• The Journal of the Korea Contents Association
• /
• v.21 no.11
• /
• pp.135-144
• /
• 2021

Recently, data-driven decision-making technology has become a key technology leading the data industry, and machine learning technology for this requires high-quality training datasets. However, real-world data contains missing values for various reasons, which degrades the performance of prediction models learned from the poor training data. Therefore, in order to build a high-performance model from real-world datasets, many studies on automatically imputing missing values in initial training data have been actively conducted. Many of conventional machine learning-based imputation techniques for handling missing data involve very time-consuming and cumbersome work because they are applied only to numeric type of columns or create individual predictive models for each columns. Therefore, this paper proposes a new data imputation technique called 'Denoising Self-Attention Network (DSAN)', which can be applied to mixed-type dataset containing both numerical and categorical columns. DSAN can learn robust feature expression vectors by combining self-attention and denoising techniques, and can automatically interpolate multiple missing variables in parallel through multi-task learning. To verify the validity of the proposed technique, data imputation experiments has been performed after arbitrarily generating missing values for several mixed-type training data. Then we show the validity of the proposed technique by comparing the performance of the binary classification models trained on imputed data together with the errors between the original and imputed values.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.1
• /
• pp.51-58
• /
• 2023

In this paper, we propose an organ segmentation technique for the automatic extraction of medical diagnostic indicators from X-ray images. In order to calculate diagnostic indicators of heart disease and spinal disease such as VHS(vertebral heart scale) and Cobb angle, it is necessary to accurately segment the thoracic spine, carina, and heart in a chest X-ray image. A deep neural network model in which the high-resolution representation of the image for each layer and the structure converted into a low-resolution feature map are connected in parallel was adopted. This structure enables the relative position information in the image to be effectively reflected in the segmentation process. It is shown that learning performance can be improved by combining the OCR module, in which pixel information and object information are mutually interacted in a multi-step process, and the channel attention module, which allows each channel of the network to be reflected as different weight values. In addition, a method of augmenting learning data is presented in order to provide robust performance against changes in the position, shape, and size of the subject in the X-ray image. The effectiveness of the proposed theory was evaluated through an experiment using 145 human chest X-ray images and 118 animal X-ray images.

• Journal of Korean Library and Information Science Society
• /
• v.53 no.2
• /
• pp.355-375
• /
• 2022

This study is to grasp LRM's feature and applications plan to reflect LRM to cataloging related standards and individual system through comparing and analyzing LRM with the FR model in terms of entities, attributes, and relationships. The application plan is suggested as follows. First, the entity can be extended by defining sub-entities of each entity in the standards and the individual system in order to reflect LRM, even though entities such as families, groups, identifiers, authorized access points, concepts, objects, events, agency and rules have been deleted in LRM. Second, the attribute should be subdivided in the standards and the individual system in order to apply LRM, though many attributes have been changed to relationships for linked data and decreased in LRM. In particular, more specific and detailed property names in the standards and the individual system should be clearly presented, and the vocabulary encoding scheme corresponding to each property should be also developed, since properties with similar functions or repetition in various entities, and material specific properties are generalized and integrated into comprehensive property names. Third, the relationship should be extended through newly declaring the refinement or subtype of the relationship and considering a multi-level relationship, since the relationship itself is general and abstract under increasing the number of relationships in comparing to the property. This study will be practically utilized in cataloging related standards and individual system for applying LRM.

• Animal Bioscience
• /
• v.36 no.3
• /
• pp.451-460
• /
• 2023

Objective: This program aimed to reveal the association of feed intrinsic molecular structure with nutrient supply to animals from canola feedstocks and co-products from bio-oil processing. The special objective of this study was to quantify the relationship between molecular spectral feature and nutrient availability and develop nutrient prediction equation with vibrational molecular structure spectral profiles. Methods: The samples of feedstock (canola oil seeds) and co-products (meals and pellets) from different bio-oil processing plants in Canada (CA) and China (CH) were submitted to this molecular spectroscopic technique and their protein and carbohydrate related molecular spectral features were associated with the nutritional results obtained through the conventional methods of analyses for chemical and nutrient profiles, rumen degradable and intestinal digestible parameters. Results: The results showed that the spectral structural carbohydrates spectral peak area (ca. 1,487.8 to 1,190.8 cm^-1) was the carbohydrate structure that was most significant when related to various carbohydrate parameters of canola meals (p<0.05, r>0.50). And spectral total carbohydrate area (ca. 1,198.5 to 934.3 cm^-1) was most significant when studying the various carbohydrate parameters of canola seeds (p<0.05, r>0.50). The spectral amide structures (ca. 1,721.2 to 1,480.1 cm^-1) were related to a few chemical and nutrient profiles, Cornell Net Carbohydrate and Protein System (CNCPS) fractions, truly absorbable nutrient supply based on the Dutch protein system (DVE/OEB), and NRC systems, and intestinal in vitro protein-related parameters in co-products (canola meals). Besides the spectral amide structures, α-helix height (ca. 1,650.8 to 1,643.1 cm^-1) and β-sheet height (ca. 1,633.4 to 1,625.7 cm^-1), and the ratio between them have shown to be related to many protein-related parameters in feedstock (canola oil seeds). Multi-regression analysis resulted in moderate to high R² values for some protein related equations for feedstock (canola seeds). Protein related equations for canola meals and carbohydrate related equations for canola meals and seeds resulted in weak R² and low p values (p<0.05). Conclusion: In conclusion, the attenuated total reflectance Fourier transform infrared spectroscopy vibrational molecular spectroscopy can be a useful resource to predict carbohydrate and protein-relates nutritional aspects of canola seeds and meals.

• Journal of the Daesoon Academy of Sciences
• /
• v.26
• /
• pp.45-75
• /
• 2016

The aim of this study is to explain characteristics of Maitreya and Maitreya belief from a point of view that 'Jeungsan is the very Maitreya(甑山卽彌勒)'. In 『Jeon-gyeong』, Maitreya is mentioned several times. Thus, new religions of Jeungsan of Daesoonjinrihoe take 'Jeungsan is the very Maitreya' belief for truth. Due to the fact that characteristics of Maitreya are so multi-layered and complicated, it is necessary to explain clearly what kind of feature Maitreya has in 『Jeon-gyeong』. If believing and following 'Jeungsan is the very Maitreya' without clarifying it, they will be faced with a problem that they regard Jeungsan of Supreme being of the Ninth Heaven as one of Maitreya and take its belief for truth. Furthermore, with respect to the characteristics of 'Jeungsan is the very Maitreya' belief, while believing in Mireukasaeng, longed-for Millenarian movement by people through Messianism and Mireukasaeng belief is found in Daesoon Thought, whereas there is a need how to understand the point that we cannot finped Messianism and Millenarian movement in Daesoon Thought. To solve this problem, I draw a conclusion that 'Jeungsan is the very Maitreya' in 『Jeon-gyeong』 has to be understood with two meanings by four demonstrations. First of all, the people perceived late Joseon dynasty as the age of decadence but Maitreya's divinity which is desired by the people is not divinity of Maitreya Sutra(Mileuggyeong). Maitreya's divinity is reflected in the people's cherished desire and it is newly created as the Messiah. Thus, the idea of Jeungsan being the very Maitreya was developed in a way that the people desired the Messiah, encompassing this inclination. That is the Messiah of the people and the divinity of Jeungsan. Although Jeungsan as Supreme being of the Ninth Heaven satisfied the people's desire, it shows a different way to salvation from the way in Maitreya Sutra(Mileuggyeong). It is 'the Great Reordering of the Universe' and 'the Great Reordering of the Three Realms'. Reordering in Jeungsan shows that divinity of Jeungsan is not limited to the people's Messiah. In other words, divinity of Jeungsan is established as The Messiah, surpassing divinity of Maitreya Sutra(Mileuggyeong). And following statements prove this divinity of Jeungsan. Jeungsan's emphasis is not only the people's desire and the Gods' appeal. Jeungsan's emphasis is that only does Supreme being of the Ninth Heaven correct heaven and earth, which is the Gods' appeal. Therefore, 'Jeungsan is the very Maitreya' belief embraces the people's Messianism and at the same time it runs with he Gods' appeal. Thus, Reordering through the Great Reordering of the Universe and the Great Reordering of the Three Realms builds up a new ideal world.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.2
• /
• pp.343-350
• /
• 2022

It is worth verifying the effectiveness of data integration between data with different features. This study investigated whether the data integration affects the accuracy of deep neural network (DNN), and which integration method shows the best improvement. This study used two different public datasets. One public dataset was taken in an actual farm in India. And another was taken in a laboratory environment in Korea. Leaf images were selected from two different public datasets to have five classes which includes normal and four different types of plant diseases. DNN used pre-trained VGG16 as a feature extractor and multi-layer perceptron as a classifier. Data were integrated into three different ways to be used for the training process. DNN was trained in a supervised manner via the integrated data. The trained DNN was evaluated by using a test dataset taken in an actual farm. DNN shows the best accuracy for the test dataset when DNN was first trained by images taken in the laboratory environment and then trained by images taken in the actual farm. The results show that data integration between plant images taken in a different environment helps improve the performance of deep neural networks. And the results also confirmed that independent use of plant images taken in different environments during the training process is more effective in improving the performance of DNN.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.2
• /
• pp.429-436
• /
• 2023

Traditional Chinese music has a deep indigenous color and has its own unique way of thinking and characteristics. A consensus has already been formed that linear thinking is a major feature of traditional Chinese music, and it has been implemented in both traditional multi-tone and single-tone music. It is mainly expressed in the form of single-tone music or single-tone music. This linear thought of traditional Chinese music is formed by influencing factors in various fields. For example, it is related to national culture, geographical and natural environment, religious and philosophical background, traditional Chinese notation, individual characteristics of traditional musical instruments, Yulje, composition, and transmission methods. This thinking is different from Western classical music that pursues three-dimensional thinking, and Western music emphasizes the harmony of harmony, harmony of tone and texture, logic and identity of structure, and emphasizes the aspect of space. However, traditional Chinese music emphasizes the horizontal development of melody, the fluency of ancestors, and the continuity of structure. We aims to analyze the causes of linear thinking of traditional Chinese music so that it can be more useful in educational aspects and promote the succession and development of traditional music by transferring knowledge of ethnic music.

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.33 no.3
• /
• pp.375-383
• /
• 2023

Objectives: Coal-fired power plants feature diverse working conditions, including multi-layered employment structures and irregular work cycles due to outsourcing and non-standardized tasks. The current uniform occupational environment measurement systems have limitations in accurately assessing and evaluating these varied conditions. This study aims to propose alternative measurement and assessment strategies to supplement existing methods. Methods: Major domestic coal-fired power plants were selected as the study targets. To prepare for the study and establish strategies, work processes were identified and existing occupational environment measurement results were compared and analyzed. The study proceeded by employing three strategies: specific exposure groups (SEGs) measurement, continuous monitoring, and supplementary measurements, which were then compared and discussed. Results: Previous exposure index evaluations (5,268 cases) indicated that crystalline silica, a type of respirable particulate matter, had detection limits below the threshold (non-detectable) in 82.6% (4,349 cases) of instances. Exposures below 10% of the exposure limit were observed at a very low concentration of 96.1%. Similar exposure group measurements yielded results where detection limits were below the threshold in 38.2% of cases, and exposures below 10% of the limit were observed in 70.6%. Continuous monitoring indicated detection limits below the threshold in 12.6% of cases, and exposures below 10% of the limit were observed in 75.6%. Instances requiring active workplace management accounted for more than 30% of cases, with SEGs at 11.8% (four cases), showing a higher proportion compared to 3.0% (four cases) in continuous monitoring. For coal dust, exposures below 10% of the limit were highest in legal measurements at 90.2% (113 cases), followed by 74.0% (91 cases) in continuous monitoring, and 47.0% (16 cases) in SEGs. Instances exceeding 30% were most prevalent in SEGs at 14.7% (five cases), followed by legal measurements at 5.0% (eight cases), and continuous monitoring at 2.4% (three cases). When examining exposure levels through arithmetic means, crystalline silica was found to be 104.7% higher in SEGs at 0.0088 mg/m³ compared to 0.0043 mg/m³ in continuous monitoring. Coal dust measurements were highest in SEGs at 0.1247 mg/m³, followed by 0.1224 mg/m³ in legal measurements, and 0.0935 mg/m³ in continuous monitoring. Conclusions: Strategies involving SEGs measurement and continuous monitoring can enhance measurement reliability in environments with irregular work processes and frequent fluctuations in working conditions, as observed in coal-fired power plants. These strategies reduce the likelihood of omitting or underestimating processes and enhance measurement accuracy. In particular, a significant reduction in detection limits below the threshold for crystalline silica was observed. Supplementary measurements can identify worker exposure characteristics, uncover potential risks in blind spots of management, and provide a complementary method for legal measurements.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.28 no.3
• /
• pp.1-9
• /
• 2024

This study proposes a multi-scale template matching technique with image pyramids (TMI) to measure structural dynamic displacement using a vision sensor under atmospheric turbulence conditions and evaluates its displacement measurement performance. To evaluate displacement measurement performance according to distance, the three-story shear structure was designed, and an FHD camera was prepared to measure structural response. The initial measurement distance was set at 10m, and increased with an increment of 10m up to 40m. The atmospheric disturbance was generated using a heating plate under indoor illuminance condition, and the image was distorted by the optical turbulence. Through preliminary experiments, the feasibility of displacement measurement of the feature point-based displacement measurement method and the proposed method during atmospheric disturbances were compared and verified, and the verification results showed a low measurement error rate of the proposed method. As a result of evaluating displacement measurement performance in an atmospheric disturbance environment, there was no significant difference in displacement measurement performance for TMI using an artificial target depending on the presence or absence of atmospheric disturbance. However, when natural targets were used, RMSE increased significantly at shooting distances of 20 m or more, showing the operating limitations of the proposed technique. This indicates that the resolution of the natural target decreases as the shooting distance increases, and image distortion due to atmospheric disturbance causes errors in template image estimation, resulting in a high displacement measurement error.

• Journal of Internet of Things and Convergence
• /
• v.10 no.2
• /
• pp.1-15
• /
• 2024

The construction of smart communities is a new method and important measure to ensure the security of residential areas. In order to solve the problem of low accuracy in face recognition caused by distorting facial features due to monitoring camera angles and other external factors, this paper proposes the following optimization strategies in designing a face recognition network: firstly, a global graph convolution module is designed to encode facial features as graph nodes, and a multi-scale feature enhancement residual module is designed to extract facial keypoint features in conjunction with the global graph convolution module. Secondly, after obtaining facial keypoints, they are constructed as a directed graph structure, and graph attention mechanisms are used to enhance the representation power of graph features. Finally, tensor computations are performed on the graph features of two faces, and the aggregated features are extracted and discriminated by a fully connected layer to determine whether the individuals' identities are the same. Through various experimental tests, the network designed in this paper achieves an AUC index of 85.65% for facial keypoint localization on the 300W public dataset and 88.92% on a self-built dataset. In terms of face recognition accuracy, the proposed network achieves an accuracy of 83.41% on the IBUG public dataset and 96.74% on a self-built dataset. Experimental results demonstrate that the network designed in this paper exhibits high detection and recognition accuracy for faces in surveillance videos.