• Journal of Korean Society on Water Environment
• /
• v.37 no.4
• /
• pp.275-285
• /
• 2021

Algal bloom is an important issue in maintaining the safety of the drinking water supply system. Fast detection and classification of algae images are essential for the management of algal blooms. Conventional visual identification using a microscope is a labor-intensive and time-consuming method that often requires several hours to several days in order to obtain analysis results from field water samples. In recent decades, various deep learning algorithms have been developed and widely used in object detection studies. YOLO is a state-of-the-art deep learning algorithm. In this study the third version of the YOLO algorithm, namely, YOLOv3, was used to develop an algae image detection model. YOLOv3 is one of the most representative one-stage object detection algorithms with faster inference time, which is an important benefit of YOLO. A total of 1,114 algae images for 30 genera collected by microscope were used to develop the YOLOv3 algae image detection model. The algae images were divided into four groups with five, 10, 20, and 30 genera for training and testing the model. The mean average precision (mAP) was 81, 70, 52, and 41 for data sets with five, 10, 20, and 30 genera, respectively. The precision was higher than 0.8 for all four image groups. These results show the practical applicability of the deep learning algorithm, YOLOv3, for algae image detection.

• The Korean Journal of Archival Studies
• /
• no.10
• /
• pp.324-370
• /
• 2004

This paper aims to research the historical development of the archival systems in the Soviet Union from 1917 to the 1920's. The Russian revolution was the turning point not only for the political and social changes but for the archival administration in Russia, as it provided the new Soviet regime with the chances to reorganize the archival institutions under Bolshevik rule. However, we must not forget the fact that the Russian Archival Workers' Union had taken part in the organization of the archival systems together with the Bolshevik revolutionaries. The Soviet government intended to transform the decentralized and poorly organized archival systems in the prerevolutionary years into the centralized and bureaucratized ones. In this meaning, the decree signed be V. I. Lenin on 1 June 1918 was estimated as the real basis for the Soviet archival centralization. Lenin's archival decree of 1918 encompassed the entire national documentary legacy, which was gradually extended to all types and categories of manuscripts and archival records by successive legislation. The concept of state proprietorship of all documentary records was designated "Single State Archival Fond"(Edinyi Gosudarstvennyi arkhivnyi fond), which was renamed as "Archival Fond of the Russian Federation" after the collapse of the USSR. The independent state administrative agency, that is, "Main Administration of Archival Affairs"(Glavnoe upravlenie arkhivnym delom) was charged with the management of the entire "Single State Archival Fond". While the Soviet Union reorganized its archival systems in the twenties, the archival institutions became under the severe control of the government. For example, M. N. Pokrovskii, the well-known Marxist historian and the political leader, headed the organizational work of archives in the Soviet Union, which resulted in the exclusion of the prerevolutionary specialists from the archival field in the Soviet Union. However, the discussions over the concept of "the Archival Fond" by B. I. Anfilov helped to develop the archival theories in the USSR in the twenties. In conclusion, the Soviet Union, having emphasized the centralization of the archives, developed its archival systems from the early period, which were the basis of the systematic archival institutions in Russia.

• Journal of Environmental Health Sciences
• /
• v.45 no.2
• /
• pp.173-185
• /
• 2019

Objectives: The objectives of this study are to evaluate the accuracy and precision of exposure model ECETOC TRA v.3.1 by comparing model predictions with repeated exposure measurements in Korean workplaces and to investigate the applicability of ECETOC TRA to Korean workplace exposure assessment in K-REACH. Methods: Measured values and work conditions for 14 kinds of chemicals collected from exposure field surveys conducted at 10 companies in Korea were utilized for this study. All possible process categories (PROCs) considered to be relevant to each work process classification were selected and applied to ECETOC TRA as major determining parameters. In order to quantify the accuracy of the model, the lack of agreement (bias, relative bias, precision) was calculated and the risk ratios for each exposure situation between estimated and measured were also compared. Results: The estimated values varied between five and 25 times according to the PROCs for all exposure situations (ESs) based on tasks/chemicals. The results showed that most of the estimated values were below the measured values, and just 13 of 53 tasks were above the measured values. The overall bias and precision were $-2.91{\pm}1.62$ with ECETOC TRA, and we found that ECETOC TRA showed a low level of conservatism when applied to Korean workplaces, similar to previous studies. Conclusions: This study demonstrates that the existed PROC codes have limitations in fully covering various ESs in Korea. In order to improve the applicability of ECETOC TRA in K-REACH, the addition of new PROCs for Korean industries are necessary.

• Journal of Cadastre & Land InformatiX
• /
• v.50 no.2
• /
• pp.37-51
• /
• 2020

With the development of machine learning and deep learning technologies, there has been increasing interest and attempt to apply these technologies to the detection of urban changes. However, the traditional methods of detecting changes and constructing spatial information are still often performed manually by humans, which is costly and time-consuming. Besides, a large number of people are needed to efficiently detect changes in buildings in urban areas. Therefore, in this study, a methodology that can detect changes by classifying road, building, and vegetation objects that are highly utilized in the geospatial information field was proposed by applying deep learning technology to point clouds. As a result of the experiment, roads, buildings, and vegetation were classified with an accuracy of 92% or more, and attributes information of the objects could be automatically constructed through this. In addition, if time-series data is constructed, it is thought that changes can be detected and attributes of existing digital maps can be inspected through the proposed methodology.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.1
• /
• pp.57-64
• /
• 2020

Recently, due to the proliferation of IoT sensors, the development of big data and artificial intelligence, time series prediction research on fine dust pollution is actively conducted. However, because the data representing fine dust contamination changes rapidly, traditional time series prediction methods do not provide a level of accuracy that can be used in the field. In this paper, we propose a method that reflects the classification results of environmental conditions through CNN when predicting micro dust contamination using LSTM. Although LSTM and CNN are independent, they are integrated into one network through the interface, so this method is easier to understand than the application LSTM. In the verification experiments of the proposed method using Beijing PM2.5 data, the prediction accuracy and predictive power for the timing of change were consistently improved in various experimental cases.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.4
• /
• pp.57-62
• /
• 2021

As the parallel computing capability for artificial intelligence improves, the field of artificial intelligence technology is expanding in various industries. In particular, artificial intelligence is being introduced to process data generated from IoT sensors that have enoumous data. However, the limitation exists when applying the AI techniques on IoT network because IoT has time series data, where the importance of data changes over time. In this paper, we propose time-weighted and user-state based artificial intelligence processing techniques to effectively process IoT sensor data. This technique aims to effectively classify IoT sensor data through a data pre-processing process that personalizes time series data and places a weight on the time series data before artificial intelligence learning and use status of personal data. Based on the research, it is possible to propose a method of applying artificial intelligence learning in various fields.

• Korean Chemical Engineering Research
• /
• v.60 no.2
• /
• pp.223-228
• /
• 2022

In chemical processes, unintended faults can make serious accidents. To tackle them, proper fault diagnosis models should be designed to identify the root cause of faults. To design a fault diagnosis model, a process and its data should be analyzed. However, most previous researches in the field of fault diagnosis just handle the data set of benchmark processes simulated on commercial programs. It indicates that it is really hard to get fresh data sets on real processes. In this study, real faulty conditions of an industrial polystyrene process are tested. In this process, a runaway reaction occurred and this caused a large loss since operators were late aware of the occurrence of this accident. To design a proper fault diagnosis model, we analyzed this process and a real accident data set. At first, a mode classification model based on support vector machine (SVM) was trained and principal component analysis (PCA) model for each mode was constructed under normal operation conditions. The results show that a proposed model can quickly diagnose the occurrence of a fault and they indicate that this model is able to reduce the potential loss.

• Journal of the Korean Geotechnical Society
• /
• v.38 no.10
• /
• pp.17-29
• /
• 2022

Recently, the interest in integrated underground geospatial information mapping (IUGIM) to ensure the safety of underground spaces and facilities has been increasing. Because IUGIM is used in the fields of underground space development and underground safety management, the up-to-dateness and accuracy of information are critical. In this study, IUGIM and field data were classified, and the accuracy of IUGIM was improved by spatial interpolation. A spatial interpolation technique was used to process borehole data in IUGIM, and a quantitative evaluation was performed with mean absolute error and root mean square error through the cross-validation of seven interpolation results according to the technique and model. From the cross-validation results, accuracy decreased in the order of nonuniform rational B-spline, Kriging, and inverse distance weighting. In the case of Kriging, the accuracy difference according to the variogram model was insignificant, and Kriging using the spherical variogram exhibited the best accuracy.

• Journal of the Society of Disaster Information
• /
• v.18 no.3
• /
• pp.495-506
• /
• 2022

Purpose: When a serious disaster occurred in East Japan on March 11, 2011, some evacuees in shock failed to avoid danger to the best of their ability. Why did they hesitate and waste their time? And why didn't they choose correct escaping routes? This study attempts to classify human behavior through psychological point of view and cognitive science and to interpret behavioral patterns based on animal behaviors from the field of biology. Method: This study first conceptually categorized walking behavior into intellectualization, automaticity and instinct based on the existing literature and matched these with empirical data. Result: The actual walking patterns observed failed to be compatible with these categories and consequently, this study suggests the following five categories: normal, busy, fast & straight, freezing and tizzy. This new classification of walking behavior is based on speed, variation of speed and change of direction. Conclusion: The method used in this study and the results can be applied to simulations of walking behavior and analysis of behavior in emergency situations.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.23 no.1
• /
• pp.7-13
• /
• 2023

This study investigated the criteria for evaluating the impact of artificial intelligence services. The study classified AI evaluation targets into two areas: AI service and AI technology, and identified influence, sustainability, efficiency, effectiveness, and appropriateness as potential evaluation criteria. The time aspect of AI service evaluation was divided into pre-evaluation and post-evaluation, with pre-evaluation focused on reviewing items during development and design. The AI service area was classified into public, private, and mixed forms, and the impact assessment was classified as vertical or horizontal. The application of AI services was divided into normative and regulatory aspects, and the purpose of the evaluation could be impact or process evaluation. The subject and field of the AI service could also be used for classification purposes. The results of this study can be used to support the creation of AI service impact policies and countermeasures. However, further research is needed to develop specific indicators based on the criteria identified in this study to evaluate the impact of AI services.