• International Journal of Computer Science & Network Security
• /
• 제22권5호
• /
• pp.175-181
• /
• 2022

COVID-19 is an acute respiratory syndrome that affects the host's breathing and respiratory system. The novel disease's first case was reported in 2019 and has created a state of emergency in the whole world and declared a global pandemic within months after the first case. The disease created elements of socioeconomic crisis globally. The emergency has made it imperative for professionals to take the necessary measures to make early diagnoses of the disease. The conventional diagnosis for COVID-19 is through Polymerase Chain Reaction (PCR) testing. However, in a lot of rural societies, these tests are not available or take a lot of time to provide results. Hence, we propose a COVID-19 classification system by means of machine learning and transfer learning models. The proposed approach identifies individuals with COVID-19 and distinguishes them from those who are healthy with the help of Deep Visual Embeddings (DVE). Five state-of-the-art models: VGG-19, ResNet50, Inceptionv3, MobileNetv3, and EfficientNetB7, were used in this study along with five different pooling schemes to perform deep feature extraction. In addition, the features are normalized using standard scaling, and 4-fold cross-validation is used to validate the performance over multiple versions of the validation data. The best results of 88.86% UAR, 88.27% Specificity, 89.44% Sensitivity, 88.62% Accuracy, 89.06% Precision, and 87.52% F1-score were obtained using ResNet-50 with Average Pooling and Logistic regression with class weight as the classifier.

• 인터넷정보학회논문지
• /
• 제20권4호
• /
• pp.91-102
• /
• 2019

자율주행 시스템에서, 카메라에 포착된 영상을 통하여 보행자를 분류하는 기능은 보행자 안전을 위하여 매우 중요하다. 기존에는 HOG(Histogram of Oriented Gradients)나 SIFT(Scale-Invariant Feature Transform) 등으로 보행자의 특징을 추출한 후 SVM(Support Vector Machine)으로 분류하는 기술을 사용했었으나, 보행자 특징을 위와 같이 수동(handcrafted)으로 추출하는 것은 많은 한계점을 가지고 있다. 따라서 본 논문에서는 CNN(Convolutional Neural Network)의 깊은 특징(deep features)과 전이학습(transfer learning)을 사용하여 보행자를 안정적이고 효과적으로 분류하는 방법을 제시한다. 본 논문은 2가지 대표적인 전이학습 기법인 고정특징추출(fixed feature extractor) 기법과 미세조정(fine-tuning) 기법을 모두 사용하여 실험하였고, 특히 미세조정 기법에서는 3가지 다른 크기로 레이어를 전이구간과 비전이구간으로 구분한 후, 비전이구간에 속한 레이어들에 대해서만 가중치를 조정하는 설정(M-Fine: Modified Fine-tuning)을 새롭게 추가하였다. 5가지 CNN모델(VGGNet, DenseNet, Inception V3, Xception, MobileNet)과 INRIA Person데이터 세트로 실험한 결과, HOG나 SIFT 같은 수동적인 특징보다 CNN의 깊은 특징이 더 좋은 성능을 보여주었고, Xception의 정확도(임계치 = 0.5)가 99.61%로 가장 높았다. Xception과 유사한 성능을 내면서도 80% 적은 파라메터를 학습한 MobileNet이 효율성 측면에서는 가장 뛰어났다. 그리고 3가지 전이학습 기법중 미세조정 기법의 성능이 가장 우수하였고, M-Fine 기법의 성능은 미세조정 기법과 대등하거나 조금 낮았지만 고정특징추출 기법보다는 높았다.

• 자동차안전학회지
• /
• 제7권3호
• /
• pp.14-18
• /
• 2015

Toughened glass is known to have about four times larger external impact resistance than that of original glass. This study is aimed to verify that ceramic-printed toughened glass does not meet of GTR(Global Technical Regulations) No. 6 and its strength is lower than that of original glass through tests. The tests were conducted on the test pieces of original glass, toughened glass, and ceramic-printed toughened glass from five glass manufacturers. In Test 1, a 227g steel ball was dropped from a height of 2 meters, and damage was checked according to the test method of GTR No. 6. In Test 2, a steel ball was freely dropped from different heights and limited damage height was determined. In the result of Test 1 according to the test method of GTR No. 6, while all five test pieces of toughened glasses were not damaged, all the ceramic-printed toughened glass from the five manufacturers were damaged. In the result of Test 2, none of the five test pieces of toughened glass were damaged by a 10m ball drop, meanwhile, the original glasses were damaged by an average of 3m ball drop. And the results of the tests show that the ceramic-printed toughened glass does not have the features of toughened glass due to its very low strength. Therefore, this study contributes to the safety of consumers by considering the GTR No. 6, and by revising the toughened glass test method.

• 한국유통학회지:유통연구
• /
• 제16권1호
• /
• pp.95-115
• /
• 2011

본 연구는 Kano모델을 활용하여 프랜차이즈 매장에 관한 다양한 품질요인들의 속성을 고객의 관점에서 분류하였다. 또한, 각 품질요인들이 고객의 만족 또는 불만족에 미치는 상대적 영향력을 분석해 보고자 만족지수와 불만족지수를 산출하였다. 자료 수집을 위해 외식 프랜차이즈 매장 방문 경험이 있는 서울 및 전국광역시 거주 성인들을 대상으로 온라인조사를 실시하였으며, 총 257개의 응답이 분석에 사용되었다. 분석 결과, 해당 품질요소가 충족이 되지 않는 경우 소비자의 불만으로 이어지는 요소에는 매장 청결도, 직원 친절도 및 숙련도, 편의시설 제공 등이 포함되는 것으로 나타났다. 프랜차이즈 사업에서 매장 간 음식메뉴의 구색, 가격, 품질수준, 인테리어, 고객서비스 절차 등의 표준화는 중요하게 생각되어 왔으나, 이 중 음식 가격의 동일성만이 고객의 불만족과 깊은 관계를 가지고 있었다. 충족이 되지 않아도 상관없지만 충족이 되는 경우 고객들의 호의적인 반응을 이끌어낼 수 있는 요소로는 외부기관으로부터의 수상 또는 인증 경력, 프랜차이즈 브랜드의 해외진출, 경품이벤트 및 사용금액에 따라 혜택을 주는 로열티 프로그램의 실시, 그리고 우수한 매장접근성이 해당되었다. 프랜차이즈 브랜드를 상대적으로 자주 이용하는 헤비유저의 경우, 정기적인 신메뉴 출시 또한 매력적인 품질요인으로 생각하고 있었다. 본 논문은 경영자가 우선적으로 관심을 두고 개선하여야 하는 부분과 경쟁력 확보를 위해 추가적으로 투자해야 할 부분이 어디인가에 대한 시사점을 제공해 준다는 점에서 연구의 의의가 있다.

• 대한인간공학회지
• /
• 제35권5호
• /
• pp.371-381
• /
• 2016

Objective:This study aims to evaluate the features of heart rate variability (HRV) and respiratory signals as indices for a driver's drowsiness and waking status in order to develop the classification model for a driver's drowsiness and waking status using those features. Background: Driver's drowsiness is one of the major causal factors for traffic accidents. This study hypothesized that the application of combined bio-signals to monitor the alertness level of drivers would improve the effectiveness of the classification techniques of driver's drowsiness. Method: The features of three heart rate variability (HRV) measurements including low frequency (LF), high frequency (HF), and LF/HF ratio and two respiratory measurements including peak and rate were acquired by the monotonous car driving simulation experiments using the photoplethysmogram (PPG) and respiration sensors. The experiments were repeated a total of 50 times on five healthy male participants in their 20s to 50s. The classification model was developed by selecting the optimal measurements, applying a binary logistic regression method and performing 3-fold cross validation. Results: The power of LF, HF, and LF/HF ratio, and the respiration peak of drowsiness status were reduced by 38%, 22%, 31%, and 7%, compared to those of waking status, while respiration rate was increased by 3%. The classification sensitivity of the model using both HRV and respiratory features (91.4%) was improved, compared to that of the model using only HRV feature (89.8%) and that using only respiratory feature (83.6%). Conclusion: This study suggests that the classification of driver's drowsiness and waking status may be improved by utilizing a combination of HRV and respiratory features. Application: The results of this study can be applied to the development of driver's drowsiness prevention systems.

• 한국융합학회논문지
• /
• 제12권7호
• /
• pp.193-205
• /
• 2021

인공지능 스피커 시장은 꾸준히 성장하고 있지만, 실제 스피커 사용자들의 만족도는 42%에 그치고 있다. 따라서, 본 연구에서는 인공지능 스피커의 세대별 토픽 변화와 감성 변화를 통해 사용자 경험을 저해하는 요소는 무엇인지 분석해 보고자 한다. 이를 위해 아마존 에코 닷 3세대와 4세대 모델에 대한 리뷰를 수집하였다. 토픽모델링 분석 기법을 사용하여 세대별로 리뷰를 이루는 주제 및 주제의 변화를 찾아내고, 딥러닝 기반 감성 분석을 통해 토픽에 대한 사용자 감성이 세대에 따라 어떻게 변화되었는지 살펴보았다. 토픽모델링 결과, 세대별로 5개의 토픽이 도출되었다. 3세대의 경우 스피커의 일반적 속성을 나타내는 토픽은 제품에 긍정적 반응 요인으로 작용했고, 사용자 편의 기능은 부정적 반응 요인으로 작용했다. 반대로 4세대에서는 일반적 속성은 부정적으로, 사용자 편의 기능은 긍정적으로 도출되었다. 이와 같은 분석은 방법론 측면에서 어휘적 특징뿐 아니라 문장 전체의 문맥적 특징이 고려된 분석결과를 제시할 수 있다는 것에 그 의의가 있다.

• Animal Bioscience
• /
• 제36권2호
• /
• pp.256-263
• /
• 2023

Objective: Feed molecular structures can affect its availability to gastrointestinal enzymes which impact its digestibility and absorption. The molecular spectroscopy-attenuated total reflectance Fourier transform infrared vibrational spectroscopy (ATR-FTIR) is an advanced technique that measures the absorbance of chemical functional groups on the infrared region so that we can identify and quantify molecules and functional groups in a feed. The program aimed to reveal the association of intrinsic molecular structure with nutrient supply to animals from canola feedstocks and co-products from bio-oil processing. The objective of this study was to characterize special intrinsic carbohydrate and protein-related molecular structure spectral profiles of feedstock and co-products (meal and pellets) from bio-oil processing from two source origins: Canada (CA) and China (CH). Methods: The samples of feedstock and co-products were obtained from five different companies in each country arranged by the Canola Council of Canada (CCC). The molecular structure spectral features were analyzed using advanced vibrational molecular spectroscopy-ATR-FTIR. The spectral features that accessed included: i) protein-related spectral features (Amide I, Amide II, α-helix, β-sheet, and their spectral intensity ratios), ii) carbohydrate-related spectral features (TC1, TC2, TC3, TC4, CEC, STC1, STC2, STC3, STC4, TC, and their spectral intensity ratios). Results: The results showed that significant differences were observed on all vibrationally spectral features related to total carbohydrates, structural carbohydrates, and cellulosic compounds (p<0.05), except spectral features of TC2 and STC1 (p>0.05) of co-products, where CH meals presented higher peaks of these structures than CA. Similarly, it was for the carbohydrate-related molecular structure of canola seeds where the difference between CA and CH occurred except for STC3 height, CEC and STC areas (p>0.05). The protein-related molecular structures were similar for the canola seeds from both countries. However, CH meals presented higher peaks of amide I, α-helix, and β-sheet heights, α-helix:β-sheet ratio, total amide and amide I areas (p<0.05). Conclusion: The principal component analysis was able to explain over 90% of the variabilities in the carbohydrate and protein structures although it was not able to separate the samples from the two countries, indicating feedstock and coproducts interrelationship between CH and CA.

• Korean Journal of Radiology
• /
• 제22권4호
• /
• pp.652-662
• /
• 2021

Objective: To investigate the association between CT imaging features and survival outcomes in patients with primary invasive mucinous adenocarcinoma (IMA). Materials and Methods: Preoperative CT image findings were consecutively evaluated in 317 patients with resected IMA from January 2011 to December 2015. The association between CT features and long-term survival were assessed by univariate analysis. The independent prognostic factors were identified by the multivariate Cox regression analyses. The survival comparison of IMA patients was investigated using the Kaplan-Meier method and propensity scores. Furthermore, the prognostic impact of CT features was assessed based on different imaging subtypes, and the results were adjusted using the Bonferroni method. Results: The median follow-up time was 52.8 months; the 5-year disease-free survival (DFS) and overall survival rates of resected IMAs were 68.5% and 77.6%, respectively. The univariate analyses of all IMA patients demonstrated that 15 CT imaging features, in addition to the clinicopathologic characteristics, significantly correlated with the recurrence or death of IMA patients. The multivariable analysis revealed that five of them, including imaging subtype (p = 0.002), spiculation (p < 0.001), tumor density (p = 0.008), air bronchogram (p < 0.001), emphysema (p < 0.001), and location (p = 0.029) were independent prognostic factors. The subgroup analysis demonstrated that pneumonic-type IMA had a significantly worse prognosis than solitary-type IMA. Moreover, for solitary-type IMAs, the most independent CT imaging biomarkers were air bronchogram and emphysema with an adjusted p value less than 0.05; for pneumonic-type IMA, the tumors with mixed consolidation and ground-glass opacity were associated with a longer DFS (adjusted p = 0.012). Conclusion: CT imaging features characteristic of IMA may provide prognostic information and individual risk assessment in addition to the recognized clinical predictors.

• 정보관리학회지
• /
• 제31권1호
• /
• pp.143-161
• /
• 2014

한국의 dCollection이 학위논문 수집기로 활용되고 있는데 반해, 일본의 레포지토리는 다양한 학술컨텐츠를 수집, 보존, 확산하고 오픈 엑세스를 실현하기 위한 개별 기관의 자발적인 운영 도구로 발전되고 있다. 본 연구는 일본의 기관 레포지토리 통계 DB인 IRDB를 통해 레포지토리의 특성을 통계적으로 분석하고 구축된 컨텐츠량, 종별 구축 비율, 그리고 종간 상관성을 살펴보았다. 또한 등록된 컨텐츠 특성을 변수로 K-means 군집 분석을 수행함으로써, 일본에 형성된 기관 레포지토리가 어떻게 유형화될 수 있는지 분석하였다. 분석 결과, 일본의 기관 레포지토리는 교내학술논문, 학위논문, 기술보고서, 의학자료, 학술잡지논문 등 다양한 컨텐츠를 수용하고 있을 뿐 아니라, 컨텐츠의 특징에 따라 5개의 차별화된 군집으로 유형화됨으로써 다양한 모습으로 발전되어 가고 있었다.

• 한국식생활문화학회지
• /
• 제26권6호
• /
• pp.673-681
• /
• 2011

The purpose of this study was to analyze the difference in bakery choice attributes according to consumers' general characteristics and purchasing behavior. Among 350 questionnaires distributed to bakery consumers, 277 complete questionnaires (79.1%) were analyzed. Bakery choice attributes were classified into five factors: "environment and image", "bakery product features", "location", "employee service", and "price and sales promotion"; the mean scores of these factors' importance levels were 3.59, 3.58, 3.49, 3.36, and 3.00, respectively. Males considered 'employee service' factor significantly more than did females. Further, the importance level of 'employee service' factor was significantly greater as consumer's age increased. The importance levels of 'bakery product features' and 'employee service' factors were considered significantly more by consumers who spent KRW10,000-15,000 than those who spent KRW5,000 or less. 'Price and sales promotion' was considered to be more important by consumers who obtained information from the Internet than from the TV and radio. 'Location' factor was considered to be more significant as purchasing frequency increased. Such differences in importance level of bakery choice attributes according to consumers' gender, age, job, and purchasing behavior should be considered and applied to the development of marketing strategies targeted at consumers.