• Tuberculosis and Respiratory Diseases
• /
• v.86 no.3
• /
• pp.226-233
• /
• 2023

Background: Inactive or old, healed tuberculosis (TB) on chest radiograph (CR) is often found in high TB incidence countries, and to avoid unnecessary evaluation and medication, differentiation from active TB is important. This study develops a deep learning (DL) model to estimate activity in a single chest radiographic analysis. Methods: A total of 3,824 active TB CRs from 511 individuals and 2,277 inactive TB CRs from 558 individuals were retrospectively collected. A pretrained convolutional neural network was fine-tuned to classify active and inactive TB. The model was pretrained with 8,964 pneumonia and 8,525 normal cases from the National Institute of Health (NIH) dataset. During the pretraining phase, the DL model learns the following tasks: pneumonia vs. normal, pneumonia vs. active TB, and active TB vs. normal. The performance of the DL model was validated using three external datasets. Receiver operating characteristic analyses were performed to evaluate the diagnostic performance to determine active TB by DL model and radiologists. Sensitivities and specificities for determining active TB were evaluated for both the DL model and radiologists. Results: The performance of the DL model showed area under the curve (AUC) values of 0.980 in internal validation, and 0.815 and 0.887 in external validation. The AUC values for the DL model, thoracic radiologist, and general radiologist, evaluated using one of the external validation datasets, were 0.815, 0.871, and 0.811, respectively. Conclusion: This DL-based algorithm showed potential as an effective diagnostic tool to identify TB activity, and could be useful for the follow-up of patients with inactive TB in high TB burden countries.

• The Journal of Bigdata
• /
• v.6 no.2
• /
• pp.85-98
• /
• 2021

The pandemic of COVID-19 further promoted the imbalance in the volume of imports and exports among countries using containers, which worsened the shortage of empty containers. Since it is important to secure as many empty containers as the appropriate demand for stable and efficient port operation, measures to predict demand for empty containers using various techniques have been studied so far. However, it was based on long-term forecasts on a monthly or annual basis rather than demand forecasts that could be used directly by ports and shipping companies. In this study, a daily and weekly prediction method using an actual artificial neural network is presented. In details, the demand forecasting model has been developed using multi-layer perceptron and multiple linear regression model. In order to overcome the limitation from the lack of data, it was manipulated considering the business process between the loaded container and empty container, which the fully-loaded container is converted to the empty container. From the result of numerical experiment, it has been developed the practically applicable forecasting model, even though it could not show the perfect accuracy.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.65 no.1
• /
• pp.15-26
• /
• 2023

This study aimed to evaluate the performance of water level classification from CCTV images in agricultural facilities such as reservoirs. Recently, the CCTV system, widely used for facility monitor or disaster detection, can automatically detect and identify people and objects from the images by developing new technologies such as a deep learning system. Accordingly, we applied the ResNet-50 deep learning system based on Convolutional Neural Network and analyzed the water level of the agricultural reservoir from CCTV images obtained from TOMS (Total Operation Management System) of the Korea Rural Community Corporation. As a result, the accuracy of water level detection was improved by excluding night and rainfall CCTV images and applying measures. For example, the error rate significantly decreased from 24.39 % to 1.43 % in the Bakseok reservoir. We believe that the utilization of CCTVs should be further improved when calculating the amount of water supply and establishing a supply plan according to the integrated water management policy.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.3C
• /
• pp.167-178
• /
• 2008

Rainfall-induced slope failures were simulated by seepage and stability analyses for actual slopes of weathered soils. After undisturbed sampling and testing on a specimen of unsaturated conditions, a seepage analysis was performed under actual rainfall and it was found that the pore water pressure increased at the boundary of soil and rock layers. The safety factor of slope stability decreased below 1.0 and the failure of actual slope could be simulated. Under design rainfall intensity, the seepage analysis could not include the effects of the antecedent rainfall and the rainfall duration. Due to these limitations, the safety factor of slope stability resulted in above 1.0, since the hydraulic head of soil layers had not be affected significantly. In the analysis of another slope failure, the parameters of unsaturated conditions were evaluated using artificial neural network (ANN). In the analysis of seepage, the boundary of soil and rock was saturated sufficiently and then the safety factor could be calculated below 1.0. It was found that the failure of actual slope can be simulated by ANN-based estimation.

• Journal of Korean Society for Quality Management
• /
• v.51 no.2
• /
• pp.283-296
• /
• 2023

Purpose: The purpose of this study is to actually implement and verify whether welding defects can be detected in real time by utilizing deep learning AI solutions in the welding process of electric vehicle hairpin winding motors. Methods: AI's function and technological elements using synthetic neural network were applied to existing electric vehicle hairpin winding motor laser welding process by making special hardware for detecting electric vehicle hairpin motor laser welding defect. Results: As a result of the test applied to the welding process of the electric vehicle hairpin winding motor, it was confirmed that defects in the welding part were detected in real time. The accuracy of detection of welds was achieved at 0.99 based on mAP@95, and the accuracy of detection of defective parts was 1.18 based on FB-Score 1.5, which fell short of the target, so it will be supplemented by introducing additional lighting and camera settings and enhancement techniques in the future. Conclusion: This study is significant in that it improves the welding quality of hairpin winding motors of electric vehicles by applying domestic artificial intelligence solutions to laser welding operations of hairpin winding motors of electric vehicles. Defects of a manufacturing line can be corrected immediately through automatic welding inspection after laser welding of an electric vehicle hairpin winding motor, thus reducing waste throughput caused by welding failure in the final stage, reducing input costs and increasing product production.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.26 no.4C
• /
• pp.255-264
• /
• 2006

The auxiliary reinforcement method is normally applied to prevent a possible collapse of the tunnel face where the ground condition is not favorable or geologic information is not sufficient. Recently, several engineering approaches have been made to choose the effective reinforcement methods using expert system such as neural network and fuzzy theory field, among others. Even if the expert system has offered many decision aid tools to properly select the reinforcement method, the quantitative assessment items are not easy to estimate and this is why the data mining technique, widely used in the field of social science, medical treatment, banking and agriculture, is introduced in this study. Using decision tree together with PDA, the decision aids for reinforcement method based on field construction data are created to derive the field rules and future study will be concentrated on the application of the proposed methods in a variety of underground development cases.

• Smart Media Journal
• /
• v.12 no.5
• /
• pp.28-35
• /
• 2023

In this study, we propose a new light-weight model RoutingConvNet with fewer parameters to improve the applicability and practicality of speech emotion recognition. To reduce the number of learnable parameters, the proposed model connects bidirectional MFCCs on a channel-by-channel basis to learn long-term emotion dependence and extract contextual features. A light-weight deep CNN is constructed for low-level feature extraction, and self-attention is used to obtain information about channel and spatial signals in speech signals. In addition, we apply dynamic routing to improve the accuracy and construct a model that is robust to feature variations. The proposed model shows parameter reduction and accuracy improvement in the overall experiments of speech emotion datasets (EMO-DB, RAVDESS, and IEMOCAP), achieving 87.86%, 83.44%, and 66.06% accuracy respectively with about 156,000 parameters. In this study, we proposed a metric to calculate the trade-off between the number of parameters and accuracy for performance evaluation against light-weight.

• Smart Media Journal
• /
• v.12 no.5
• /
• pp.17-27
• /
• 2023

Recently, automatic text summarization, which automatically summarizes only meaningful information for users, is being studied steadily. Especially, research on text summarization using Transformer, an artificial neural network model, has been mainly conducted. Among various studies, the GSG method, which trains a model through sentence-by-sentence masking, has received the most attention. However, the traditional GSG has limitations in selecting a sentence to be masked based on the degree of overlap of tokens, not the meaning of a sentence. Therefore, in this study, in order to improve the quality of text summarization, we propose SbGSG (Semantic-based GSG) methodology that selects sentences to be masked by GSG considering the meaning of sentences. As a result of conducting an experiment using 370,000 news articles and 21,600 summaries and reports, it was confirmed that the proposed methodology, SbGSG, showed superior performance compared to the traditional GSG in terms of ROUGE and BERT Score.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.363-363
• /
• 2022

LID 시설에 대한 모니터링은 인력을 활용한 실강우 모니터링을 진행하고 있으나 LID 시설은 소규모 분산형시설로서 인력을 동원한 식생고사, 강우시 모니터링, 현장답사 등 꾸준한 시설확인에 한계가 있으며, LID 시설을 조성한 이후 적정한 유지관리 방법(주기, 빈도, 항목 등)을 인지하지 못하여 막힘현상, 효율저하, 식물고사 등의 문제가 발생한다. 따라서 본연구에서는 딥러닝 분석을 활용하여 강우시 강우모니터링 자료와 LID 시설 내 센서를 통해 측정된 자료를 통해 침투도랑 내 유입수 성상에 대한 예측분석을 수행하였다. 심지 내 LID 시설에 유입되는 오염물질을 예측을 위한 딥러닝 분석을 위해 과거 실강우시 모니터링 자료(TSS, COD, TN, TP)와 대기센서(대기습도, 대기온도, 강수량, 미세먼지) 데이터를 활용하여 딥러닝 모델에 대한 적용가능성 평가를 수행하였다. 측정항목에 대한 상관성 분석을 수행하였으며, 딥러닝 모델은 Tenser Flow를 이용하여 DNN(Deep Neural Network)모델을 활용하여 분석하였다. DNN 모델에 대한 MSE값은 0.31로 분석되었으며, TSS에 대한 평균 50.6mg/L로 분석되었으며, COD 평균 98.7 mg/L로 나타났다. TN의 평균 2.21 mg/L로 분석되었으며, TP 평균 0.67 mg/L로 나타났다. 상관계수분석결과 TSS는 0.53로 분석되었으며, TN과 TP의 상관계수는 0.10, 0.56으로 나타났다. COD의 상관계수는 0.63으로 TSS와 COD, TP에 대한 예측이 된 것으로 분석되었다. 딥러닝을 통한 LID 시설 내 농도변화 예측시 강우시 센서데이터 값은 조밀해야하며 오염물질 농도와 상관성이 높은 항목들에 대해 계측과 실강우 모니터링 자료를 축적하여 미래에 대한 활용성을 높여야 한다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.232-234
• /
• 2022

Big data is widely used in customer relationship management, relationship marketing, financial business improvement, credit information and risk management. Moreover, as non-face-to-face financial transactions have become more active recently due to the COVID-19 virus, the use of financial big data is more demanded in terms of relationships with customers. In terms of customer relationship, financial big data has arrived at a time that requires an emotional rather than a technical approach. In relational marketing, it was necessary to emphasize the emotional aspect rather than the cognitive, rational, and rational aspects. Existing traditional financial data was collected and utilized through text-type customer transaction data, corporate financial information, and questionnaires. In this study, the customer's emotional image data, that is, atypical data based on the customer's cultural and leisure activities, is acquired through SNS and the customer's activity image is analyzed with an artificial intelligence CNN algorithm. Activity analysis is again applied to the annotated AI, and the AI big data model is designed to analyze the behavior model shown in the annotation.