service quality factors of health information sites, and the factors' effect on users' satisfaction on the web site via providing knowledge and trust on the web site. As a result, usability, site aesthetic, responsiveness and security are the 4 factors to measure e-service quality of health information web site. All factors except site aesthetic have significant effects on providing knowledge, security only effects on trust on the web site.

• International Journal of Knowledge Content Development & Technology
• /
• v.4 no.2
• /
• pp.5-23
• /
• 2014

The instant messenger (IM) is frequently encountered as a facilitator of communication in both social and working contexts. Nevertheless, there are concerns about the extent to which IMs bring organizational benefits, thereby overcoming interruptions to work. In this study, we focus on how IM tools contribute to knowledge transfer at work, as well as their influence on team performance. Our research model is validated by a survey of work professionals (n=41) at a small sized company. Implications and conclusions are discussed.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.35 no.1
• /
• pp.63-76
• /
• 2011

The potential for brand dilution occurs when new brands merge with an identical image of a parent brand. Despite the significant attention regarding the effects of dilution on brand extension, there is limited research on brand dilution in the beauty industry. This study analyzes the effect of the typicality of extensional type on brand attitude or purchase intention for potential dilution toward beauty brand extension. In addition, we examine the moderating effect of knowledge transfer when customers evaluate the typicality of the extensional type for potential dilution toward beauty brand extensions. For the experiment, 4 description type factorial designs were performed. A total of 219 students participated in the experiment who had experience of visiting a beauty salon. The results are as follows. First, there are significant independent and interaction effects between the typicality and extensional type that reveals differential influences on the attitude or purchase intentions toward extended beauty brands. Second, the knowledge of the parent brand is transferred to extended brands. Third, there is a significant moderating effect of knowledge transfer, as customers evaluate the typicality of the extensional type for the potential dilution toward beauty brand extension. This study provides some theoretical and practical perspectives with some limitations.