• The Journal of the Convergence on Culture Technology
• /
• v.9 no.4
• /
• pp.487-494
• /
• 2023

In this study, performance comparison was performed on R and TensorFlow, which are free deep learning tools. In the experiment, six types of deep neural networks were built using each tool, and the neural networks were trained using the 10-year Korean temperature dataset. The number of nodes in the input layer of the constructed neural network was set to 10, the number of output layers was set to 5, and the hidden layer was set to 5, 10, and 20 to conduct experiments. The dataset includes 3600 temperature data collected from Gangnam-gu, Seoul from March 1, 2013 to March 29, 2023. For performance comparison, the future temperature was predicted for 5 days using the trained neural network, and the root mean square error (RMSE) value was measured using the predicted value and the actual value. Experiment results shows that when there was one hidden layer, the learning error of R was 0.04731176, and TensorFlow was measured at 0.06677193, and when there were two hidden layers, R was measured at 0.04782134 and TensorFlow was measured at 0.05799060. Overall, R was measured to have better performance. We tried to solve the difficulties in tool selection by providing quantitative performance information on the two tools to users who are new to machine learning.

• Korean Journal of Remote Sensing
• /
• v.38 no.6_1
• /
• pp.1109-1124
• /
• 2022

This study proposes a marine debris monitoring methods using satellite and drone multispectral images. A multi-layer perceptron (MLP) model was applied to detect marine debris using Sentinel-2 satellite image. And for the detection of marine debris using drone multispectral images, performance evaluation and comparison of U-Net, DeepLabv3+ (ResNet50) and DeepLabv3+ (Inceptionv3) among deep learning models were performed (mIoU 0.68). As a result of marine debris detection using satellite image, the F1-Score was 0.97. Marine debris detection using drone multispectral images was performed on vegetative debris and plastics. As a result of detection, when DeepLabv3+ (Inceptionv3) was used, the most model accuracy, mean intersection over union (mIoU), was 0.68. Vegetative debris showed an F1-Score of 0.93 and IoU of 0.86, while plastics showed low performance with an F1-Score of 0.5 and IoU of 0.33. However, the F1-Score of the spectral index applied to generate plastic mask images was 0.81, which was higher than the plastics detection performance of DeepLabv3+ (Inceptionv3), and it was confirmed that plastics monitoring using the spectral index was possible. The marine debris monitoring technique proposed in this study can be used to establish a plan for marine debris collection and treatment as well as to provide quantitative data on marine debris generation.

• Environmental Analysis Health and Toxicology
• /
• v.22 no.2 s.57
• /
• pp.129-135
• /
• 2007

Polycyclic aromatic hydrocarbons(PAHs) have been analyzed to assess vertical distribution of them with different land uses. The soils were collected from three layers; surface $(0{\sim}5cm)$, intermediate $(6{\sim}10cm)$, and deep $(11{\sim}15cm)$ layer, respectively considering land use; paddy, upland, and mountain in each site. Total 89 samples of soil from 10 sites were analyzed. Overall mean of ${\sum}PAHs$ were 137 (range $8.87{\sim}625{\mu}g\;kg^{-1}$), 203 (range $16.5{\sim}645{\mu}g\;kg^{-1}$), and $83.4{\mu}g\;kg^{-1}$ (range $6.65{\sim}667{\mu}g\;kg^{-1}$) for paddy, upland, and mountain soil, respectively. The dominant PAHs were fluoroanthene/benzo(b)fluoroanthene>pyrene>indeno(1, 2, 3-cd) pyrene in paddy, fluoroanthene/pyrene>benzo(b)fluoroanthene>chrysene in upland, and benzo(b)fluoroanthene>pyrene>chrysene in mountain soil, whereas the profile was quite similar for each other except that indeno(1, 2, 3-cd)pyrene and benzo(ghi)perylene are relatively higher in the paddy soils. Although the concentration gradient by depth was not observed in the paddy and upland soils because perturbation of soil layer by tillage, significant decrease was in the deep layer relative to the surface and intermediate layer. However, the concentration gradient of PAHs by soil depth was clearly shown in mountain soil without experiencing disturbance of tillage.

• 한국해양학회지
• /
• v.12 no.2
• /
• pp.75-81
• /
• 1977

Distribution of drifting larvae of Anadara broughtnoii SCHRENCK was studied based on the planktonic sampling which has been collected in fifteen sampling areas of southern coast of Korea and Ulsan Bay during summer season from 1973 to 1977. Vertical and horizontal occurrence was analyzed related to the environmental factors such as surface water temperature, current velocity and depth of water column. High density of the larvae was observed in the Chinhae Bay which included the sampling areas Rampo, Sockcheon, Majeon, Changpo, Dangdong, Bedun, Changchoa, and Wonmun. Maximum occurrence of the farvae was accompanied with the highest water temperature of the summer season, and it was usually August when the water temperature was over 27$^{\circ}C$. In August, 1975, the highest density of the farvae was observed, when the mean surface water temperature was the highest compared to those of other years. The first appearence of the drifting larvae was also related to the surface water temperature. Each year the larae begin to appear from the late July and the ready-to-fall larvae appear in abundance from the mid-August. Vertical distribution patterns of the larvae are closely related to the depth of the water column as well as to the current velocity. In shallow water the larvae tend to aggregate in the bottom layer, while they are diffused to some extent in deep water. In shallow water column ( 8m) more or less 75% of the total larvae individuals was observed in the lower 4m layer and in deep water column ( 16m) only 45% of those was found in the lower 4m layer. In the water of lower velocity a large fraction of the larvae population is distributed in the lower depth layer.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.6
• /
• pp.301-307
• /
• 2023

Deep learning requires building a deep neural network, collecting a large amount of training data, and then training the built neural network for a long time. If training does not proceed properly or overfitting occurs, training will fail. When using deep learning tools that have been developed so far, it takes a lot of time to collect training data and learn. However, due to the rapid advent of the mobile environment and the increase in sensor data, the demand for real-time deep learning technology that can dramatically reduce the time required for neural network learning is rapidly increasing. In this study, a real-time deep learning system was implemented using an Arduino system equipped with a fine dust sensor. In the implemented system, fine dust data is measured every 30 seconds, and when up to 120 are accumulated, learning is performed using the previously accumulated data and the newly accumulated data as a dataset. The neural network for learning was composed of one input layer, one hidden layer, and one output. To evaluate the performance of the implemented system, learning time and root mean square error (RMSE) were measured. As a result of the experiment, the average learning error was 0.04053796, and the average learning time of one epoch was about 3,447 seconds.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.1 no.2
• /
• pp.73-79
• /
• 1996

This study was conducted to investigate the faunal composition and distribution patterns of macrobenthic community in the deep sea sediments of the KODOS area, the northeastern Pacific Ocean during May 1996. Benthic animals were collected at 25 stations using a spade type box corer. Sediments were sieved through -.3 mm mesh screen. A total of 17 faunal groups in 9 phyla and 363 specimens were identified. Nematoda was the most abundant faunal group which accounted for 30.0% of total abundance. Other dominant faunal groups were foraminiferans (25.1%), harpacticoids (10.2%), xenophyophores (5.2%), and polychaetes (4.7%), Polychaeta was a typically dominant component of macrobenthic community in the study area except traditionally recognized meiofauna taxa. Mean occurrence number of faunal taxa was ca. 6 per 0.01 m$\^$2/, and mean density was estimated as 1,288 indiv./m$\^$2/. The abundance of whole fauna and that of each faunal group was highest at the surface layer of sediment, and decreased monotonously along the sediment depth; 98% of faunal abundance was found within 10 cm depth layer.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.7
• /
• pp.11-18
• /
• 2023

In this paper, we evaluate deep learning time series forecasting models. Recent studies show that those models perform better than the traditional prediction model such as ARIMA. Among them, recurrent neural networks to store previous information in the hidden layer are one of the prediction models. In order to solve the gradient vanishing problem in the network, LSTM is used with small memory inside the recurrent neural network along with BI-LSTM in which the hidden layer is added in the reverse direction of the data flow. In this paper, we compared the performance of Informer by comparing with other models (LSTM, BI-LSTM, and Transformer) for real Nitrogen dioxide (NO2) data. In order to evaluate the accuracy of each method, mean square root error and mean absolute error between the real value and the predicted value were obtained. Consequently, Informer has improved prediction accuracy compared with other methods.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.32 no.10
• /
• pp.791-799
• /
• 2008

We investigate the flow characteristics around a series of rectangular bodies ($40^d{\times}80^w{\times}80^h$, $80^d{\times}80^w{\times}80^h$ and $160^d{\times}80^w{\times}80^h$) placed in a deep turbulent boundary layer. The study is aiming to understand the surface pressure distribution around the bodies such as the suction pressure in the leading edge, when the flow is normal, which is responsible for producing extreme suction pressures on the roof. The experiment includes wind tunnel work by using HWA (Hot-Wire anemometry) and pressure transducers. The experiments are carried out at three different Reynolds numbers, based on the velocity U at the body height h, of $2.4{\times}10^4$, $4.6{\times}10^4$ and $6.7{\times}10^4$, and large enough that the mean flow is effectively Reynolds number independent. The results include the measurements of the growth of the turbulent boundary layer in the wind tunnel and the surface pressure around the bodies.

• Ocean and Polar Research
• /
• v.26 no.2
• /
• pp.311-321
• /
• 2004

Relationships between sedimentary environments and abundance of benthic animals were examined on the deep-sea floor, the Clarion-Clipperton Fracture Zone, in the northeast equatorial Pacific Ocean. Specimens were collected using a box corer at 8 stations by sieving through 0.3 mm mesh screen. Sediments showed finer grain size ranged from 5.63 to $7.97{\varphi}$, 83.1% of mean porosity, 1.81 kPa of mean shear strength and organic carbon content in sediment ranged from 0.97 to $1.87\;mg/cm^3$. Manganese nodules covered on the bottom layer from 4 to 57% of coverages. A total of 26 faunal groups in 6 phyla was sampled and comprised 1,467 individuals. Mean biomass were calibrated to 0.5 gWWt/$0.06\;m^2$. Small-sized animals including foraminiferans and nematods were dominated among the faunal group which comprised 49.1% (892 ind.) and 11.5% (320 ind.), respectively. In SPI-analysis, vertical bio-disturbance marks were not observed except to Beggiatoa-type bacterial mats. As the results of relationship between environments and benthos, abundance of benthic animals, especially nematode, showed only a negative correlation to the coverage of nodules, and any other sedimentary factors analyzed in this study were rarely affected to the spatial distribution of benthic animals.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.30 no.11 s.254
• /
• pp.1425-1432
• /
• 2006

Coating layers on a coated sheet steel frequently affect distributions of strain rate of sheets and deteriorate the frictional characteristics between sheets and tools in sheet metal forming. Thus, it is important to identify the deformation behavior of these coatings to ensure the success of the sheet forming operation. In this study, the technique using nano-indentation test, FE-simulation and Artificial Neural Network(ANN) were proposed to determine the power law stress-strain behavior of coating layer and the power law behavior of extracted coating layers was examined using FE-simulation of drawing and nano-indentation process. Also, deep drawing test was performed to estimate the formability and frictional characteristic of coated sheet, which was calculated using the linear relationship between drawing force and blank holding force obtained from the deep drawing test. FE-simulations of the drawing process were respectively carried out for single-behavior FE-model having one stress-strain behavior and for layer-behavior FE-model which consist of coating and substrate separately. The results of simulations showed that layer-behavior model can predict drawing forces with more accuracy in comparison with single-behavior model. Also, mean friction coefficients used in FE-simulation signify the value that can occur maximum drawing force in a drawing test.