• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.1
• /
• pp.1-15
• /
• 2022

The MobileNetV3 is specially designed for mobile devices with limited memory and computing power. To reduce the network parameters and improve the network inference speed, a new lightweight network is proposed based on MobileNetV3. Firstly, to reduce the computation of residual blocks, a partial residual structure is designed by dividing the input feature maps into two parts. The designed partial residual structure is used to replace the residual block in MobileNetV3. Secondly, a dual-path feature extraction structure is designed to further reduce the computation of MobileNetV3. Different convolution kernel sizes are used in the two paths to extract feature maps with different sizes. Besides, a transition layer is also designed for fusing features to reduce the influence of the new structure on accuracy. The CIFAR-100 dataset and Image Net dataset are used to test the performance of the proposed partial residual structure. The ResNet based on the proposed partial residual structure has smaller parameters and FLOPs than the original ResNet. The performance of improved MobileNetV3 is tested on CIFAR-10, CIFAR-100 and ImageNet image classification task dataset. Comparing MobileNetV3, GhostNet and MobileNetV2, the improved MobileNetV3 has smaller parameters and FLOPs. Besides, the improved MobileNetV3 is also tested on CPU and Raspberry Pi. It is faster than other networks

• Korean Journal of Social Welfare Studies
• /
• v.40 no.2
• /
• pp.253-283
• /
• 2009

This study examined empirically the effect mechanism of labor market structure on access to social insurance. Data were derived from the second wave of Korea Welfare Panel Study in 2007 with a sub-sample of 3,790 of wage workers. Path analysis and structural equation model were used to analyze the effect mechanism. Findings confirmed the inequality of labor market by revealing high causal relationships among occupations and jobs, size of the firm, employment forms, and income. The results suggest that access to social insurance depends on the structural interaction of labor market rather than on the individual worker's trait. Small enterprises and non-standard employment were indicated as the main element explaining the unequal access to social insurance. Finally this study identified women's experience of intensive inequalities in the labor market status and access to social insurance.

• Journal of Astronomy and Space Sciences
• /
• v.11 no.1
• /
• pp.71-80
• /
• 1994

It is generally known that the measurement of the ionospheric total electron content(TEC) by GPS can more accurately monitor the broader area of the ionosphere than other current methods. \Ve measured the TEC along a slant path considering the arrival time differences of P-code which is transmitted from GPS satellites with the modulation on two L-band carrier frequencies, L1 (1574.42MHz) and L2 (1227.60MHz). Under the assumptions that the ionosphere is uniformly distributed and its average height is 350km, we transformed the slant TEC to the vertical TEC at the point that the line-of-sight direction to GPS satellite cut across the average height of the ionosphere. Because there is no dual frequency P-code GPS receiver in Korea, we used the data observed at the TAIW GPS station ($N25^{\circ},E121.5^{\circ}$) in Taiwan which is one of the core stations in International GPS and Geodynamics Services (IGS). The TEC values obtained in this work showed a typical daily variation of the ionosphere which is high in the daytime and low in the nighttime. Our results are found to be consistent with the SOLAR-DAILY data of NOAA and the Klobuchar's model for the ionospheric correction of GPS. In addition, in the cornparision with SOLAR-DAILY data, we estimated the precision of our TEC measurement as 2 TEC.

• Journal of Biomedical Engineering Research
• /
• v.20 no.2
• /
• pp.173-182
• /
• 1999

Contrast-enhanced CT has an important role in assessing liver lesions, the optimal protocol to get most effective result is not clear. The mein goal when deciding injention protocol is to optimize lesion detectability with rapid scanning when lesion to liver contrast is maximum. For this purpose, we developed a physiological model of the contrast medium enhancement based on the compartment modeling and pharmacokinetics. Blood supply to liver is achieved in two paths. This dual supply characteristic distinguishes the CT enhancement of liver from that of the other organs. The first path is by hepatic artery and to second, by portal vein. However, it is assumed that only gepatic artery can supply blood to hepatocellular carcinoma(HCC) compartment, thus, the difference of contrast enhancement is resulted between normal liver tissue and hepatic tumor. By solving differential equations for each compartment simultaneously using the computer program Matlab, CT contrast-enhancement curves were simulated. The simulated enhancement curves for aortic, hepatic, portal vein, and HCC compartments were compared with the mean enhancement curves from 24 patients exposed to the same protocols as the simulation. These enhancement curves showed a good agreement. Furthermore, we simulated lesion-to-liver curves for various injection protocols, and the effects were analyzed. The variables to be considered in the injection protocol were injection rate, dose, and concentration of contrast material. These data may help to optimize scanning protocols for better diagnosis.

• Progress in Medical Physics
• /
• v.13 no.1
• /
• pp.1-8
• /
• 2002

Accurate knowledge of the distribution of contamination electrons ( which comes from mainly gantry head by Compton scattering, pair production, and tray: henceforth called leptons ) at the surface and in the first centimeters of tissue is essential for the clinical practice of radiation oncology. Such lepton tends to reduce or eliminate the ‘skin-sparing’ advantage of megavoltage photon beam radiotherapy, This information is needed to prescribe a absorbed dose to a skin volume at a few millimeter depth in high energy therapeutic radiation photon beam All experiments were done with 15 MV photon beam from a dual energy linear accelerator (Clinac 1800, Varian). Field size is defined by ranged from 10.0$\times$10.0 to 30.0$\times$30.0 $\textrm{cm}^2$. The absorbed dose and distribution of leptons in therapeutic radiation beam (15 MV) are investigated by means of variable blocked beams of 30.0$\times$30.0 $\textrm{cm}^2$ and dose beam profiles partly removed leptons with a copper plate. A numerous leptons mainly are distributed as shape of broad cone in the central photon beam and leptons path length in the water are shorter than 2.5 cm because of the leptons energy having around 3.0 MeV. These results clearly appears that the subtraction of leptons from the total depth dose curve not only lower the absolute dose in the buildup region and surface dose, it also causes a shift of d$_{max}$ to a deeper depth.

• Korean Journal of Remote Sensing
• /
• v.25 no.1
• /
• pp.45-59
• /
• 2009

This research presents the effect of Asian dust on the derived sea surface temperature (SST) from measurements of the Advanced Very High Resolution Radiometer (AVHRR) instrument flown onboard NOAA polar orbiting satellites. To analyze the effect, A VHRR infrared brightness temperature (TB) is estimated from simulated radiance calculated from radiative transfer model on various atmospheric conditions. Vertical profiles of temperature, pressure, and humidity from radiosonde observation are used to build up the East Asian atmospheric conditions in spring. Aerosol optical thickness (AOT) and size distribution are derived from skyradiation measurements to be used as inputs to the radiative transfer model. The simulation results show that single channel TB at window region is depressed under the Asian dust condition. The magnitude of depression is about 2K at nadir under moderate aerosol loading, but the magnitude reaches up to 4K at slant path. The dual channel difference (DCD) in spilt window region is also reduced under the Asian dust condition, but the reduction of DCD is much smaller than that shown in single channel TB simulation. Owing to the depression of TB, SST has cold bias. In addition, the effect of AOT on SST is amplified at large satellite zenith angle (SZA), resulting in high variance in derived SSTs. The SST depression due to the presence of Asian dust can be expressed as a linear function of AOT and SZA. On the basis of this relationship, the effect of Asian dust on the SST retrieval from the conventional daytime multi-channel SST algorithm can be derived as a function of AOT and SZA.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.29 no.4D
• /
• pp.491-497
• /
• 2009

This paper presents the results of accelerated pavement tests (APT) that simulate permanent deformation (rutting) of asphalt concrete pavements under different temperatures and loading courses. Also, finite element (FE) analysis has been conducted to predict the test results. Test section for APT is the same as one of test sections at Korea Expressway Corporation test road and is subjected to a constant moving dual tire wheel load of APT at three different temperatures: 30, 40, $50^{\circ}C$. The moving wheel is applied at different loading courses within a 75cm wide wheel path to account for traffic wandering. Also, the effect of wandering on permanent deformation development is investigated numerically with three wandering schemes. In this study, ABAQUS is adopted to model APT pavement section with plain stain elements and creep strain rate model is used to take into account viscoplastic stain of asphalt concrete mixtures, and elastic layer properties are back-calculated from FWD measurements. Plus, the effect of boundary condition and subgrade on FE permanent deformation predictions is investigated. A full FE model that accounted for subgrade provided more realistic rut depth predictions, indicating subgrade has contributed to surface rutting.

• (The) Korean Journal of Educational Psychology
• /
• v.32 no.1
• /
• pp.99-130
• /
• 2018

The purpose of this study was to examine the longitudinal relationship between resilience and social support of school violence exposure types including school bullying, victimization, and dual experience. The study used data obtained from the third year (2012) of the Seoul Education Longitudinal Study of 1,137 elementary school students in grade 6 who reported experiencing school violence. The results of the autoregressive cross-lagged model are as follows. First, as a result of measuring the self-regression coefficients of resilience and social support of the youth exposed to school violence at 3 time points (2012, 2014, and 2016), it was found for all types of violence that resilience and social support at the previous time point showed a signigicant positive effect on the same variable at the next time point. Second, in the case of the cross-lagged effects of resilience and social support, the effect of previous social support on resilience at the next time point was statistically significant for the victimization group, but not for the bullying or dual experience groups. Third, considering the opposite path from resilience to social support, resilience at the previous time point had a significant influence on the social support at the next time point for both the bullying and victimization groups. This result is new and can be complementary to the cross-sectional studies so far using a longitudinal view. The results of this study suggest that the bullying and victimized students who are relatively more resilient are less likely to perceive social support than those who are not resilient. Finally, we discuss the longitudinal relationship between resilience and social support, the limitations of this study, and implications for future research.

• Asia-Pacific Journal of Business Venturing and Entrepreneurship
• /
• v.19 no.1
• /
• pp.75-91
• /
• 2024

Recently, global business environment is changing dramatically along with the acceleration of technological innovation amid the war, climatic change, and geopolitical instability. Accordingly, it is difficult to predict or plan for the future as the volatility, complexity, ambiguity, and uncertainty of the industrial ecosystem continue to increase. Therefore, organizations are undergoing inevitable restructuring in accordance with their survival strategy, for instance, removing marginal businesses or firing. Accordingly, office workers are seeking a startup as an alternative for their continuous economic activity amid rising anxiety factors that make them think they would lose their jobs unintentionally. Here, this study is aimed to verify through what paths office workers' self-determination influences the process of converting to a startup. For this study, an online survey was carried out, and 310 respondents' valid data were analyzed through SPSS and AMOS. To sum up the results, first, office workers' self-determination did not have significant effects on entrepreneurial intention. However, it was confirmed that self-determination had positive (+) effects on innovativeness and perception of the startup support system. This result shows that their psychology works to prepare step by step by accumulating innovative experiences and increasing perception of the startup support system from a long-term life path perspective rather than challenging startups right way. Second, innovativeness is found to have positive (+) effects on entrepreneurial intention. Also, perception of the startup support system had positive (+) effects on entrepreneurial intention. This implies that when considering startups, they are highly aware of the government's various startup support systems. Third, innovativeness is found to have positive (+) effects on perception of the startup support system. It is judged that perception of the startup support system is valid for prospective founders to exhibit their innovativeness and realize new ideas. Fourth, it was confirmed that innovativeness and perception of the startup support system mediated correlation between self-determination and entrepreneurial intention, and perception of the startup support system mediated correlation between innovativeness and entrepreneurial intention, which shows that it is a crucial factor in entrepreneurial intention. Although previous studies related to startups deal with students mostly, this study targets office workers who form a great part in economic activities, which makes it academically valuable in terms of being differentiated from others and extending the scope of research. Also, when we consider the fact that the motivation for self-determination alone fails to stimulate entrepreneurial intention and the complete mediation of innovativeness and the startup support system, it has great implications in practical aspects such as the government's human and material support systems. In the selection and analysis of samples, this study exhibits a limitation that the problem of common method bias is not completely resolved. Also, additional definitive research is needed on whether entrepreneurial intention is formed and converted into startup behavior. Academically and practically, this study deals with the relationship between humans' psychological motives and startups which has not been handled sufficiently in previous studies. The conversion of office workers to startups is expected to have effects on individuals' economic stability and the state's job creation; therefore, it needs to be investigated continuously for its great value.