• 대한원격탐사학회지
• /
• 제36권6_3호
• /
• pp.1711-1720
• /
• 2020

최근 급속한 산업화와 도시화로 인해 인위적으로 발생하는 미세먼지(Particulate matter, PM)는 기상 조건에 따라 이동 및 분산되면서 피부와 호흡기 등 인체에 악영향을 미친다. 본 연구는 기상인자를 multiple linear regression(MLR), support vector machine(SVM), 그리고 random forest(RF) 모델의 입력자료로 하여 서울시 PM10 농도를 예측하고, 모델 간 성능을 비교 평가하는데 그 목적을 둔다. 먼저 서울시에 소재한 39개소 대기오염측정망(air quality monitoring sites, AQMS)에서 관측된 PM10 농도 자료를 8:2 비율로 구분하여 모델 훈련과 검증 데이터셋으로 사용되었다. 또한 기상관측소(automatic weather system, AWS)에서 관측되고 있는 자료 중 9개 기상인자(평균기온, 최고기온, 최저기온, 일 강수량, 평균풍속, 최대순간풍속, 최대순간풍속풍향, 황사발생유무, 상대습도)가 모델의 입력자료로 선정되었다. 각 AQMS에서 관측된 PM10 농도와 MLR, SVM, 그리고 RF 모델에 의해 예측된 PM10 농도 간 결정계수(R2)는 각각 0.260, 0.772, 그리고 0.793이었고, RF 모델이 PM10 농도 예측에 가장 높은 성능을 나타냈다. 특히 모델 검증에 사용되는 AQMS 중 관악구와 강남대로 AQMS는 상대적으로 AWS에 가까워 SVM과 RF 모델에서 높은 정확도를 나타냈다. 종로구 AQMS는 AWS에서 비교적 멀리 떨어져 있지만, 인접한 두 AQMS 데이터가 모델 학습에 사용되었기 때문에 두 모델에서 높은 정확도를 나타냈다. 반면 용산구 AQMS는 AQMS 및 AWS에서 비교적 멀리 떨어져 있기에 두 모델의 성능이 낮게 나타냈다.

• 여성건강간호학회지
• /
• 제29권4호
• /
• pp.317-327
• /
• 2023

Purpose: This study investigated the influence of cervical cancer knowledge, human papillomavirus (HPV) knowledge, self-efficacy, and uncertainty on the intention to engage in cervical cancer preventive behavior in HPV-infected women. Methods: This descriptive correlational study was conducted among 129 adult women aged 20 to 65 years who received positive HPV results at a general hospital in Changwon, Korea. The dataset was analyzed using descriptive statistics, the independent t-test, analysis of variance, the Pearson correlation coefficient, and multiple regression. Results: The mean score for the intention to engage in cervical cancer preventive behavior was high (4.43±0.65). This intention was significantly different according to age at first sexual intercourse (F=7.38, p=.001), HPV type (F=4.79, p=.010), vaccination (t=3.19, p=.002), and condom use (t=3.03, p=.003). The intention to engage in cervical cancer preventive behavior showed significant, weak-to-moderate positive correlations with HPV knowledge (r=.22, p=.012) and self-efficacy (r=.42, p<.001). Self-efficacy (β=.46, p<.001), first sexual intercourse at <20 years (β=.45, p<.001), first sexual intercourse at 20-24 years (β=.29, p=. 018), HPV high- and low-risk group infection (β=.26, p=.019), HPV high-risk group infection (β=.26, p=.026), and vaccination (β=.21, p=.007) significantly influenced the intention to engage in cervical cancer preventive behavior. These variables explained 34.6% of variance in intention. Conclusion: Study findings support the need to develop a program that effectively conveys accurate information about cervical cancer prevention to HPV-infected women and helps them enhance self-efficacy to boost the intention to engage in cervical cancer preventive behavior.

• 한국전자파학회논문지
• /
• 제29권7호
• /
• pp.550-559
• /
• 2018

본 논문은 딥뉴럴네트워크(deep neural network: DNN)를 이용해 사람 걸음 및 배경잡음원에 의해 발생한 마이크로 도플러 신호를 탐지 및 분류 처리하는 연구를 제안한다. 기존 분류처리 연구는 경험 및 통계적인 방법을 통해 분류기 성능에 직접적으로 영향을 미치는 의미있는 특징을 추출하기 위한 복잡한 과정을 포함한다. 그러나 딥뉴럴네트워크는 다수의 레이어 층을 단계적으로 통과하는 과정을 통해 점진적으로 특징을 재구성 및 생성하므로, 별도의 특징 추출과정을 생략할 수 있으며, 자연스럽게 네트워크상에서 특징을 생성할 수 있는 이점이 있다. 따라서 본 논문에서는 마이크로 도플러 신호 인식을 위한 딥뉴럴네트워크 효과성 입증을 위해, 이진분류기와 다층클래스 분류기를 다층퍼셉트론과 딥뉴럴네트워크를 통해 설계하고 비교분석한다. 실험 결과, 다층퍼셉트론은 이진분류기의 경우 테스트세트에 대한 분류 정확도가 90.3 %로 측정되었고, 다층클래스 분류기의 경우 테스트세트에 대한 분류정확도가 86.1 %로 측정되었다. 딥뉴럴네트워크는 이진분류기의 경우 테스트세트에 대한 분류 정확도가 97.3 %로 측정되었고, 다층클래스 분류기의 경우 테스트세트에 대한 분류정확도가 96.1 %로 측정되었다.

• 한국과학교육학회지
• /
• 제43권3호
• /
• pp.237-251
• /
• 2023

본 연구는 합성곱 신경망을 활용한 자동 채점 모델을 설계하고 학생의 운동학 그래프 답안에 적용함으로써, 과학 그래프 답안에 대한 자동 채점의 가능성을 탐색하였다. 연구자가 작성한 2,200개의 답안을 2,000개의 훈련 데이터와 200개의 검증 데이터로 데이터셋을 구성하고, 202개의 학생 답안을 100개의 훈련 데이터와 102개의 시험 데이터로 데이터셋을 구성하여 연구를 진행하였다. 먼저, 자동 채점모델을 설계하고 성능을 검증하는 과정에서는 연구자가 작성한 답안 데이터셋을 활용하여 그래프 이미지 분류에 최적화되도록 자동 채점모델을 완성하였다. 다음으로 자동 채점 모델에 훈련 데이터셋을 여러 유형으로 학습시키면서 학생의 시험 데이터셋에 대한 채점을 수행하여 훈련 데이터의 양이 많고 다양할수록 자동 채점 모델의 성능이 향상된다는 것을 확인하였고, 최종적으로 인간 채점과의 일치율은 97.06%, 카파 계수는 0.957, 가중 카파 계수는 0.968을 얻었다. 한편, 훈련 데이터로 학습되지 않은 유형의 답안의 경우 인간 채점자들 간에는 채점이 거의 일치하였으나, 자동 채점 모델은 일치하지 않게 채점하는 것을 확인하였다.

• 한국환경보건학회지
• /
• 제47권5호
• /
• pp.454-461
• /
• 2021

Background: Fish and fishery products (FFPs) unintentionally contaminated with various environmental pollutants are major exposure pathways for humans. To protect human health from the consumption of contaminated FFPs, it is essential to develop a systematic tool for evaluating exposure and risks. Objectives: To regularly, accurately, and quickly evaluate adverse health outcomes due to FFPs contamination, we developed an automated dietary exposure and risk assessment system called HERA (the Human Exposure and Risk Assessment system for chemicals in FFPs). The aim of this study was to develop an overall architecture design and demonstrate the major features of the HERA system. Methods: For the HERA system, the architecture framework consisted of multi-layer stacks from infrastructure to fish exposure and risk assessment layers. To compile different contamination levels and types of seafood consumption datasets, the data models were designed for the classification codes of FFP items, contaminants, and health-based guidance values (HBGVs). A systematic data pipeline for summarizing exposure factors was constructed through down-scaling and preprocessing the 24-hour dietary recalls raw dataset from the Korea National Health and Nutrition Examination Survey (KNAHES). Results: According to the designed data models for the classification codes, we standardized 167 seafood items and 2,741 contaminants. Subsequently, we implemented two major functional workflows: 1) preparation and 2) main process. The HERA system was developed to enable risk assessors to accumulate the concentration databases sustainably and estimate exposure levels for several populations linked to seafood consumption data in KNAHES in a user-friendly manner and in a local PC environment. Conclusions: The HERA system will support policy-makers in making risk management decisions based on a nation-wide risk assessment for FFPs.

• IMMUNE NETWORK
• /
• 제16권3호
• /
• pp.189-194
• /
• 2016

Obesity is characterized as an accumulation of adipose tissue mass represented by chronic, low-grade inflammation. Obesity-derived inflammation involves chemokines as important regulators contributing to the pathophysiology of obesity-related diseases such as cardiovascular disease, diabetes and some cancers. The obesity-driven chemokine network is poorly understood. Here, we identified the profiles of chemokine signature between human preadipocytes and adipocytes, using PCR arrays and qRT-PCR. Both preadipocytes and adipocytes showed absent or low levels in chemokine receptors in spite of some changes. On the other hand, the chemokine levels of CCL2, CCL7-8, CCL11, CXCL1-3, CXCL6 and CXCL10-11 were dominantly expressed in preadipocytes compared to adipocytes. Interestingly, CXCL14 was the most dominant chemokine expressed in adipocytes compared to preadipocytes. Moreover, there is significantly higher protein level of CXCL14 in conditioned media from adipocytes. In addition, we analyzed the data of the chemokine signatures in adipocytes obtained from healthy lean and obese postmenopausal women based on Gene Expression Omnibus (GEO) dataset. Adipocytes from obese individuals had significantly higher levels in chemokine signature as follows: CCL2, CCL13, CCL18-19, CCL23, CCL26, CXCL1, CXCL3 and CXCL14, as compared to those from lean ones. Also, among the chemokine networks, CXCL14 appeared to be the highest levels in adipocytes from both lean and obese women. Taken together, these results identify CXCL14 as an important chemokine induced during adipogenesis, requiring further research elucidating its potential therapeutic benefits in obesity.

• 한국융합학회논문지
• /
• 제13권1호
• /
• pp.81-88
• /
• 2022

대학수학능력시험 국어 과목에서 중요한 비중을 차지하는 독서 영역의 주된 교육 목표는 주어진 지문을 온전히 이해할 수 있는가를 평가하는 데에 있다. 따라서 해당 지문에 포함된 질의를 주어진 지문만으로 풀이할 수 있는지는 해당 영역의 교육 목표와 관련이 깊다. 본 연구에서는 처음으로, 교육학 분야와 딥러닝을 접목하여 이러한 교육 목표가 실제로도 타당하게 실현 가능한지를 입증하고자 한다. 대학수학능력시험의 독서 영역의 개별지문과 그에 수반된 다수의 문장 쌍(sentence pair)을 정제하여 추출하고, 해당 문장 쌍을 주어진 지문에 비추어 적절하거나(T), 적절하지 않은지(F)를 판단하는 이진 분류 태스크(binary classification task)에 적용하여 평가하고자 한다. 그 결과, F1 스코어 기준 59.2%의 human performance를 뛰어넘는 성능을 62.49%의 KoELECTRA를 비롯한 대부분의 언어 모델에서 확인할 수 있었으며, 또한 데이터 전처리 과정에 변화를 줌으로써 언어 모델의 구조적 한계를 극복할 수 있었다.

• 대한의용생체공학회:의공학회지
• /
• 제45권1호
• /
• pp.1-9
• /
• 2024

Communication with others plays an important role in human social interaction and information exchange in modern society. However, some individuals have difficulty in communicating due to dysarthria. Therefore, it is necessary to develop effective diagnostic techniques for early treatment of the dysarthria. In the present study, we propose a mobile device-based methodology that enables to automatically classify dysarthria type. The light-weight CNN model was trained by using the open audio dataset of Korean patients with dysarthria. The trained CNN model can successfully classify dysarthria into related subtype disease with 78.8%~96.6% accuracy. In addition, the user-friendly mobile application was also developed based on the trained CNN model. Users can easily record their voices according to the selected inspection type (e.g. word, sentence, paragraph, and semi-free speech) and evaluate the recorded voice data through their mobile device and the developed mobile application. This proposed technique would be helpful for personal management of dysarthria and decision making in clinic.

• Bulletin of the Korean Chemical Society
• /
• 제32권7호
• /
• pp.2397-2404
• /
• 2011

Variable selection k nearest neighbor QSAR modeling approach was applied to a data set of 80 3-arylisoquinolines exhibiting cytotoxicity against human lung tumor cell line (A-549). All compounds were characterized with molecular topology descriptors calculated with the MolconnZ program. Seven compounds were randomly selected from the original dataset and used as an external validation set. The remaining subset of 73 compounds was divided into multiple training (56 to 61 compounds) and test (17 to 12 compounds) sets using a chemical diversity sampling method developed in this group. Highly predictive models characterized by the leave-one out cross-validated $R^2$ ($q^2$) values greater than 0.8 for the training sets and $R^2$ values greater than 0.7 for the test sets have been obtained. The robustness of models was confirmed by the Y-randomization test: all models built using training sets with randomly shuffled activities were characterized by low $q^2{\leq}0.26$ and $R^2{\leq}0.22$ for training and test sets, respectively. Twelve best models (with the highest values of both $q^2$ and $R^2$) predicted the activities of the external validation set of seven compounds with $R^2$ ranging from 0.71 to 0.93.

• 전기학회논문지
• /
• 제67권9호
• /
• pp.1224-1231
• /
• 2018

Automatic age estimation has been used in many social network applications, practical commercial applications, and human-computer interaction visual-surveillance biometrics. However, it has rarely been explored. In this paper, we propose an automatic age estimation system, which includes face detection and convolutional deep learning based on an inception module. The latter is a 22-layer-deep network that serves as the particular category of the inception design. To evaluate the proposed approach, we use 4,000 images of eight different age groups from the Adience age dataset. k-fold cross-validation (k = 5) is applied. A comparison of the performance of the proposed work and recent related methods is presented. The results show that the proposed method significantly outperforms existing methods in terms of the exact accuracy and off-by-one accuracy. The off-by-one accuracy is when the result is off by one adjacent age label to the above or below. For the exact accuracy, the age label of "60+" is classified with the highest accuracy of 76%.