• The Korean Journal of Petroleum Geology
• /
• v.5 no.1_2 s.6
• /
• pp.9-15
• /
• 1997

The study area is located in the Central Bransfield Basin, Antarctica. To analyze the morphology of seafloor, structure of basement, and seismic stratigraphy of the sedimentary layers, we have acquired, processed, and interpreted the multi-channel seismic data. The northwest-southeastern back-arc extension dramatically changes seafloor morphology, volcanic and fault distribution, and basin structure along the spreading ridges. The northern continental shelf shows a narrow, steep topography. In contrast, the continental shelf or slope in the south, which is connected to the Antarctic Peninsula, has a gentle gradient. Volcanic activities resulted in the formation of large volcanos and basement highs near the spreading center, and small-scale volcanic diapirs on the shelf. A very long, continuous normal fault characterizes the northern shelf, whereas several basinward synthetic faults probably detach into the master fault in the south. Four transfer faults, the northwest-southeastern deep-parallel structures, controlled the complex distributions of the volcanos, normal faults, depocenters, and possibly hydrocarbon provinces in the study area. They have also deformed the basement structure and depositional pattern. Even though the Bransfield Basin was believed to be formed in the Late Cenozoic (about 4 Ma), the hydrocarbon potential may be very high due to thick sediment accumulation, high organic contents, high heat flow resulted from the active tectonics, and adequate traps.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Journal of Korean Society of Environmental Engineers
• /
• v.31 no.4
• /
• pp.249-255
• /
• 2009

The performance of microbial fuel cell (MFC) can be affected by many factors including the rate of organic matter oxidation, the electron transfer to electrode by electrochemical bacteria, proton diffusion, the concentration of electron acceptor, the rate of electron acceptor reduction and internal resistance. the performance of MFC using oxygen as electron acceptor can be influenced by oxygen concentration as limit factors in cathode compartment. Many studies have been performed to enhance electricity production from MFC. The series or parallel stacked MFC connected several MFC units can use to increase voltages and currents produced from MFCs. In this study, a single MFC (S-MFC) and a stacked MFC (ST-MFC) using acetate as electron donor and oxygen as electron acceptor were used to investigate the influence of dissolved oxygen (DO) concentrations in cathode compartment on MFC performance. The power density (W/$m^3$) of S-MFC was in order DO 5 > 3 > 7 > 9 mg/L, the maximum power density (W/$m^3$) of S-MFC was 42 W/$m^3$ at DO 5 mg/L. The power density (W/$m^3$) of ST-MFC was in order DO 5 > 7 > 9 > 3 mg/L and the maximum power density (W/$m^3$) of STMFC was 20 W/$m^3$ at DO 5 mg/L. These results suggest that the DO concentration of cathode chamber should be considered as important limit factor of MFC operation and design for stacked MFC as well as single MFC. The results of ST-MFC operation showed the voltage decrease of some MFC units by salt formation on the surface of anode, resulting in decrease total voltage of ST-MFC. Therefore, connecting MFC units in parallel might be more appropriate way than series connections to enhance power production of stacked MFC.

• Journal of Practical Agriculture & Fisheries Research
• /
• v.4 no.1
• /
• pp.119-127
• /
• 2002

This study was conducted to develop the heat exchanger by utilizing the heat energy of underground water(15℃), which might be used for cooling and heating system of the agricultural facilities. We developed the heat exchanger, parallel type plat fin tube made of Aluminum(Al 6063), which was named Aloo-Heat(No. of The registration design : 0247164, by Korean Intellectual property Office). The fin of exchanger was design of the granulated surface for minimizing fouling factor and dew forms, and also placed parallel to the tube in order to minimized the resistance of flows. 1. Aloo-heat was designed to have 0.03m for inside diameter, 0.036m for outside diameter of tube, 0.0012m for thickness of fin and 0.032m for length of plat fin. 2. t was also designed to have 1.5248m²/m for outside area of heat transfer, 0.0942m²/m for inside area contacting hot liquid, and the ratio (Ra) was 16.1869. 3. Efficiency of the fin was 93 percentage when fin length was 0.032m, and the fin thickness satisfied equation $\frac{h{\rho}}{k}$< 0.2 when it was 0.0012m. 4. According to the performance test of Aloo-heat, as the temperature and rate increased, the heating value also increased, heating value was 504kJ/h·m and 6,048kJ/h·m when it was 60℃, 10 𝑙/min and 80℃, 40 𝑙/min respectively. 5. The test of heating value was confident, because correlation value(R²) was 0.9898 for the temperature and 0.9721 for flow rate of hot liquid, respectively.

• Korean Journal of Optics and Photonics
• /
• v.26 no.3
• /
• pp.162-167
• /
• 2015

In this paper, we developed a MTF (Modulation Transfer Function) measurement system using a knife-edge scanning method for infrared optics. It consists of an objective part to generate the target image, a collimator to make the beam parallel, and a detector to analyze the image. We used a tungsten filament as the light source and MCT (Mercury Cadmium Telluride) to detect the mid-infrared(wavelength $3-5{\mu}m$) image. We measured the MTF of a standard lens (f=5, material ZnSe) to test this instrument and compared the result to the theoretical value calculated using the ZEMAX commercial software. It was found that the difference was within ${\pm}0.035$ at the cut-off frequency (50 1/mm). Also, we calculated the A-type measurement uncertainty to check the reliability of the measurement. The result showed only 0.002 at 20 1/mm in spatial frequency, which means very little variation in the MTF measurement under the same conditions.

• Journal of The Korean Association of Information Education
• /
• v.20 no.6
• /
• pp.629-644
• /
• 2016

The purpose of this study was to investigate the effects of whole-task sequencing emphasis manipulation on expertise acquisition in web based complex task. To achieve the purpose, emphasis manipulation sequencing type is composed of a simple emphasis manipulation, a snowballing manipulation, and a full emphasis manipulation sequencing and participants was drawn from a pool of 93 undergraduate students sampled for the study. According to the findings, a snowballing manipulation group invested significantly lower cognitive load than a full emphasis manipulation group but did not a simple emphasis manipulation group. Based on these findings, though complex task is included of high interactivity owing to real task, learner cannot suffer cognitive overload because emphasis manipulation which can view the whole task and the part task in parallel provides meta cognition for learner. And whole-task sequencing emphasis manipulation affects to transfer. The snowballing emphasis manipulation group invested significantly higher than simple emphasis manipulation group and full emphasis manipulation group. Based on these findings, the snowballing manipulation which learner use whole-task sequencing and part-task sequencing simultaneously contribute to understandings and ability to solve problems for complex task and it will in turn, lead to expertise acquisition.

• Journal of the Korean Wood Science and Technology
• /
• v.44 no.5
• /
• pp.776-784
• /
• 2016

In this study, green Larix kaempferi lumber was heat-treated by using superheated steam (SHS) at a pilot scale and then various physico-mechanical properties of the heat-treated wood were evaluated and compared with the properties of conventional hot air (HA) heat-treated wood. Decay resistance of brown rot fungi and compressive strength parallel to the grain of the SHS heat-treated wood without occurrence of drying check from green lumber were increased. On the other hand, density, equilibrium moisture content, shrinkage, and bending strength of the SHS heat-treated wood were lower than those of the conventional HA heat-treated wood. Because heat transfer and thermal hydrolysis of SHS heat treatment was accelerated by a large amount of water, the effect of SHS heat treatment on the physico-mechanical properties was higher than that of HA heat treatment at the similar conditions of temperature and time. From the results of this study, because green lumber can be heat-treated without occurrence of cracks or checks by using SHS and similar heat treatment effect on the physico-mechanical properties of wood can be produced despite a low temperature or short time of heat treatment, it is expected that heat time and energy consumption could be reduced by using SHS.

• Journal of Environmental Science International
• /
• v.13 no.11
• /
• pp.965-985
• /
• 2004

The investigation of driving mechanism for the formation of tropical night in the coastal region, defined as persistent high air temperature over than 25$^{\circ}C$ at night was carried out from August 14 through 15, 1995. Convective boundary layer (CBL) of a 1 km depth with big turbulent vertical diffusion coefficients is developed over the ground surface of the inland basin in the west of the mountain and near the top of the mountain, while a depth of thermal internal boundary layer (TIBL) like CBL shrunken by relatively cool sea breeze starting at 100 km off the eastern sea is less than 150 m from the coast along the eastern slope of the mountain. The TIBL extends up to the height of 1500 m parallel to upslope wind combined with valley wind and easterly sea breeze from the sea. As sensible heat flux convergences between the surface and lower atmosphere both at the top of mountain and the inland coast are much greater than on the coastal sea, sensible heat flux should be accumulated inside both the TIBL and the CBL near the mountain top and then, accumulated sensible heat flux under the influence of sea breeze circulation combined with easterly sea breeze from sea to inland and uplifted valley wind from inland to the mountain top returning down toward the eastern coastal sea surface should be transported into the coast, resulting in high air temperatures near the coastal inland. Under nighttime cooling of ground surface after sunset, mountain wind causes the daytime existed westerly wind to be an intensified westerly downslope wind and land breeze further induces it to be strong offshore wind. No sensible heat flux divergence or very small flux divergence occurs in the coast, but the flux divergences are much greater on the top of the mountain and along its eastern slope than on the coastal inland and sea surfaces. Thus, less cooling down of the coastal surface than the mountain surface and sensible heat transfer from warm pool over the coast into the coastal surface produce nocturnal high air temperature on the coastal inland surfaces, which is not much changed from daytime ones, resulting in the persistence of tropical night (nocturnal thermal high) until the early in the morning.

• Fisheries and Aquatic Sciences
• /
• v.22 no.12
• /
• pp.30.1-30.11
• /
• 2019

Marine natural products isolated from the sponge Fascaplysinopsis cf reticulata, in French Polynesia, were investigated as an alternative to antibiotics to control pathogens in aquaculture. The overuse of antibiotics in aquaculture is largely considered to be an environmental pollution, because it supports the transfer of antibiotic resistance genes within the aquatic environment. One environmentally friendly alternative to antibiotics is the use of quorum sensing inhibitors (QSIs). Quorum sensing (QS) is a regulatory mechanism in bacteria which control virulence factors through the secretion of autoinducers (AIs), such as acyl-homoserine lactone (AHL) in gram-negative bacteria. Vibrio harveyi QS is controlled through three parallel pathways: HAI-1, AI-2, and CAI-1. Bioassay-guided purification of F. cf reticulata extract was conducted on two bacterial species, i.e., Tenacibaculum maritimum and V. harveyi for antibiotic and QS inhibition bioactivities. Toxicity bioassay of fractions was also evaluated on the freshwater fish Poecilia reticulata and the marine fish Acanthurus triostegus. Cyclohexanic and dichloromethane fractions of F. cf reticulata exhibited QS inhibition on V. harveyi and antibiotic bioactivities on V. harveyi and T. maritimum, respectively. Palauolide (1) and fascaplysin (2) were purified as major molecules from the cyclohexanic and dichloromethane fractions, respectively. Palauolide inhibited QS of V. harveyi through HAI-1 QS pathway at 50 ㎍ ml^-1 (26 μM), while fascaplysin affected the bacterial growth of V. harveyi (50 ㎍ ml^-1) and T. maritimum (0.25 ㎍). The toxicity of fascaplysin-enriched fraction (FEF) was evaluated and exhibited a toxic effect against fish at 50 ㎍ ml^-1. This study demonstrated for the first time the QSI potential of palauolide (1). Future research may assess the toxicity of both the cyclohexanic fraction of the sponge and palauolide (1) on fish, to confirm their potential as alternative to antibiotics in fish farming.

• Journal of Biomedical Engineering Research
• /
• v.39 no.5
• /
• pp.220-228
• /
• 2018

The internal resistance of microbial fuel cell (MFC) using stainless steel skein for oxidizing electrode was investigated and the factors affecting the voltage generation were identified. We also investigated the effect of power management system (PMS) on the usability for MFC and the removal efficiency of organic pollutants. The performance of a stack microbial fuel cell connected with (PMS) or PMS+LED was analyzed by the voltage generation and organic matter reduction. The maximum power density of the unit cells was found to be $5.82W/m^3$ at $200{\Omega}$. The maximum current density was $47.53A/m^3$ without power overshoot even under $1{\Omega}$. The ohmic resistance ($R_s$) and the charge transfer resistance ($R_{ct}$) of the oxidation electrode using stainless steel skein electrode, were $0.56{\Omega}$ and $0.02{\Omega}$, respectively. However, the sum of internal resistance for reduction electrode using graphite felts loaded Pt/C catalyst was $6.64{\Omega}$. Also, in order to understand the internal resistance, the current interruption method was used by changing the external resistance as $50{\Omega}$, $300{\Omega}$, $5k{\Omega}$. It has been shown that the ohm resistance ($R_s$) decreased with the external resistance. In the case of a series-connected microbial fuel cell, the reversal phenomenon occurred even though two cells having the similar performance. However, the output of the PMS constantly remained for 20 hours even when voltage reversal occurred. Also the removal ability of organic pollutants (SCOD) was not reduced. As a result of this study, it was found that buffering effect for a certain period of time when the voltage reversal occurred during the operation of the microbial fuel cell did not have a serious effect on the energy loss or the operation of the microbial fuel cell.