• The Transactions of the Korea Information Processing Society
• /
• v.13 no.4
• /
• pp.189-198
• /
• 2024

Lipreading is one of the important parts of speech recognition, and several studies have been conducted to improve the performance of lipreading in lipreading systems for speech recognition. Recent studies have used method to modify the model architecture of lipreading system to improve recognition performance. Unlike previous research that improve recognition performance by modifying model architecture, we aim to improve recognition performance without any change in model architecture. In order to improve the recognition performance without modifying the model architecture, we refer to the cues used in human lipreading and set other regions such as chin and cheeks as regions of interest along with the lip region, which is the existing region of interest of lipreading systems, and compare the recognition rate of each region of interest to propose the highest performing region of interest In addition, assuming that the difference in normalization results caused by the difference in interpolation method during the process of normalizing the size of the region of interest affects the recognition performance, we interpolate the same region of interest using nearest neighbor interpolation, bilinear interpolation, and bicubic interpolation, and compare the recognition rate of each interpolation method to propose the best performing interpolation method. Each region of interest was detected by training an object detection neural network, and dynamic time warping templates were generated by normalizing each region of interest, extracting and combining features, and mapping the dimensionality reduction of the combined features into a low-dimensional space. The recognition rate was evaluated by comparing the distance between the generated dynamic time warping templates and the data mapped to the low-dimensional space. In the comparison of regions of interest, the result of the region of interest containing only the lip region showed an average recognition rate of 97.36%, which is 3.44% higher than the average recognition rate of 93.92% in the previous study, and in the comparison of interpolation methods, the bilinear interpolation method performed 97.36%, which is 14.65% higher than the nearest neighbor interpolation method and 5.55% higher than the bicubic interpolation method. The code used in this study can be found a https://github.com/haraisi2/Lipreading-Systems.

• Journal of Internet Computing and Services
• /
• v.25 no.1
• /
• pp.99-107
• /
• 2024

Along with economic growth and industrial development, there is an increasing demand for various electronic components and device production of semiconductor, SMT component, and electrical battery products. However, these products may contain foreign substances coming from manufacturing process such as iron, aluminum, plastic and so on, which could lead to serious problems or malfunctioning of the product, and fire on the electric vehicle. To solve these problems, it is necessary to determine whether there are foreign materials inside the product, and may tests have been done by means of non-destructive testing methodology such as ultrasound ot X-ray. Nevertheless, there are technical challenges and limitation in acquiring X-ray images and determining the presence of foreign materials. In particular Small-sized or low-density foreign materials may not be visible even when X-ray equipment is used, and noise can also make it difficult to detect foreign objects. Moreover, in order to meet the manufacturing speed requirement, the x-ray acquisition time should be reduced, which can result in the very low signal- to-noise ratio(SNR) lowering the foreign material detection accuracy. Therefore, in this paper, we propose a five-step approach to overcome the limitations of low resolution, which make it challenging to detect foreign substances. Firstly, global contrast of X-ray images are increased through histogram stretching methodology. Second, to strengthen the high frequency signal and local contrast, we applied local contrast enhancement technique. Third, to improve the edge clearness, Unsharp masking is applied to enhance edges, making objects more visible. Forth, the super-resolution method of the Residual Dense Block (RDB) is used for noise reduction and image enhancement. Last, the Yolov5 algorithm is employed to train and detect foreign objects after learning. Using the proposed method in this study, experimental results show an improvement of more than 10% in performance metrics such as precision compared to low-density images.

• Journal of the Korean Society for Library and Information Science
• /
• v.58 no.1
• /
• pp.5-30
• /
• 2024

The purpose of this study is to extract topics from experimental data using the topic modeling methods(LDA, Top2Vec, and BERTopic) and compare the characteristics and differences between these models. The experimental data consist of 55,442 papers published in 85 academic journals in the field of library and information science, which are indexed in the Web of Science(WoS). The experimental process was as follows: The first topic modeling results were obtained using the default parameters for each model, and the second topic modeling results were obtained by setting the same optimal number of topics for each model. In the first stage of topic modeling, LDA, Top2Vec, and BERTopic models generated significantly different numbers of topics(100, 350, and 550, respectively). Top2Vec and BERTopic models seemed to divide the topics approximately three to five times more finely than the LDA model. There were substantial differences among the models in terms of the average and standard deviation of documents per topic. The LDA model assigned many documents to a relatively small number of topics, while the BERTopic model showed the opposite trend. In the second stage of topic modeling, generating the same 25 topics for all models, the Top2Vec model tended to assign more documents on average per topic and showed small deviations between topics, resulting in even distribution of the 25 topics. When comparing the creation of similar topics between models, LDA and Top2Vec models generated 18 similar topics(72%) out of 25. This high percentage suggests that the Top2Vec model is more similar to the LDA model. For a more comprehensive comparison analysis, expert evaluation is necessary to determine whether the documents assigned to each topic in the topic modeling results are thematically accurate.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.6
• /
• pp.534-541
• /
• 2009

Word recognition test (WRT) for the children can be useful for diagnosing the degree of communication disability, prescribing hearing instruments, planning aural rehabilitation and speech therapy, and determination of site of lesions. The Korean standard monosyllabic word lists for preschoolers (KS-MWL-P) were developed considering the criteria given by the literatures. However, the authors of KS-MWL-P suggested more children should be included to verify homogeneity of the lists using psychometric function curve since only 8 children participated in the developing process. The purpose of this study was to explore the homogeneity of KS-MWL-P for supplementing the limitations of the lists employing psychometric analysis. To 23 preschoolers who have normal-hearing, 100 monosyllabic KS-MWL-P words were examined with the pictures. Psychometric function curve with linear slopes of 20% and 80%'s correct rates through accounting recognition scores of each monosyllabic word at variable intensities from -10 to 40 dBHL was obtained and analyzed. As a result, s-shaped psychometric function curve was presented with increasing correct rate depending on intensity and showed no statistical significant differences among each word and list. The congruous graph shapes among lists also indicated good homogeneity and the list 1,2,3,4's average slopes were 4.48, 3.86, 4.65, 4.50. It was verified that the homogeneity was suitable because the analysis of variance showed no statistical significance among lists (p>0.05). However, KS-MWL-P's order of slope according to the order of the number of items, $1{\sim}10$, $1{\sim}20$, $1{\sim}25$ showed no difference with the p-value of 0.93, 0.59, 0.91, 0.70 for the lists 1,2,3, and 4, respectively. Although KS-MWL-P was assumed that the lower-numbered items were easy for testing younger ages, this study's results could not agree with the author's conclusion. Considering this matter, rearranging of the number of items should be performed according to the analysis of slope suggested by this study for testing younger children with easier items. Other than this, in conclusion, KS-MWL-P was proved to be useful for clinical and rehabilitative evaluating and training tools for preschoolers.

• Food Science and Preservation
• /
• v.30 no.4
• /
• pp.703-715
• /
• 2023

This study was conducted to investigate the fermentation characteristics and anti-obesity effects of acetic acid fermentation products of coffee wine. The live cell counts, soluble solids, pH and total acidity of the acetic acid unfermented coffee wine (AUFCW; day 0, before fermentation) were 6.35 log CFU/mL, 8.10 °Brix, 3.88, and 1.29%, respectively, while the acetic acid fermented coffee wine (AFCW; day 15, after fermentation) were 4.40 log CFU/mL, 8.57 °Brix, 3.07, and 7.45%, respectively. Pancreatic lipase inhibitory activity tended to increase as the acetic acid fermentation period increased. The anti-obesity effects of AFCW on 3T3-L1 cells, which was induced by MDI, were evaluated based on the lipid accumulation rate, leptin expression, and fat production-related gene expression (PPAR-γ and SREBP-1c) at the mRNA level. In the case of AFCW, the lipid accumulation rate and leptin expression were decreased to 69.37% and 50.20% at a concentration of 200 ㎍/mL, respectively, and the expression levels of PPAR-γ and SREBP-1c at the mRNA level were decreased to 79.89% and 48.81%, respectively. These results indicate that anti-obesity effect of acetic acid fermentation products could be increased by acetic acid fermentation of coffee wine.

• The Korean Society of Law and Medicine
• /
• v.24 no.3
• /
• pp.59-104
• /
• 2023

This essay attempts to analyze the public health welfare conception within the text of the Beveridge Report and its realization via the NHS in Great Britain. Once referring to the influence of the Report to create the foundations of the 20th-century welfare system, the quest to scrutinize the original intentions of the Report and its succession to the NHS is certainly intriguing. Furthermore, when regarding the significance of public health policy for a modern state, the effort to engage in Beveridge's conception and its realization is more than timely. In light of such a premise, this paper indulges in its study by the following methods. First, the historical background of the Report - namely, the role of the spirit of the age and the experience of the Emergency Medical Service are to be analyzed to identify the origins of the welfare policies proposed by Beveridge. Furthermore, the public health welfare conception of the Report conceived from its time is reflected upon by engaging on the goal towards social welfare and public health scheme. Lastly, the aims of the NHS and its management, treatment classification, and rehabilitation program are reviewed for comparative analysis with the Report to survey the realization of Beveridge's design. In this process, this paper not only takes into account the original text of the Report - but also other essential works of law and public policy, including the NHS Constitution for England and the National Health Service Act of 1946. The intentions of this study are not bound by merely coinciding with the Report, but resonate significance via reflecting upon the Beveridgian legacy on the modern welfare state from the current perspective. The structured analysis to research the aims and policies of the Report and to compare them to the reality of the NHS may provide an opportunity to confirm the realization of Beveridge's scheme in British society. In addition, this essay is part of an academic endeavor to critically assess the past and the present of the welfare institution in the public health sector. As such, it is hopeful that the essay sheds light on further studies concerning the constructive remedies of the Korean welfare system as well.

• Journal of the Korean Society of Radiology
• /
• v.18 no.3
• /
• pp.293-300
• /
• 2024

On October 31, 2023, the revision of the Medical Technologist Act made it mandatory to complete field training courses in order to obtain a license as a radiologic technologist. Therefore, we would like to survey the actual situation of field training in medical institutions to inform the revised Medical Technologist Act and propose improvement measures to increase the effectiveness of field training. A survey was conducted from March to April, 2023, among radiologic technologists working in medical institutions. The questionnaire was sent through a form on a domestic portal site, Company N, and 120 respondents completed it. Eighty-two respondents, or 68.3 percent, had experience in educating on-the-job training students. 58% of the respondents were aware of the fact that the amendment to the Act on Medical Technologist etc. made field training mandatory to obtain a radiologic technologist license. In accordance with Article 9 of the Medical Technologist Act, which prohibits unlicensed persons from practicing, 50% of the respondents were aware that those who are in training to complete an education course equivalent to the license they are seeking to obtain at a university or other institution are allowed to practice as medical Technologists. When asked what is currently taught during fieldwork, 6% of respondents said that they are required to perform radiation-generating activities in addition to observing, guiding patients, and positioning and moving patients. When asked about the future direction of education as fieldwork becomes mandatory for licensure, 77% of respondents said that they will teach more than they currently do. When asked about the appropriate total length of fieldwork, 35% said 12 weeks and 480 hours, 33% said 8 weeks and 320 hours, and 27% said 16 weeks and 640 hours. It can be seen that the current on-the-job training is inadequate according to various regulations, and students' satisfaction is low. However, with the revision of the Act on Medical Technologists, field training has become mandatory to obtain a license as a radiologist, and it is necessary to improve the educational conditions of field training. Therefore, it is necessary to comply with the Nuclear Safety Act and the Rules on the Safety Management of Diagnostic Radiation Generating Devices, introduce standardized training objectives and evaluation systems, designate training hospitals and radiologists in charge of training, and introduce extended training periods and simulation exercises to internalize field training.

• Asia-Pacific Journal of Business Venturing and Entrepreneurship
• /
• v.19 no.1
• /
• pp.55-73
• /
• 2024

Unicorn companies are attracting attention around the world as they are recognized for their high corporate value in a short period of time as an innovative business models. Their growth process presents good lessons for the startup ecosystem and have a positive impact on national economic development and job creation. However, previous studies related to unicorn companies are focused on 'event studies' and 'case studies' such as characteristics of founders, environmental factors, business models and success/failure cases of companies already recognized as unicorns rather than a multifaceted approach. The occurrence of unicorn companies and Macroscopic analysis of related factors is lacking. Against this background, this study are considering the characteristics of unicorns examined through previous research and the current status unicorns with a high proportion of technology companies, the purpose was to analyze the impact of the country's technological competitiveness, such as 'technology human resource index', 'R&D index', and 'technology infrastructure index', on the increase in unicorn companies. For statistical analysis, data published by various international organizations, the Bank of Korea, and Statistics Korea from 2017 to 2020 and unicorn company data compiled by CB Insights were used as panel data for 44 countries to be tested by multiple regression analysis. As a result of the study, it was confirmed that the number of science majors had a positive (+) effect on the increase of unicorn companies in the case of technology human resource index, and in the case of R&D index, the total amount of R&D investment had a positive (+) effect on the increase of unicorn companies, while the number of Triad Patents Families and the number of scientific and technological papers published had a negative (-) effect on the increase of unicorn companies. Finally, in the case of technology infrastructure index, it was confirmed that the number of the world's 500th-ranked universities had a positive (+) effect on the increase of unicorn companies. This study is the first to reveal the causal relationship between national technological competitiveness and unicorn company growth based on country-specific and time-series empirical data, which were insufficiently covered in previous studies. and compared to the UN's ranking of the global industrial competitiveness index and the OECD's total R&D investment by country, Korea is considered to have technological and growth potential, while the number of unicorn companies driving growth as leaders of the innovative economy is relatively small, so the research results can be used when establishing policies to discover and foster unicorn companies in the future.

• Korean Journal of Heritage: History & Science
• /
• v.57 no.3
• /
• pp.24-35
• /
• 2024

Recently, research using deep learning technologies such as artificial intelligence, convolutional neural networks, etc. has been actively conducted in various fields including healthcare, manufacturing, autonomous driving, and security, and is having a significant influence on society. In line with this trend, the present study attempted to apply deep learning to the classification of archaeological artifacts, specifically ancient Korean roof-end tiles. Using 100 images of roof-end tiles from each of the Goguryeo, Baekje, and Silla dynasties, for a total of 300 base images, a dataset was formed and expanded to 1,200 images using data augmentation techniques. After building a model using transfer learning from the pre-trained EfficientNetB0 model and conducting five-fold cross-validation, an average training accuracy of 98.06% and validation accuracy of 97.08% were achieved. Furthermore, when model performance was evaluated with a test dataset of 240 images, it could classify the roof-end tile images from the three dynasties with a minimum accuracy of 91%. In particular, with a learning rate of 0.0001, the model exhibited the highest performance, with accuracy of 92.92%, precision of 92.96%, recall of 92.92%, and F1 score of 92.93%. This optimal result was obtained by preventing overfitting and underfitting issues using various learning rate settings and finding the optimal hyperparameters. The study's findings confirm the potential for applying deep learning technologies to the classification of Korean archaeological materials, which is significant. Additionally, it was confirmed that the existing ImageNet dataset and parameters could be positively applied to the analysis of archaeological data. This approach could lead to the creation of various models for future archaeological database accumulation, the use of artifacts in museums, and classification and organization of artifacts.

• Journal of the Korean Chemical Society
• /
• v.68 no.5
• /
• pp.259-273
• /
• 2024

This study aimed to investigate changes in metamodeling recognition among chemistry teachers through a teacher educational program related to redox models and water electrolysis experiments. To this end, a science model education program was developed for 9 chemistry teachers and conducted over 10 lessons for a total of 40 hours. In addition, a pre- and post-survey was administered to determine teachers' metamodeling recognition in non-contextual and contextual situations. As a result of the study, through the science model education program, teachers showed educational effects in both non-contextual and contextual situations. In the case of non-contextual situations, the stages of scientific metamodeling knowledge development of chemistry teachers came out differently depending on the type of question. For example, the nature or purpose of the model, the modeling process, or the evaluation and improvement of the model improved from low to high, but there was no significant change because the perception of model change and diversity was already high in advance. In the case of contextual situations, the stage of scientific metamodeling knowledge development improved from objectivity to subjectivity in both the redox model theory class and the water electrolysis model experiment class. Therefore, through the 5E circular learning model-based education program, chemistry teachers' perception of metamodeling was clearly improved. However, the modeling activities of teachers in the water electrolysis model experiment class were different from the change in metamodeling perception. The types that teachers selected as additional experiments for modeling were analyzed in two ways. The first type is when they are interested in finding an ideal condition in which the ratio of hydrogen and oxygen gas is close to 2:1 through additional experiments. The second type is when additional experiments are designed with interest in why the experimental results are coming out like that. It was analyzed that the second type was the experiment necessary for modeling. In addition, modeling activities were analyzed into two types. The first was a type in which water molecules were directly decomposed in two electrodes to generate hydrogen and oxygen gas. This type was the case of regression to the textbook model regardless of the experimental results, and 6 chemistry teachers out of 8 were analyzed as this type. The second type was the type in which water reacted at the (+) electrode to generate other substances, and hydrogen ions reacted at the (-) electrode to generate hydrogen gas. Teachers who performed these modeling activities corresponded to the second type in additional experiments, and 2 chemistry teachers out of 8 corresponded to this. Therefore, it is necessary to provide an experience of activities corresponding to the second type of experiment and modeling through an educational program that provides an experience of directly modeling through experiments in order to develop modeling capabilities, unlike the development of metamodeling knowledge.