• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.40 no.1
• /
• pp.19-29
• /
• 2003

This paper presents a novel hybrid classifier for face recognition with artificially generated virtual training samples. We utilize both the nearest neighbor approach in feature angle space and a connectionist model to obtain a synergy effect by combining the results of two heterogeneous classifiers. First, a classifier called the nearest feature angle (NFA), based on angular information, finds the most similar feature to the query from a given training set. Second, a classifier has been developed based on the recall of stored frontal projection of the query feature. It uses a frontal recall network (FRN) that finds the most similar frontal one among the stored frontal feature set. For FRN, we used an ensemble neural network consisting of multiple multiplayer perceptrons (MLPs), each of which is trained independently to enhance generalization capability. Further, both classifiers used the virtual training set generated adaptively, according to the spatial distribution of each person's training samples. Finally, the results of the two classifiers are combined to comprise the best matching class, and a corresponding similarit measure is used to make the final decision. The proposed classifier achieved an average classification rate of 96.33% against a large group of different test sets of images, and its average error rate is 61.5% that of the nearest feature line (NFL) method, and achieves a more robust classification performance.

• Journal of IKEEE
• /
• v.14 no.2
• /
• pp.51-57
• /
• 2010

A CMOS rail-to-rail high voltage buffer amplifier is proposed to drive the gamma correction reference voltage of large TFT LCD panels. It is operating by a single supply and only shows current consumption of 0.5mA at 18V power supply voltage. The circuit is designed to drive the gamma correction voltage of 8-bit or 10-bit high resolution TFT LCD panels. The buffer has high slew rate, 0.5mA static current and 1k$\Omega$ resistive and capacitive load driving capability. Also, it offers wide supply range, offset voltages below 50mV at 5mA constant output current, and below 2.5mV input referred offset voltage. To achieve wide-swing input and output dynamic range, current mirrored n-channel differential amplifier, p-channel differential amplifier, a class-AB push-pull output stage and a input level detector using hysteresis comparator are applied. The proposed circuit is realized in a high voltage 0.18um 18V CMOS process technology for display driver IC. The circuit operates at supply voltages from 8V to 18V.

• Korean Journal of Soil Science and Fertilizer
• /
• v.36 no.5
• /
• pp.290-303
• /
• 2003

Modern agriculture has been heavily dependent on chemical fertilizers to meet the food demands of ever increasing population. Progressive depletion of major plant nutrients in soil due to intensive cultivation practices has also necessitated the use of higher dose of chemical fertilizers particularly in soils where the organic matter content is very low. Indiscriminate use of chemical fertilizers and pressure on agriculturists to enhance per area crop yields has led to fast depletion of fossil fuel resources with concomitant increase in the prices of chemical fertilizers and also led to environmental pollution. Hence, the current trend throughout the world is to explore the possibility of using alternate nutrient sources or increasing the efficiency of chemical fertilizers by supplementing them with organic fertilizers and bioinoculants comprising largely microbes like, bacteria, fungi, algae etc to enhance nitrogen and phosphates in the soil thus creating a sustainable agricultural environment. Among the different microbial inoculants or biofertilizers, Azospirillum could be a potential candidate due to its non specific host root colonization. It had the capability to fix $N_2$ in wide pH regimes and even in presence of combined nitrogen. Azospirillum inoculation can increase the crop yield to 10-25% and substitute 25% of recommended doses of nitrogenous fertilizers. Apart from nitrogen fixation, Azospirillum is also involved in the root improvement, the activity which was attributed to an increase in the rate of water and mineral uptake by roots. The ability of Azospirillum to produce phytohormones was reported to enhance the root respiration rate, metabolism and root proliferation. They have also been reported to produce polyhydroxybutyrate, that can be used as a biodegradable thermosplastic. A lot of studies have addressed improvements in enhancing its efficiency to fix nitrogen fixation and hormone production.

• Tunnel and Underground Space
• /
• v.29 no.3
• /
• pp.197-213
• /
• 2019

We simulated the fault reactivation experiment conducted at 'Main Fault' intersecting the low permeability clay formations of Mont Terri Underground Research Laboratory in Switzerland using TOUGH-FLAC simulator. The fluid flow along a fault was modelled with solid elements and governed by Darcy's law with the cubic law in TOUGH2, whereas the mechanical behavior of a single fault was represented by creating interface elements between two separating rock blocks in FLAC3D. We formulate the hydro-mechanical coupling relation of hydraulic aperture to consider the elastic fracture opening and failure-induced dilation for reproducing the abrupt changes in injection flow rate and monitoring pressure at fracture opening pressure. A parametric study was conducted to examine the effects of in-situ stress condition and fault deformation and strength parameters and to find the optimal parameter set to reproduce the field observations. In the best matching simulation, the fracture opening pressure and variations of injection flow rate and monitoring pressure showed good agreement with field experiment results, which suggests the capability of the numerical model to reasonably capture the fracture opening and propagation process. The model overestimated the fault displacement in shear direction and the range of reactivated zone, which was attributed to the progressive shear failures along the fault at high injection pressure. In the field experiment results, however, fracture tensile opening seems the dominant mechanism affecting the hydraulic aperture increase.

• Korean Chemical Engineering Research
• /
• v.57 no.1
• /
• pp.118-123
• /
• 2019

In this study, the electrochemical characteristics of Graphite/Silicon/Pitch anode composites were analyzed to improve the low theoretical capacity of graphite as a lithium ion battery. The Graphite/Silica composites were synthesized by bonding silica onto polyvinylpyrrolidone coated graphite. The surface of used silica was treated with (3-Aminopropyl)triethoxysilane(APTES). Graphite/Silicon/Pitch composites were prepared by carbonization of petroleum pitch, the fabrication processes including the magnesiothermic reduction of nano silica to obtain silicon and varying the mass ratio of silica. The Graphite/Silicon/Pitch composites were analysed by XRD, SEM and XRD. Also the electrochemical performances of Graphite/Silicon/Pitch composite as the anode of lithium ion battery were investigated by constant current charge/discharge, rate performance, cyclic voltammetry and electrochemical impedance tests in the electrolyte of $LiPF_6$ dissolved in organic solvents (EC:DMC:EMC=1:1:1 vol%). The Graphite/Silicon/Pitch anode composite (silica 28.5 in weight) has better capacity (537 mAh/g). The cycle performance has an excellent capacity retention to 30th cycle of 95% and the retention rate capability of 98% in 0.1 C/0.2 C.

• Clean Technology
• /
• v.24 no.4
• /
• pp.332-338
• /
• 2018

Currently graphite is used as an anode active material for lithium ion battery. However, since the maximum theoretical capacity of graphite is limited to $372mA\;h\;g^{-1}$, a new anode active material is required for the development of next generation high capacity and high energy density lithium ion battery. The maximum theoretical capacity of Si is $4200mA\;h\;g^{-1}$, which is about 10 times higher than the maximum theoretical capacity of graphite. However, since the volume expansion rate is almost 400%, the irreversible capacity increases as the cycle progresses and the discharge capacity relative to the charge is remarkably reduced. In order to solve these problems, it is possible to control the particle size of the Si anode active material to reduce the mechanical stress and the volume change of the reaction phase, thereby improving the cycle characteristics. Therefore, in order to minimize the decrease of the charge / discharge capacity according to the volume expansion rate of the Si particles, the improvement of the cycle characteristics was carried out by pulverizing Si by a dry method with excellent processing time and cost. In this paper, Si is controlled to nano size using vibrating mill and the physicochemical and electrochemical characteristics of the material are measured according to experimental variables.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.14 no.3
• /
• pp.201-207
• /
• 2021

Deep learning, a type of machine learning, performs learning while changing the weights as it progresses through each learning process. Tensor Flow and Keras provide the results of the end of the learning in graph form. Thus, If an error occurs, the result must be discarded. Consequently, existing technologies provide a function to roll back learning results, but the rollback function is limited to results up to five times. Moreover, they applied the concept of MLOps to track the deep learning process, but no rollback capability is provided. In this paper, we construct a system that manages the intermediate value of the learning process by blockchain to record the intermediate learning process and can rollback in the event of an error. To perform the functions of blockchain, the deep learning process and the rollback of learning results are designed to work by writing Smart Contracts. Performance evaluation shows that, when evaluating the rollback function of the existing deep learning method, the proposed method has a 100% recovery rate, compared to the existing technique, which reduces the recovery rate after 6 times, down to 10% when 50 times. In addition, when using Smart Contract in Ethereum blockchain, it is confirmed that 1.57 million won is continuously consumed per block creation.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.6
• /
• pp.147-154
• /
• 2021

Recently, it has been observed that as the level of Science and Technology in Korea is improving, the demand for technology transfer from overseas buyers is also increasing. A technical value is an important factor for the technology transfer process and the valuation of technology should be performed reasonably. Specifically, a non-economical value has to be examined thoroughly when conducting the valuation for a technology that depends on public values. Since public technology has public benefit as its purpose when compared to technology from the private sector, its discount rate should be appropriately assessed and reflected in its valuation process. In this context, this study presents the methodology of valuation of public technology particularly relating to the transfer of technology from the national defense industries. To be specific, both an application method of the discount rate according to the characteristics and the purpose of the target technology and a qualitative and quantitative evaluation method to reflect the public values are presented. The proposed method for the valuation of defense technology could be used practically both in strengthening bargaining capability based on the reasonably derived technical values for transfer of national defense technology abroad and in the compilation of budgets for technology development in the future.

• Applied Chemistry for Engineering
• /
• v.32 no.1
• /
• pp.83-90
• /
• 2021

In this study, effects of the physical and chemical properties of low temperature heated carbon on electrochemical behavior as a secondary battery anode material were investigated. A heat treatment at 600 ℃ was performed for coking of petroleum based pitch, and the manufactured coke was heat treated with different heat temperatures at 700~1,500 ℃ to prepare low temperature heated anode materials. The physical and chemical properties of carbon anode materials were studied through nitrogen adsorption and desorption, X-ray diffraction (XRD), Raman spectroscopy, elemental analysis. Also the anode properties of low temperature heated carbon were considered through electrochemical properties such as capacity, initial Coulomb efficiency (ICE), rate capability, and cycle performance. The crystal structure of low temperature (≤ 1500 ℃) heated carbon was improved by increasing the crystal size and true density, while the specific surface area decreased. Electrochemical properties of the anode material were changed with respect to the physical and chemical properties of low temperature heated carbon. The capacity and cycle performance were most affected by H/C atomic ratio. Also, the ICE was influenced by the specific surface area, whereas the rate performance was most affected by true density.

• The Journal of the Korea Contents Association
• /
• v.21 no.3
• /
• pp.184-193
• /
• 2021

This paper investigates the performance of technology innovation activities performed by firms in 16 major regions in Korea using 2002-2010 survey data by STEPI. The theoretical and empirical analysis is carried out via the 2 models which are the simple R&D - total revenue model and Cobb-Douglas model based on the simple model adding labor variable. The main results shows that for simple model, the R&D elasticity for total revenue is 0.42 for all areas and Ul-San shows the highest elasticity level, 0.66 and Bu-San the lowest level, 0.2. In case of Cobb Douglas model the R&D elasticities are not statistically significant for many regions. To overcome the low statistical significance, we grouped the 15 regions for 3 wider regions using ANOVA based on the R&D intensity for the homogeneity of R&D activities. By grouping, each region has more observations to analyze and the results from the empirical analysis shows higher statistical significance level and data explanation capability. In this case, Group 3 which shows larger firm size and slightly higher export share shows the highest level of R&D elasticity, 0.088 and Group 1 which has the smallest firm size and the lowest revenue growth rate shows the lowest level, 0.31. For the labor elasticity, Group 1 shows the higest level, 1.16 and Group2 the lowest level, 1.096. These results show that the regions which have many middle and small firms reveal low R&D-revenue elasticity and high labor-revenue elasticity.