• Proceedings of the Korean Magnestics Society Conference
• /
• 2007.05a
• /
• pp.61.2-61.2
• /
• 2007

• Journal of the Semiconductor & Display Technology
• /
• v.7 no.3
• /
• pp.11-16
• /
• 2008

The authors investigated an overhead shuttle (OHS) system for automatic transferring the liquid crystal display (LCD) panels. The constructed tracks of OHS system include the linear and curve regions and have been installed on the ceiling to transfer the cassette of LCD glass along the closed-loop and open-loop tracks. In this study, the OHS system was implemented by a proposed linear motor to solve encoder installation and the system cost problems of the long distance transfer system. In addition, we utilized a new algorithm of the position detection and a new control algorithm for driving linear motor. The newly developed control algorithm was demonstrated from both a computer simulation and an experimentation, indicating that the highly reliable and speedy transfer system can enhance the LCD panel productivity of commercial OHS system.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.46 no.4
• /
• pp.39-52
• /
• 2023

In this study, project information of government-funded research institute in the food field was collected and analyzed to systematically identify the factors affecting the process of transferring technological achievements of public research institute to the private sector. This study hypothesized that human resources, financial resources, and technological characteristics as input factors of R&D projects affect output factors, such as research papers or patents produced by R&D projects. Moreover, these outputs would serve as drivers of the technology transfer as one of the R&D outcomes. Linear Regression Analysis and Poisson Regression Analysis were conducted to empirically and sequentially investigate the relationship between input factors and output and outcome of R&D projects and the results are as follows: First, the principle investigator's career and participating researcher's size as human resource factors have an influence on both the number of SCI (science citation index) papers and patent registration. Second, the research duration and research expenses for the current year have an influence on the number of SCI papers and patent registrations, which are the main outputs of R&D projects. Third, the technology life cycle affects the number of SCI papers and patent registrations. Lastly, the higher the number of SCI papers and patent registrations, the more it affected the number of technology transfers and the amount of technology transfer contract.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.35 no.2
• /
• pp.1-8
• /
• 2012

The primary goal of this study is to develop proper performance indicators for evaluating the results of industrial R&D program. We define R&D results include output, transfer, outcome, and impact in the performance of R&D program. Performance indicators are focused on the results of R&D commercialization, building the industrial foundations as performance proxy of R&D program of Ministry of Knowledge Economy. And we first try to design the coupled structure as logical model between R&D program goals and performance indicators. The coupled structure is represented with objectives of group unit of R&D program, implementation process, R&D practical stage, and the ultimate goal of R&D program.

• Proceedings of the Korean Magnestics Society Conference
• /
• 2008.12a
• /
• pp.4.2-4.2
• /
• 2008

• Proceedings of the KSAR Conference
• /
• 2001.03a
• /
• pp.64-64
• /
• 2001

To facilitate the widespread application of somatic cell cloning, improvements in blastocyst production efficiency and subsequent fetal viability are required. Area where technical improvements are needed include donor cell treatments, starvation and passage numbers. This study was carried out to investigate the effect of serum-starvation and passage on the development of ear skin fibroblast cells cloned embryos. A skin biopsy was obtained from the ear of a 2-year-old Korean Hanwoo female. The cells were cultured in 10% FBS＋DMEM up to 2-3 months(up to 10 passages) and then used. In Experiment 1, the Korean bovine Ear Skin Fibroblast cells (KbESF) were either serum starved (culture in 0.05% FBS＋DMEM) or serum fed (10% FBS＋DMEM) for 4-7 days Prior to NT In Experiment 2, the KbESF cells used for nuclear transfer in these experiments were from passages 2 to 10. The development of 208 nuclear transfer (NT) embryos reconstructed from either serum starved or serum fed ear skin fibroblast was assessed. NT embryos reconstructed from serum starved and serum fed cells showed the same developmental rate (cleavage 80.16 vs. 85.37%; blastocyst 20.63 vs. 19,51%). The development of 590 nuclear transfer (NT) embryos reconstructed from passage 2 to 10 was assessed. We observed the same developmental rates for embryos derived from later Passages as compared with those embryos from early passages(blastocyst from 16.69 to 27.91%, average 20.17%). There was no significant difference between serum-fed and serum-starved donor cells. We observed no difference in developmental rates for embryos derived from 2 to 10 passages. These data show that prolonged culture and serum starvation does not affects the cloning competence of adult somatic cells.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.24 no.4
• /
• pp.382-391
• /
• 2021

Unsupervised domain adaptation often gives impressive solutions to handle domain shift of data. Most of current approaches assume that unlabeled target data to train is abundant. This assumption is not always true in practices. To tackle this issue, we propose a general solution to solve the domain gap minimization problem without any target data. Our method consists of two regularization steps. The first step is a pixel regularization by arbitrary style transfer. Recently, some methods bring style transfer algorithms to domain adaptation and domain generalization process. They use style transfer algorithms to remove texture bias in source domain data. We also use style transfer algorithms for removing texture bias, but our method depends on neither domain adaptation nor domain generalization paradigm. The second regularization step is a feature regularization by feature alignment. Adding a feature alignment loss term to the model loss, the model learns domain invariant representation more efficiently. We evaluate our regularization methods from several experiments both on small dataset and large dataset. From the experiments, we show that our model can learn domain invariant representation as much as unsupervised domain adaptation methods.

• Journal of Korea Foundry Society
• /
• v.24 no.5
• /
• pp.281-289
• /
• 2004

The present study focused on the estimation of the interfacial heat transfer coefficient as a function of the surface temperature of the aluminum casting at the mold/casting interface to investigate the effects of superheat and coating layer. The casting experiments of aluminum into a cylindrical copper mold were systematically conducted to obtain the thermal history during solidification. The thermal history recorded by four thermocouples embedded both in the mold and the casting was used to solve the inverse heat conduction problem using Beck's method. The effects of superheat and coating on the interfacial heat transfer coefficient in the liquid state, during the solidification, and in the solid state were comparatively discussed. In the liquid state, the interfacial heat transfer coefficient is thought to be affected by the roughness of the mold, the wettability of the casting on the mold surface, and the thermophysical properties of the coating layer. When the solidification begins, the air gap forms between the casting and the mold, and the interfacial heat transfer coefficient becomes a function of the air gap as well as surface roughness and the superheat. In the solid phase, it depends only upon the thermal conductivity and the thickness of the air gap. The coating layer reduces seriously the interfacial heat transfer coefficient in the liquid state and during the solidification.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.40 no.2
• /
• pp.31-40
• /
• 2017

In this paper, we compared the efficiencies of national R&D investments between NT (Nanotechnology) areas in terms of papers, patents, and commercializations, and found ways to improve the efficiencies of national R&D investments for each NT area. This is in response to huge R&D investments government has made recently in NT areas. Here, we collected data on investments, papers, patents, and commercializations for the R&D projects in NT areas through National Science & Technology Information Service. Based on the data, we analyzed the investment and performances (papers, patents, and commercializations) for each NT area, calculated the efficiency for each NT area, and compared the efficiencies between NT areas. Next, using cluster analysis, we identified several NT areas with similar characteristics in terms of paper efficiency, patent efficiency and commercialization efficiency. Finally, we derived implications for the efficiency enhancement for each grouping. The cluster analysis showed that there could be two groups, one being low in terms of technological outcome (papers and patents) efficiencies and high in terms of commercialization efficiencies, while the other being high in terms of technological outcome (papers and patents) efficiencies and low in terms of commercialization efficiencies. Therefore, the strategy for one group calls for support for technology transfer or technology introduction from other R&D performers and grant of guidance for improving R&D performers' commercialization ability to other R&D performers while the strategy for the other group calls for R&D support for transfer of technology to other R&D performers, activation of technology transfer and support for commercialization of R&D performers.

• Knowledge Management Research
• /
• v.22 no.3
• /
• pp.91-106
• /
• 2021

With the recent rapid development of deep learning technology, the demand for analyzing huge text documents in the national R&D field from various perspectives is rapidly increasing. In particular, interest in the application of a BERT(Bidirectional Encoder Representations from Transformers) language model that has pre-trained a large corpus is growing. However, the terminology used frequently in highly specialized fields such as national R&D are often not sufficiently learned in basic BERT. This is pointed out as a limitation of understanding documents in specialized fields through BERT. Therefore, this study proposes a method to build an R&D KoBERT language model that transfers national R&D field knowledge to basic BERT using further pre-training. In addition, in order to evaluate the performance of the proposed model, we performed classification analysis on about 116,000 R&D reports in the health care and information and communication fields. Experimental results showed that our proposed model showed higher performance in terms of accuracy compared to the pure KoBERT model.