• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.175-197
• /
• 2021

Recently, with the development of deep learning technology, research on unstructured data analysis is being actively conducted, and it is showing remarkable results in various fields such as classification, summary, and generation. Among various text analysis fields, text classification is the most widely used technology in academia and industry. Text classification includes binary class classification with one label among two classes, multi-class classification with one label among several classes, and multi-label classification with multiple labels among several classes. In particular, multi-label classification requires a different training method from binary class classification and multi-class classification because of the characteristic of having multiple labels. In addition, since the number of labels to be predicted increases as the number of labels and classes increases, there is a limitation in that performance improvement is difficult due to an increase in prediction difficulty. To overcome these limitations, (i) compressing the initially given high-dimensional label space into a low-dimensional latent label space, (ii) after performing training to predict the compressed label, (iii) restoring the predicted label to the high-dimensional original label space, research on label embedding is being actively conducted. Typical label embedding techniques include Principal Label Space Transformation (PLST), Multi-Label Classification via Boolean Matrix Decomposition (MLC-BMaD), and Bayesian Multi-Label Compressed Sensing (BML-CS). However, since these techniques consider only the linear relationship between labels or compress the labels by random transformation, it is difficult to understand the non-linear relationship between labels, so there is a limitation in that it is not possible to create a latent label space sufficiently containing the information of the original label. Recently, there have been increasing attempts to improve performance by applying deep learning technology to label embedding. Label embedding using an autoencoder, a deep learning model that is effective for data compression and restoration, is representative. However, the traditional autoencoder-based label embedding has a limitation in that a large amount of information loss occurs when compressing a high-dimensional label space having a myriad of classes into a low-dimensional latent label space. This can be found in the gradient loss problem that occurs in the backpropagation process of learning. To solve this problem, skip connection was devised, and by adding the input of the layer to the output to prevent gradient loss during backpropagation, efficient learning is possible even when the layer is deep. Skip connection is mainly used for image feature extraction in convolutional neural networks, but studies using skip connection in autoencoder or label embedding process are still lacking. Therefore, in this study, we propose an autoencoder-based label embedding methodology in which skip connections are added to each of the encoder and decoder to form a low-dimensional latent label space that reflects the information of the high-dimensional label space well. In addition, the proposed methodology was applied to actual paper keywords to derive the high-dimensional keyword label space and the low-dimensional latent label space. Using this, we conducted an experiment to predict the compressed keyword vector existing in the latent label space from the paper abstract and to evaluate the multi-label classification by restoring the predicted keyword vector back to the original label space. As a result, the accuracy, precision, recall, and F1 score used as performance indicators showed far superior performance in multi-label classification based on the proposed methodology compared to traditional multi-label classification methods. This can be seen that the low-dimensional latent label space derived through the proposed methodology well reflected the information of the high-dimensional label space, which ultimately led to the improvement of the performance of the multi-label classification itself. In addition, the utility of the proposed methodology was identified by comparing the performance of the proposed methodology according to the domain characteristics and the number of dimensions of the latent label space.

• Korean Journal of Remote Sensing
• /
• v.37 no.3
• /
• pp.379-394
• /
• 2021

Air pollution data in South Korea is provided on a real-time basis by Air Korea stations since 2005. Previous studies have shown the feasibility of gridding air pollution data, but they were confined to a few cities. This paper examines the creation of nationwide gridded maps for PM10 concentration using 333 Air Korea stations with variogram optimization and ordinary kriging. The accuracy of the spatial interpolation was evaluated by various sampling schemes to avoid a too dense or too sparse distribution of the validation points. Using the 114,745 matchups, a four-round blind test was conducted by extracting random validation points for every 365 days in 2019. The overall accuracy was stably high with the MAE of 5.697 ㎍/m³ and the CC of 0.947. Approximately 1,500 cases for high PM10 concentration also showed a result with the MAE of about 12 ㎍/m³ and the CC over 0.87, which means that the proposed method was effective and applicable to various situations. The gridded maps for daily PM10 concentration at the resolution of 0.05° also showed a reasonable spatial distribution, which can be used as an input variable for a gridded prediction of tomorrow's PM10 concentration.

• Korean Journal of Remote Sensing
• /
• v.37 no.3
• /
• pp.531-542
• /
• 2021

Open-pit mines require constant monitoring as they can cause surface changes and environmental disturbances. In open-pit mines, there is little vegetation at the mining site and can be monitored using InSAR (Interferometric Synthetic Aperture Radar) coherence imageries. In this study, activities occurring in mine were analyzed by applying the recently developed InSAR coherence-based NDAI (Normalized Difference Activity Index). The March 5 Youth Mine is a North Korean mine whose development has been expanded since 2008. NDAI analysis was performed with InSAR coherence imageries obtained using Sentinel-1 SAR images taken at 12-day intervals in the March 5 Youth Mine. First, the area where the elevation decreased by about 75.24 m and increased by about 9.85 m over the 14 years from 2000 was defined as the mining site and the tailings piles. Then, the NDAI images were used for time series analysis at various time intervals. Over the entire period (2017-2019), average mining activity was relatively active at the center of the mining area. In order to find out more detailed changes in the surface activity of the mine, the time interval was reduced and the activity was observed over a 1-year period. In 2017, we analyzed changes in mining operations before and after artificial earthquakes based on seismic data and NDAI images. After the large-scale blasting that occurred on 30 April 2017, activity was detected west of the mining area. It is estimated that the size of the mining area was enlarged by two blasts on 30 September 2017. The time-averaged NDAI images used to perform detailed time-series analysis were generated over a period of 1 year and 4 months, and then composited into RGB images. Annual analysis of activity confirmed an active region in the northeast of the mining area in 2018 and found the characteristic activity of the expansion of tailings piles in 2019. Time series analysis using NDAI was able to detect random surface changes in open-pit mines that are difficult to identify with optical images. Especially in areas where in situ data is not available, remote sensing can effectively perform mining activity analysis.

• Sleep Medicine and Psychophysiology
• /
• v.28 no.1
• /
• pp.27-33
• /
• 2021

Objectives: The purpose of this study is to investigate the difference in the effects of eye-mask and light-off on sleep status according to a commercial fitness tracker and a sleep diary of psychiatric in-patients in correctional facilities where nocturnal light is compulsory. Methods: This study was conducted over 3 consecutive nights. In-patients of the National Forensic Psychiatric Hospital (n = 29) were assigned random subject numbers and slept as usual in the light-on condition on the first night. The subjects slept with eye-masks in the light-on condition on another night and without an eye-mask in the light-off condition on the other night. Subjects were asked to sleep wearing a commercial fitness tracker and to keep a sleep diary. The order of these changes in bedroom lighting condition on the second and third nights was assigned randomly to participants. Results: In comparison of the sleep variables between the light-on condition and the eye-mask condition, the Wakefullness After Sleep Onset (WASO) was shorter and sleep satisfaction was higher in the latter.(respectively, Z = 3.66, p < 0.017 ; Z = 2.69, p < 0.017) In comparison of the sleep variables between the light-on and light-off conditions, the WASO was shorter and sleep efficiency and sleep satisfaction were higher in the latter (respectively, Z = 2.40, p < 0.017 ; Z = 3.02, p < 0.017 ; Z = 3.88, p < 0.017). However, there were no differences in the sleep variables between the eye-mask condition and the light-off condition. Conclusion: Subjective improvements in sleep variables were noted in sleep diaries of institutionalized psychiatric patients under either the 'eye-mask' or 'light-off' condition. However, there were no significant differences between the 'eye-mask' and 'light-off' conditions. Therefore, we suggest that psychiatric patients in correctional facilities use eye-masks when sleeping.

• Journal of Internet Computing and Services
• /
• v.21 no.6
• /
• pp.23-31
• /
• 2020

This study was carried out to generate various images of railroad surfaces with random defects as training data to be better at the detection of defects. Defects on the surface of railroads are caused by various factors such as friction between track binding devices and adjacent tracks and can cause accidents such as broken rails, so railroad maintenance for defects is necessary. Therefore, various researches on defect detection and inspection using image processing or machine learning on railway surface images have been conducted to automate railroad inspection and to reduce railroad maintenance costs. In general, the performance of the image processing analysis method and machine learning technology is affected by the quantity and quality of data. For this reason, some researches require specific devices or vehicles to acquire images of the track surface at regular intervals to obtain a database of various railway surface images. On the contrary, in this study, in order to reduce and improve the operating cost of image acquisition, we constructed the 'Defective Railroad Surface Regeneration Model' by applying the methods presented in the related studies of the Generative Adversarial Network (GAN). Thus, we aimed to detect defects on railroad surface even without a dedicated database. This constructed model is designed to learn to generate the railroad surface combining the different railroad surface textures and the original surface, considering the ground truth of the railroad defects. The generated images of the railroad surface were used as training data in defect detection network, which is based on Fully Convolutional Network (FCN). To validate its performance, we clustered and divided the railroad data into three subsets, one subset as original railroad texture images and the remaining two subsets as another railroad surface texture images. In the first experiment, we used only original texture images for training sets in the defect detection model. And in the second experiment, we trained the generated images that were generated by combining the original images with a few railroad textures of the other images. Each defect detection model was evaluated in terms of 'intersection of union(IoU)' and F1-score measures with ground truths. As a result, the scores increased by about 10~15% when the generated images were used, compared to the case that only the original images were used. This proves that it is possible to detect defects by using the existing data and a few different texture images, even for the railroad surface images in which dedicated training database is not constructed.

• Korean Journal of Optics and Photonics
• /
• v.33 no.4
• /
• pp.167-176
• /
• 2022

Adaptive optics (AO) systems compensate for atmospheric disturbance, especially phase distortion, by introducing counter-wavefront deformation calculated from real-time wavefront sensing or prediction. Because AO system implementations are time-consuming and costly, it is highly desirable to estimate the system's performance during the development of the AO system or its parts. Among several techniques, we mostly apply statistical analysis, computational simulation, and optical-bench tests. Statistical analysis estimates performance based on the sum of performance variances due to all design parameters, but ignores any correlation between them. Computational simulation models every part of an adaptive optics system, including atmospheric disturbance and a closed loop between wavefront sensor and deformable mirror, as close as possible to reality, but there are still some differences between simulation models and reality. The optical-bench test implements an almost identical AO system on an optical bench, to confirm the predictions of the previous methods. We are currently developing an AO system for a 1.6-m ground telescope using a deformable mirror that was recently developed in South Korea. This paper reports the results of the statistical analysis and computer simulation for the system's design and confirmation. For the analysis, we apply the Strehl ratio as the performance criterion, and the median seeing conditions at the Bohyun observatory in Korea. The statistical analysis predicts a Strehl ratio of 0.31. The simulation method similarly reports a slightly larger value of 0.32. During the study, the simulation method exhibits run-to-run variation due to the random nature of atmospheric disturbance, which converges when the simulation time is longer than 0.9 seconds, i.e., approximately 240 times the critical time constant of the applied atmospheric disturbance.

• The Journal of Korean Society for Radiation Therapy
• /
• v.33
• /
• pp.89-97
• /
• 2021

Purpose: The purpose of this study is to compare and evaluate the dose change according to the gas volume variations in the rectum, which was not included in the treatment plan during radiation therapy for cervical cancer. Materials and methods: Static Intensity Modulated Radiation Therapy (S-IMRT) using a 9-field and Volumetric Modulated Arc Therapy (VMAT) using 2 full-arcs were established with treatment planning system on Computed Tomography images of a human phantom. Random gas parameters were included in the Planning Target Volume(PTV) with a maximum change of 2.0 cm in increments of 0.5 cm. Then, the Conformity Index (CI), Homogeneity Index (HI) and PTV D_max for the target volume were calculated, and the minimum dose (D_min), mean dose (D_mean) and Maximum Dose (D_max) were calculated and compared for OAR(organs at risk). For statistical analysis, T-test was performed to obtain a p-value, where the significance level was set to 0.05. Result: The HI coefficients of determination(R²) of S-IMRT and VMAT were 0.9423 and 0.8223, respectively, indicating a relatively clear correlation, and PTV D_max was found to increase up to 2.8% as the volume of a given gas parameter increased. In case of OAR evaluation, the dose in the bladder did not change with gas volume while a significant dose difference of more than D_mean 700 cGy was confirmed in rectum using both treatment plans at gas volumes of 1.0 cm or more. In all values except for D_mean of bladder, p-value was less than 0.05, confirming a statistically significant difference. Conclusion: In the case of gas generation not considered in the reference treatment plan, as the amount of gas increased, the dose difference at PTV and the dose delivered to the rectum increased. Therefore, during radiation therapy, it is necessary to make efforts to minimize the dose transmission error caused by a large amount of gas volumes in the rectum. Further studies will be necessary to evaluate dose transmission by not only varying the gas volume but also where the gas was located in the treatment field.

• The Journal of Korean Society for Radiation Therapy
• /
• v.33
• /
• pp.9-14
• /
• 2021

This study examined dose change depending on the reposition error of the junction at the time of treatment with multi-isocenter volumetric modulated arc therapy. This study selected a random treatment region in the Arccheck Phantom and established the treatment plan for multi-isocenter volumetric modulated arc therapy. Then, after setting the error of the junction at 0 ~ 4 mm in the X (left), Y (upper), and Z (inner and outer) directions, the area was irradiated using a linear accelerator; the point doses and gamma indexes obtained through the Phantom were subsequently analyzed. It was found that when errors of 2 and 4 mm took place in the X and Y directions, the gamma pass rates (point doses) were 99.3% (2.085) and 98% (2.079 Gy) in the former direction and 98.5% (2.088) and 95.5% (2.093 Gy) in the latter direction, respectively. In addition, when errors of 1, 2, and 4 mm occurred in the inner and outer parts of the Z direction, the gamma pass rates (point doses) were found to be 94.8% (2.131), 82.6% (2.164), and 72.8% (2.22 Gy) in the former part and 93.4% (2.069), 90.6% (2.047), and 79.7% (1.962 Gy) in the latter part, respectively. In the X and Y directions, errors up to 4 mm were tolerable; however, in the Z direction, error values exceeding 1 mm were beyond the tolerance level. This suggests that for high and low dose areas, errors in the direction same as the progress direction in the treatment region have a more sensitive dose distribution. If the guidelines for set-up errors are established at the institutional level through continuous research in the future, it will be possible to provide good quality treatment using junctions.

• Journal of Bio-Environment Control
• /
• v.31 no.4
• /
• pp.343-355
• /
• 2022

This study investigated the consumption tendency and awareness of paprika in order to expand and promote the consumption of Capsicum annuum L. The research investigated the relationship of preference and loyalty based on emotional response of paprika according to the semantic differential scale. The survey was conducted from January to February 2022 using a random sampling method targeting 155 general people, and a total of 142 questionnaires were analyzed excluding 13 wrong answers. The nine items on the awareness of paprika showed to be consisted of three factors such as 'Food taste', 'Usability', and 'Economics' by factor analysis. Regarding to the awareness of paprika the positive answer that 'I think paprika is good for health' among the nine questions was the highest at 92.3%. In the preference aspect of shape, blocky type had the highest preference for the shape of paprika, followed by mini and conical types in order of preference (p < 0.001). As for color preference, yellow paprika was the most preferred, followed by orange, red, and green, showing statistical significance. The emotional response of paprika by paprika image showed a statistically significant difference in the four colors. The words such as 'bright', 'clean', and 'spirited' appeared as representative emotional vocabulary for paprika. Multiple regression analysis was performed to examine the effect of paprika on the three factors of awareness, preference, and loyalty due to the quality of life. As a result, the higher the paprika preference and quality of life, and the higher the taste and availability factors, the higher the paprika awareness and loyalty. As the variable that has the most influence on the loyalty of the survey respondents, preference was found to have the highest explanatory power at 43%. From these results, it was judged as a very important factor in the survey on the shape and color preference of paprika. Therefore, the recent increase in awareness that paprika is good for health is thought to act as a positive factor in revitalizing the domestic market and increasing consumption of paprika in the future. Also, among the three types of paprika, the yellow blunt type showed the highest preference. Therefore, in order to produce and promote this type of paprika, it is also important to increase the cultivation to suit the purchasing propensity of consumers.

• Korean Journal of Remote Sensing
• /
• v.38 no.6_1
• /
• pp.1371-1383
• /
• 2022

In this study, the amount of coal waste dump was calculated using six Digital Elevation Models (DEMs) produced between 2006 and 2018 in Jangseong-dong, Taebaek-si, Gangwon-do, and the subsidence was observed by applying the Persistent Scatterer Interferometric SAR (PSInSAR) technique on the Sentinel-1 SAR images. As a result of depositing activities using DEMs, a total of 1,668,980 m³ of coal waste was deposited over a period of about 12 years from 2006 to 2018. The observed subsidence rate from PSInSAR was -32.3 mm/yr and -40.2 mm/yr from the ascending and descending orbits, respectively. As the thickness of the waste pile increased, the rate of subsidence increased, and the more recent the completion of the deposit, the faster the subsidence tended to occur. The subsidence rates from the ascending and descending orbits were converted to vertical and horizontal east-west components, and 22 random reference points were set to compare the subsidence rate, the waste rock thickness, and the time of depositing completion. As a result, the subsidence rate of the reference point tended to increase as the thickness of the waste became thicker, similar to the PSInSAR results in relation to the waste thickness. On the other hand, there was no clear correlation between the completion time of the deposits and the rate Of subsidence at the reference points. This is because the time of completion of the deposits at all but 5 of the 22 reference points was too biased in 2010 and the correlation analysis was meaningless. As in this study, the use of DEM and PSInSAR is expected to be an effective alternative to compensate for the lack of field data in the safety management of coal waste deposits.