• Journal of Advanced Technology Convergence
• /
• v.3 no.2
• /
• pp.9-15
• /
• 2024

Due to the high inflation rate of dining out, along with changes in group meals or cafeteria services, office workers are increasingly using workplace cafeterias to reduce their meal expenses even slightly. With the recent development of ICT technology, various fields are realizing that not only are smartphones becoming more popular, but they are also becoming an integration of the latest technologies. In this paper, we analyze the current status of cafeterias with a large number of customers and propose ways to improve problems or difficulties. Since most people always carry their smartphones for urgent communication or work tasks, we aim to develop a cafeteria management system that utilizes the NFC function of smartphones. By presenting the process from customer entry to menu selection, it will enable more efficient use of the cafeteria.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.3
• /
• pp.79-87
• /
• 2024

Recently, the importance of pre-trained language models (PLM) has been emphasized for natural language processing (NLP) such as text classification, sentiment analysis, and question answering. Korean PLM shows high performance in NLP in general-purpose domains, but is weak in domains such as finance, medicine, and law. The main goal of this study is to propose a language model learning process and method to build a financial-specific language model that shows good performance not only in the financial domain but also in general-purpose domains. The five steps of the financial-specific language model are (1) financial data collection and preprocessing, (2) selection of model architecture such as PLM or foundation model, (3) domain data learning and instruction tuning, (4) model verification and evaluation, and (5) model deployment and utilization. Through this, a method for constructing pre-learning data that takes advantage of the characteristics of the financial domain and an efficient LLM training method, adaptive learning and instruction tuning techniques, were presented.

• Smart Media Journal
• /
• v.13 no.2
• /
• pp.44-51
• /
• 2024

The construction of a stress measurement detaset plays a curcial role in various modern applications. In particular, for the efficient training of artificial intelligence models for stress measurement, it is essential to compare various biases and construct a quality-controlled dataset. In this paper, we propose the construction of a stress measurement dataset with quality management through the comparison of various biases. To achieve this, we introduce strss definitions and measurement tools, the process of building an artificial intelligence stress dataset, strategies to overcome biases for quality improvement, and considerations for stress data collection. Specifically, to manage dataset quality, we discuss various biases such as selection bias, measurement bias, causal bias, confirmation bias, and artificial intelligence bias that may arise during stress data collection. Through this paper, we aim to systematically understand considerations for stress data collection and various biases that may occur during the construction of a stress dataset, contributing to the construction of a dataset with guaranteed quality by overcoming these biases.

• KOREAN JOURNAL OF PACKAGING SCIENCE & TECHNOLOGY
• /
• v.29 no.3
• /
• pp.223-228
• /
• 2023

Sustainable energy supplies without the recharging and replacement of the charge storage device have become increasingly important. Among various energy harvesters, the triboelectric nanogenerator (TENG) has attracted considerable attention due to its high instantaneous output power, broad selection of available materials, eco-friendly and inexpensive fabrication process, and various working modes customized for target applications. In this study, the amount of voltage and current generated was measured by applying the PSD profile random vibration test of the electronic vibration tester and ISTA 3A according to the time of Anodized Aluminum Oxide (AAO) pore widening of the manufactured TENG device Teflon and AAO. The discharge and charging tests of the integrated module during the random simulated transport environment and the recognition distance of RFID were measured while agricultural products (onion) were loaded into the returnable folding plastic box. As a result, it was found that AAO alumina etching processing time to maximize TENG performance was optimal at 31 min in terms of voltage and current generation, and the integrated module applied with the TENG module showed a charging effect even during the continuous use of RFID, so the voltage was kept constant without discharge. In addition, the RFID recognition distance of the integrated module was measured as a maximum of 1.4 m. Therefore, it was found that the surface condition of AAO, a TENG element, has a great influence on the power generation of the integrated module, and due to the characteristics of TENG, the power generation increases as the surface dries, so it is judged that the power generation can be increased if the surface drying treatment (ozone treatment, etc.) of AAO is applied in the future.

• Korean Journal of Clinical Laboratory Science
• /
• v.56 no.2
• /
• pp.105-114
• /
• 2024

Clinical laboratories endeavor to secure quality by establishing effective quality management systems. However, laboratory environments are complex, and single quality control procedures may inadequately detect many errors. Patient-based real-time quality control (PBRTQC) is a laboratory tool that monitors the testing process using algorithms such as Bull's algorithm and several variables, such as average of normal, moving median, moving average, and exponentially weighted moving average. PBRTQC has many advantages over conventional quality control, including low cost, commutability, continuous real-time performance monitoring, and sensitivity to pre-analytical errors. However, PBRTQC is not easily implemented as it requires statistical algorithm selection, the design of appropriate rules and protocols, and performance verification. This review describes the basic concepts, methods, and procedures of PBRTQC and presents guidelines for implementing a patient-based quality management system. Furthermore, we propose the combined use of PBRTQC when the performance of internal quality control is limited. However, clinical evaluations were not conducted during this review, and thus, future evaluation is required.

• Anatomy and Cell Biology
• /
• v.56 no.1
• /
• pp.86-93
• /
• 2023

Age at death estimation has always been a crucial yet challenging part of identification process in forensic field. The use of human skeletons have long been explored using the principle of macro and micro-architecture change in correlation with increasing age. The clavicle is recommended as the best candidate for accurate age estimation because of its accessibility, time to maturation and minimal effect from weight. Our study applies pre-trained convolutional neural network in order to achieve the most accurate and cost effective age estimation model using clavicular bone. The total of 988 clavicles of Thai population with known age and sex were radiographed using Kodak 9000 Extra-oral Imaging System. The radiographs then went through preprocessing protocol which include region of interest selection and quality assessment. Additional samples were generated using generative adversarial network. The total clavicular images used in this study were 3,999 which were then separated into training and test set, and the test set were subsequently categorized into 7 age groups. GoogLeNet was modified at two layers and fine tuned the parameters. The highest validation accuracy was 89.02% but the test set achieved only 30% accuracy. Our results show that the use of medial clavicular radiographs has a potential in the field of age at death estimation, thus, further study is recommended.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.83-88
• /
• 2024

In this research, we propose a Semi-Supervised learning based railroad surface defect detection method. The Resnet50 model, pretrained on ImageNet, was employed for the training. Data without labels are randomly selected, and then labeled to train the ResNet50 model. The trained model is used to predict the results of the remaining unlabeled training data. The predicted values exceeding a certain threshold are selected, sorted in descending order, and added to the training data. Pseudo-labeling is performed based on the class with the highest probability during this process. An experiment was conducted to assess the overall class classification performance based on the initial number of labeled data. The results showed an accuracy of 98% at best with less than 10% labeled training data compared to the overall training data.

• KIPS Transactions on Computer and Communication Systems
• /
• v.13 no.1
• /
• pp.10-20
• /
• 2024

As the online commerce market continues to expand with an increase of diverse products and content, users find it challenging in navigating and in the selection process. Thereafter both platforms and shopping malls are actively working in conducting continuous research on recommendations system to select and present products that align with user preferences. Most existing recommendation studies have relied on user data which is relatively easy to obtain. However, these studies only use a single type of event and their reliance on time dependent data results in issues with reliability and complexity. To address these challenges, this paper proposes a recommendation system that analysis user preferences in consideration of the relationship between various types of event data. The proposed recommendation system analyzes the correlation of multiple events, extracts weights, learns the recommendation model, and provides recommendation services through it. Through extensive experiments the performance of our system was compared with the previously studied algorithms. The results confirmed an improvement in both complexity and performance.

• Journal of the Korea Fashion and Costume Design Association
• /
• v.25 no.4
• /
• pp.59-76
• /
• 2023

The COVID-19 pandemic has brought about environmental severity and new social, economic, and cultural changes. Conscious fashion, which is oriented to sustainable and valuable consumption, has become a trend to consume products produced using eco-friendly and ethical processes, from the selection of the product materials to the manufacturing process. The purpose of this study is to identify the concepts and characteristics of conscious fashion and zero waste, and to explore design trends of zero wastein the new normal era of conscious fashion through the analysis of various cases. The research method is a literature review on conscious fashion based on relevant professional and academic books and articles, designer collections, and campaigns from 2010 to the present, when conscious fashion as eco-friendliness and sustainable fashion became a trend. The concept and characteristics of conscious fashion were examined them in terms of environmental, ethical, social, and cultural aspects and the concept and characteristics of zero waste through previous studies and case analysis. Through this, the trends of zero-waste design in conscious fashion were categorized into: first, an eco-friendly design orientation that utilizes reuse and reduce methods of clothing and fabric; second, a variable design orientation that practices zero waste designs by using diversity of patterns through deconstruction, disassembly, and various cutting methods. Third, long-term circulation of design through the recycling of resources by second-hand trade, the utilization of stock clothing, resale, and availability of eco-friendly materials through the development of new technologies. As an active practice for the sustainable fashion industry expands, it is expected that continuous research will be conducted as a future core value to realize the possibility of long-term circular zero-waste design through social responsibility and conscious recycling, reuse, and reproduction.

• Journal of Korean Society for Quality Management
• /
• v.52 no.1
• /
• pp.77-93
• /
• 2024

Purpose: The purpose of this study is to examine machine learning use cases in manufacturing companies from a digital quality management (DQM) perspective and to analyze and present machine learning research patterns from a quality management perspective. Methods: This study was conducted based on systematic literature review methodology. A comprehensive and systematic review was conducted on manufacturing papers covering the overall quality management process from 2015 to 2022. A total of 3 research questions were established according to the goal of the study, and a total of 5 literature selection criteria were set, based on which approximately 110 research papers were selected. Based on the selected papers, machine learning research patterns according to quality management were analyzed. Results: The results of this study are as follows. Among quality management activities, it can be seen that research on the use of machine learning technology is being most actively conducted in relation to quality defect analysis. It suggests that research on the use of NN-based algorithms is taking place most actively compared to other machine learning methods across quality management activities. Lastly, this study suggests that the unique characteristics of each machine learning algorithm should be considered for efficient and effective quality management in the manufacturing industry. Conclusion: This study is significant in that it presents machine learning research trends from an industrial perspective from a digital quality management perspective and lays the foundation for presenting optimal machine learning algorithms in future quality management activities.