KJC Korea
Journal Central

In this paper, instead of using recurrent neural network, we compare a classification performance of time series imaging algorithms using convolution neural network. There are traditional algorithms that imaging time series data (e.g. GAF(Gramian Angular Field), MTF(Markov Transition Field), RP(Recurrence Plot)) in TSC(Time Series Classification) community. Furthermore, we compare STFT(Short Time Fourier Transform) algorithm that can acquire spectrogram that visualize feature of voice data. We experiment CNN’s performance by adjusting hyper parameters of imaging algorithms. When evaluate with GunPoint dataset in UCR archive, STFT(Short-Time Fourier transform) has higher accuracy than other algorithms. GAF has 98~99% accuracy either, but there is a disadvantage that size of image is massive.

In this paper, the actual data of VOC reduction facilities was analyzed through a model that detects and predicts data anomalies. Using the USAD model, which shows stable performance in the field of anomaly detection, anomalies in real-time data are detected and sensors that cause anomalies are searched. In addition, we propose a method of predicting and warning, when abnormalities that time will occur by predicting future outliers with an auto-regressive model. The experiment was conducted with the actual data of the VOC reduction facility, and the anomaly detection test results showed high detection rates with precision, recall, and F1-score of 98.54%, 89.08%, and 93.57%, respectively. As a result, averaging of the precision, recall, and F1-score for 8 sensors of detection rates were 99.64%, 99.37%, and 99.63%. In addition, the Hamming loss obtained to confirm the validity of the detection experiment for each sensor was 0.0058, showing stable performance. And the abnormal prediction test result showed stable performance with an average absolute error of 0.0902.

In this paper, we propose an AI-based system for automatically classifying industry and occupation codes in the population census. The accurate classification of industry and occupation codes is crucial for informing policy decisions, allocating resources, and conducting research. However, this task has traditionally been performed by human coders, which is time-consuming, resource-intensive, and prone to errors. Our system represents a significant improvement over the existing rule-based system used by the statistics agency, which relies on user-entered data for code classification. In this paper, we trained and evaluated several models, and developed an ensemble model that achieved an 86.76% match accuracy in industry and 81.84% in occupation, outperforming the best individual model. Additionally, we propose process improvement work based on the classification probability results of the model. Our proposed method utilizes an ensemble model that combines transfer learning techniques with pre-trained models. In this paper, we demonstrate the potential for AI-based systems to improve the accuracy and efficiency of population census data classification. By automating this process with AI, we can achieve more accurate and consistent results while reducing the workload on agency staff.

In this paper, we propose an Internal/External Knowledge Distillation (IEKD), which utilizes both external correlations between feature maps of heterogeneous models and internal correlations between feature maps of the same model for transferring knowledge from a teacher model to a student model. To achieve this, we transform feature maps into a sequence format and extract new feature maps suitable for knowledge distillation by considering internal and external correlations through a transformer. We can learn both internal and external correlations by distilling the extracted feature maps and improve the accuracy of the student model by utilizing the extracted feature maps with feature matching. To demonstrate the effectiveness of our proposed knowledge distillation method, we achieved 76.23% Top-1 image classification accuracy on the CIFAR-100 dataset with the “ResNet-32×4/VGG-8” teacher and student combination and outperformed the state-of-the-art KD methods.

In this paper, we propose a technique of multi-time window feature extraction for anger detection in gait data. In the previous gait-based emotion recognition methods, the pedestrian's stride, time taken for one stride, walking speed, and forward tilt angles of the neck and thorax are calculated. Then, minimum, mean, and maximum values are calculated for the entire interval to use them as features. However, each feature does not always change uniformly over the entire interval but sometimes changes locally. Therefore, we propose a multi-time window feature extraction technique that can extract both global and local features, from long-term to short-term. In addition, we also propose an ensemble model that consists of multiple classifiers. Each classifier is trained with features extracted from different multi-time windows. To verify the effectiveness of the proposed feature extraction technique and ensemble model, a public three-dimensional gait dataset was used. The simulation results demonstrate that the proposed ensemble model achieves the best performance compared to machine learning models trained with existing feature extraction techniques for four performance evaluation metrics.

The utilization of department information in bibliometric analysis using scientific and technological literature is highly advantageous. In this paper, the department information dataset was built through the screening, data refinement, and classification processing of authors’ department type belonging to university institutions appearing in academic journals in the field of science and technology published in Korea, and the automatic classification model based on deep learning was developed using the department information dataset as learning data and verification data. In addition, we analyzed the co-authorship structure and network in the field of computer science using the department information dataset and affiliation information of authors from domestic academic journals. The research resulted in a 98.6% accuracy rate for the automatic classification model using Korean department information. Moreover, the co-authorship patterns of Korean researchers in the computer science and engineering field, along with the characteristics and centralities of the co-author network based on institution type, region, institution, and department type, were identified in detail and visually presented on a map.

In this paper, we propose a data augmentation method based on CNN(Convolutional Neural Network) learning for efficiently obtaining concrete crack image datasets. Real concrete crack images are not only difficult to obtain due to their unstructured shape and complex patterns, but also may be exposed to dangerous situations when acquiring data. In this paper, we solve the problem of collecting datasets exposed to such situations efficiently in terms of cost and time by using vector and thickness-based data augmentation techniques. To demonstrate the effectiveness of the proposed method, experiments were conducted in various scenes using U-Net-based crack detection, and the performance was improved in all scenes when measured by IoU accuracy. When the concrete crack data was not augmented, the percentage of incorrect predictions was about 25%, but when the data was augmented by our method, the percentage of incorrect predictions was reduced to 3%.

Since the introduction of smartphones, the introduction of charging cable infrastructure that can be used for public use is underway. Thanks to this, people use public cables comfortably without doubt, but most people are not aware of the dangers of public cables. These public cables can lead to infringement accidents such as personal information exposure due to the development of hacking cables, and in the worst case, hackers can take control of smartphones and laptops. This study analyzed the operating principles and attack principles of hacking cables that seem like these general charging cables, but contain malicious scripts or hardware inside. In addition, physical and logical countermeasures were considered based on the analysis.

In this paper, we propose a robustness sensitivity index (RSI) of highway networks to analyze the effect of congestion in a specific section on the entire highway. The newly proposed RSI is defined as the change in the total mileage of the transportation network per extended unit length when the length of a particular section is extended. When the RSI value is large, traffic congestion in the section has a worse effect on the entire network than in other sections. The existing network robustness index (NRI) simply observes changes in transportation networks with and without specific sections, but the RSI proposed in this study is a kind of performance indicator that allows quantitative analysis of the ripple effect of the entire network according to the degree of congestion in a specific section. While changing the degree of congestion in a particular section, it is possible to calculate how the traffic volume increases, decreases, and the size and location of the congestion section change. This analysis proves the superiority of RSI as it cannot be analyzed with NRI. Various properties of RSI are analyzed using data from the domestic highway network. In addition, using the RSI concept, it is shown that the ripple effect on other sections in which a change in the degree of congestion of a specific section occurs can be analyzed.

Recently, the technology of drones is developing remarkably. The role of military drones is so great that they can cause serious damage to the enemy's important strategic assets without any damage to our allies in all battlefield environments (land, sea, air). However, the battleship combat management system currently operated by the Korean Navy is vulnerable to defense because there is no customized defense system against drones. As drones continue to develop, they are bound to pose a major threat to navy in the future. This paper proposes a way for the warfare software of naval combat management system sets a combat mode suitable for anti-drone battle, evaluates the threat priority in order to preemptively respond to drone threats and eliminate drone threats through automatic allocation of self-ship-mounted weapons and sensors, and through a test of the improved warfare software in a simulated environment, it was proved that the time to respond to the drone was improved by 62%.

In this paper, we propose a new method for displaying colored defects by measuring the relative density with the wide-area and local densities of X-ray. The relative density of one pixel represents a relative difference from the surrounding pixels, and we also suggest a colorization of X-ray images representing these pixels as normal and defective. The traditional method mainly inspects materials such as plastics and metals, which have large differences in transmittance to the object. Our proposed method can be used to detect defects such as sprouts or holes in images obtained by an inspection machine that detects X-rays. In the experiment, the products that could not be seen with the naked eye were colored with pests or sprouts in a specific color so that they could be used in the agricultural product selection system. Products that are uniformly filled with a single ingredient inside, such as potatoes, carrots, and apples, can be detected effectively. However, it does not work well with bumpy products, such as peppers and paprika. The advantage of this method is that, unlike machine learning, it doesn't require large amounts of data. The proposed method could be applied to a screening system using X-rays and used not only in agricultural product screening systems but also in manufacturing processes such as processed food and parts manufacturing, so that it can be actively used to select defective products.

Smart farms are steadily increasing in research to minimize labor, energy, and quantity put into crops as IoT technology and artificial intelligence technology are combined. However, research on efficiently managing crop growth information in smart farms has been insufficient to date. In this paper, we propose a management technique that can efficiently monitor crop growth information by applying autonomous sensors to smart farms. The proposed technique focuses on collecting crop growth information through autonomous sensors and then recycling the growth information to crop cultivation. In particular, the proposed technique allocates crop growth information to one slot and then weights each crop to perform load balancing, minimizing interference between crop growth information. In addition, when processing crop growth information in four stages (sensing detection stage, sensing transmission stage, application processing stage, data management stage, etc.), the proposed technique computerizes important crop management points in real time, so an immediate warning system works outside of the management criteria. As a result of the performance evaluation, the accuracy of the autonomous sensor was improved by 22.9% on average compared to the existing technique, and the efficiency was improved by 16.4% on average compared to the existing technique.

The Chinese three northeastern three provinces - Heilongjiang, Liaoning, and Jilin - hold significant geographical, geopolitical, and commercial importance due to their location allowing for cross-border trade and transportation with North Korea. These provinces are crucial for establishing a complex Eurasian logistics network in line with South Korea's new northern policy. The Chinese three northeastern three provinces, as this region is known, boasts excellent maritime transportation links between South Korea, China, and North Korea, making it an logistics hub for transporting goods to Eurasia and Europe through multimodal transport. This study highlights the importance of securing a logistics hub area by fostering cooperation and friendly relations with China's three northeastern three provinces, which are crucial to the success of the New Northern Policy. In particular, the study aims to analyze current status of trade with these region and freight volume transported by ships and recommend political advice for securing logistics hub and revitalizing maritime transport. As the policy suggestion, this study is to establish a logistics hub by implementing joint port operations, constructing port infrastructure jointly, and operating shipping companies together. Additionally, we propose ways to revitalize the maritime passenger transport business between Korea and China, which involves expanding cultural exchanges and developing content.

National parks are one of the important types of protected area management systems established by IUCN and a management model for implementing effective conservation and sustainable use of natural and cultural heritage in countries around the world, and they assume important roles in conservation, scientific research, education, recreation and driving community development. In the context of big data, this study takes China's Liaohe National Park, a typical representative of global coastal wetlands, as a case study, and using Python technology to collect tourists' travelogues and reviews from major OTA websites in China as a source. The text spans from 2015 to 2022 and contains 2998 reviews with 166,588 words in total. The results show that wildlife resources, natural landscape, wetland ecology and the fishing and hunting culture of northern China are fully reflected in the perceptions of visitors to Liaohe National Park; visitors have strong positive feelings toward Liaohe National Park, but there is still much room for improvement in supporting services and facilities, public education and visitor experience and participation.

As the use of mobile devices such as PDAs and smart-phones increases, the Mobile Payment System(MPS) is widely used as a new payment method. Accordingly, domestic and foreign MPS users can conveniently purchase or use the products and services they want, free from the constraints of time and space. In this situation, the study was to investigate how the perception of MPS value varies depending on the characteristics of MPS users(demographic characteristics, MPS usage characteristics), and to analyze whether the perception of MPS value ultimately has a significant effect on user satisfaction. To accomplish these research purposes, the study conducted a statistical analysis using a questionnaire for people with experience using MPS in Korea. The analysis results of the study could be useful in seeking ways to successfully spread of MPS at home and abroad and improve user satisfaction in the future.

This study aims to explore the perceptions of MZ‘s on eco-friendly consumption and their preferences for eco-friendly souvenirs. The top reasons for buying upcycling products include design, price, differentiation, and environmentalism, with design having the most influence on product purchase. Design is also a top consideration when buying upcycling souvenirs. Regarding environmental issues, they were most aware of the seriousness of environmental pollution and thought that recycling of fashion products was necessary. When it comes to eco-friendly fashion purchasing behaviors, the most common choice is to buy clothes that will last longer than those that are in fashion. Upcycling products are more likely to be purchased when there is a concern for the environment or an interest in eco-friendly products. In addition, those who have purchased upcycling products are more likely to be aware of and interested in eco-friendly fashion products and recycling of fashion products. Women are more likely than men to be concerned about environmental issues, and women are also more likely to be aware of upcycling souvenirs. In future research, it would be useful to study the relationship between upcycling products, environmental issues, consumer behavior, and upcycling souvenirs.

Due to the Fourth Industrial Revolution, the influence applied to all areas of our society is continuing to develop at a rapid pace as the days go by. Recently, in the field of legal services, artificial intelligence technology has been introduced mainly in the United States, an advanced country, leading innovation in the legal market. As such, artificial intelligence is expected to rapidly grow as a means of replacing people, leaving the auxiliary role of people at a rapid pace, and the purpose of this study is to examine necessary measures for Korean professional legal professionals to survive in this legal market. After analyzing it based on prior research by domestic researchers and various data in Korea, the law was revised to prohibit non-lawyers from handling legal affairs, active state intervention in public information cases, and ways for the state and the private sector to check each other. Therefore, the above research is expected to throw a lot of discussion points in terms of legal services using artificial intelligence in the future.

This study was conducted to explore the necessary conditions and support for stable operation of an expanded AI curriculum in education. A high school that has implemented an AI curriculum since 2020 was targeted, and students and teachers were surveyed on their perceptions of the AI curriculum, implementation and support strategies. The survey items were categorized into 1) experience with AI education, 2) implementation direction of AI education, and 3) expected effects through AI education, and the results were derived focusing on frequency analysis to identify trends. The analysis resulted in three implications. First, it was suggested that the activation of AI education. Second, the need to develop a hands-on AI curriculum and incorporate AI throughout the entire curriculum was highlighted. Third, it was emphasized that efforts to enhance the capabilities of teachers to implement AI teaching and learning, along with the expansion of physical infrastructure for hands-on education, are necessary.

This study propose to develop artificial intelligence liberal arts courses for college students in the humanities and social sciences majors using the entry artificial intelligence model. A group of experts in computer, artificial intelligence, and pedagogy was formed, and the final artificial intelligence liberal arts course was developed using previous research analysis and Delphi techniques. As a result of the study, the educational topics were largely composed of four categories: image classification, image recognition, text classification, and sound classification. The training consisted of 1) Understanding the principles of artificial intelligence, 2) Practice using the entry artificial intelligence model, 3) Identifying the Ethical Impact, and 4) Based on learned, team idea meeting to solve real-life problems. Through this course, understanding the principles of the core technology of artificial intelligence can be directly implemented through the entry artificial intelligence model, and furthermore, based on the experience of solving various real-life problems with artificial intelligence, and it can be expected to contribute positively to understanding technology, exploring the ethics needed in the artificial intelligence era.

In this study, we attempted to examine the changing ways of thinking about lecture evaluation before and after COVID-19. To this end, decision tree analysis(Decision Tree) was used among data mining techniques based on lecture evaluation data for liberal arts and major classes conducted before and after COVID-19 for A university. According to the results of the study, liberal arts changed from 'method' to 'content', and 'knowledge improvement' was an important factor both before and after majors. In particular, 'Assignment' was found to be an important factor after the COVID-19 in common in the evaluation of lectures in the liberal arts department, which means that in the future, professors will be provided with appropriate teaching methods during class, interaction with students, and feedback on assignments or test results, indicates the need for competence. Based on the results of this study, a plan to improve communication with students and activation of blended learning was suggested.

In this paper, artificial intelligence agents ChatGPT and Stable Diffusion were used to explore the possibility of educational use by implementing a program to recommend picture materials for teaching and learning according to the class topic entered by teachers. The average time spent recommending all picture materials is about 6 minutes. In general, pictures related to keywords were recommended, and the letters in the recommended pictures could only know the intention to represent the letters, and the letters could not be recognized and the meaning could not be known. However, further research seems to be needed on the fact that the type or content of the recommended picture depends entirely on the response of ChatGPT and that it is not possible to accurately recommend the picture for all keywords. In addition, it was concluded that it is true that the recommended picture is related to the keyword, but the evaluation of whether it has educational value is the subject of discussion that should be left to the judgment of human teachers.