• 스마트미디어저널
• /
• 제9권3호
• /
• pp.80-89
• /
• 2020

Out - of - distribution (OOD) samples are frequently encountered when deploying a classification model in plenty of real-world machine learning-based applications. Those samples are normally sampling far away from the training distribution, but many classifiers still assign them high reliability to belong to one of the training categories. In this study, we address the problem of removing OOD examples by estimating marginal density estimation using variational autoencoder (VAE). We also investigate other proper methods, such as temperature scaling, Gaussian discrimination analysis, and label smoothing. We use Chonnam National University (CNU) weeds dataset as the in - distribution dataset and CIFAR-10, CalTeach as the OOD datasets. Quantitative results show that the proposed framework can reject the OOD test samples with a suitable threshold.

• Journal of Information Processing Systems
• /
• 제17권6호
• /
• pp.1057-1070
• /
• 2021

Case-related news filtering is crucial in legal text mining and divides news into case-related and case-unrelated categories. Because case-related news originates from various fields and has different writing styles, it is difficult to establish complete filtering rules or keywords for data collection. In addition, the labeled corpus for case-related news is sparse; therefore, to train a high-performance classification model, it is necessary to annotate the corpus. To address this challenge, we propose topic-enhanced positive-unlabeled learning, which selects positive and negative samples guided by topics. Specifically, a topic model based on a variational autoencoder (VAE) is trained to extract topics from unlabeled samples. By using these topics in the iterative process of positive-unlabeled (PU) learning, the accuracy of identifying case-related news can be improved. From the experimental results, it can be observed that the F1 value of our method on the test set is 1.8% higher than that of the PU learning baseline model. In addition, our method is more robust with low initial samples and high iterations, and compared with advanced PU learning baselines such as nnPU and I-PU, we obtain a 1.1% higher F1 value, which indicates that our method can effectively identify case-related news.

• Nuclear Engineering and Technology
• /
• 제54권4호
• /
• pp.1261-1270
• /
• 2022

This study was conducted to investigate the possibility of using gaze entropy to evaluate an operator's situation awareness in an emergency accident situation of a nuclear power plant. Gaze entropy can be an effective measure for evaluating an operator's situation awareness at a nuclear power plant because it can express gaze movement as a single comprehensive number. In order to determine the relationship between situation awareness and gaze entropy for an emergency accident situation of a nuclear power plant, an experiment was conducted to measure situation awareness and gaze entropy using simulators created for emergency accident situations LOCA, SGTR, SLB, and LOV. The experiment was to judge the accident situation of nuclear power plants presented in the simulator. The results showed that situation awareness and Shannon, dwell time, and Markov entropy had a significant negative correlation, while visual attention entropy (VAE) did not show any significant correlation with situation awareness. The results determined that Shannon entropy, dwell time entropy, and Markov entropy could be used as measures to evaluate situation awareness.

• International journal of advanced smart convergence
• /
• 제12권2호
• /
• pp.84-89
• /
• 2023

In recent years, with the breakthrough of Artificial Intelligence (AI) technology in deep learning algorithms such as Generative Adversarial Networks (GANs) and Variational Autoencoders (VAE), AI generation technology has rapidly expanded in various sub-sectors in the art field. 2022 as the explosive year of AI-generated art, especially in the creation of AI-generated art creative design, many excellent works have been born, which has improved the work efficiency of art design. This study analyzed the application design characteristics of AI generation technology in two sub fields of artistic creative design of AI painting and AI animation production , and compares the differences between traditional painting and AI painting in the field of painting. Through the research of this paper, the advantages and problems in the process of AI creative design are summarized. Although AI art designs are affected by technical limitations, there are still flaws in artworks and practical problems such as copyright and income, but it provides a strong technical guarantee in the expansion of subdivisions of artistic innovation and technology integration, and has extremely high research value.

• Structural Engineering and Mechanics
• /
• 제63권3호
• /
• pp.317-327
• /
• 2017

The addition of different types of polymers such as SBR, VAE, Acrylic, etc. in concrete and mortar leads to an increase in compressive, tensile and bond strength and decrease in permeability of polymer modified mortar (PMM) and concrete (PMC). The improvement in properties such as bond strength and impermeability makes PMM/PMC suitable for use as repair/retrofitting and water proofing material. In the present study effect of addition of fluoropolymer on the strength and permeability properties of mortar has been studied. In the cement mortar different percentages viz. 10, 20 and 30 percent of fluoropolymer by weight of cement was added. It has been observed that on addition of fluoropolymer in mortar the workability of mortar increases. In the present study all specimens were cast keeping the workability constant, i.e., flow value $105{\pm}5mm$, by changing the amount of water content in the mortar suitably. The specimens were cured for two different curing conditions. Firstly, these were cured wet for one day and then cured dry for 27 days. Secondly, specimens were cured wet for 7 days and then cured dry for 21 days. It has been observed that compressive strength and split tensile strength of specimens cured wet for 7 days and then cured dry for 21 days is 7-13 percent and 12-15 percent, respectively, higher than specimens cured one day dry and 27 days wet. The sorptivity of fluoropolymer modified mortar decreases by 88.56% and 91% for curing condtion one and two, respectively. However, It has been observed that on addition of 10 percent fluoropolymer both compressive and tensile strength decreases, but with the increase in percentage addition from 10 to 20 and 30 percent both the strengths starts increasing and becomes equal to that of the control specimen at 30 percent for both the curing conditions. It is further observed that percentage decrease in strength for second curing condition is relatively less as compared to the first curing condition. However, for both the curing conditions chloride ion permeability of polymer modified mortar becomes very low.

• The Journal of Korean Physical Therapy
• /
• 제14권1호
• /
• pp.27-37
• /
• 2002

This is the study of the knee joint injured patients at the orthopaedic surgery clinic where is located in Daejon, who has MCL combine injured ACL reconstruction caused by sport activity and accident during the period from Jan. 2001 to Oct. 2001. By comparing with groups between 7th case of I-group for MCL combined stitch and II-group for ACL reconstruction since 6weeks cast. We have been concluded with that following results. 1. Range of motion for the knee was not limited at 5th case(37%) of I-group, 6th case(42%) of II-group and the cases of Flexion deficit less then 10 -degree were 2nd case(13%) of I-group and II-group 1st case(8%) with no extension deficit more then 5 -degree. 2. The level of activity that tells you whether you are capable of exercise for six month after operation. It han been divided by 3 levels. The case of capable of doing low risk exercise(swimming, cycling, etc.) was 5th case of I-group, the case of capable of doing medium risk exercise(jogging, etc.) was 3rd case of I-group and 4th case of II-group and the case of capable of doing high risk exercise(football, etc.) were 3rd case of I-group and 3rd case of II-group. 3. The timing of the return to their job were average 6.4 weeks for I-group and average 22.9 weeks for II-group(P<.05, statistical difference). 4. There was no statistical difference between I-group and II-group for the timing of the return to their job(P>.05). 5. By using VAS to compare them there was no statistical difference between I-group and II-group of clinical results according to Lysholm scale.

• 산업경영시스템학회지
• /
• 제46권2호
• /
• pp.133-142
• /
• 2023

The large process plant is currently implementing predictive maintenance technology to transition from the traditional Time-Based Maintenance (TBM) approach to the Condition-Based Maintenance (CBM) approach in order to improve equipment maintenance and productivity. The traditional techniques for predictive maintenance involved managing upper/lower thresholds (Set-Point) of equipment signals or identifying anomalies through control charts. Recently, with the development of techniques for big analysis, machine learning-based AAKR (Auto-Associative Kernel Regression) and deep learning-based VAE (Variation Auto-Encoder) techniques are being actively applied for predictive maintenance. However, this predictive maintenance techniques is only effective during steady-state operation of plant equipment, and it is difficult to apply them during start-up and shutdown periods when rises or falls. In addition, unlike processes such as nuclear and thermal power plants, which operate for hundreds of days after a single start-up, because the pumped power plant involves repeated start-ups and shutdowns 4-5 times a day, it is needed the prediction and alarm algorithm suitable for its characteristics. In this study, we aim to propose an approach to apply the optimal predictive alarm algorithm that is suitable for the characteristics of Pumped Storage Power Plant(PSPP) facilities to the system by analyzing the predictive maintenance techniques used in existing nuclear and coal power plants.

• 한국음향학회지
• /
• 제42권6호
• /
• pp.500-510
• /
• 2023

수동소나 신호에는 정상신호와 비정상 신호가 같이 존재하는 경우가 대부분이다. 정상신호와 혼재된 비정상 신호는 주로 정상신호만을 학습하는 오토인코더를 이용하여 탐지된다. 하지만 기존의 오토인코더는 혼재된 신호로부터 왜곡된 정상신호를 복원하므로 부정확한 탐지를 수행할 수 있다. 이러한 한계를 개선하고자, 본 논문에서는 순환신경망과 벡터 양자화 기반의 비정상 신호 탐지 모델을 제안한다. 제안된 모델은 학습된 잠재벡터들을 대표하는 코드 북을 생성하고, 제안된 코드벡터의 탐색을 통해 보다 정확하게 비정상 신호를 탐지한다. 공개된 수중 음향 데이터를 이용한 실험에서 제안된 기법이 적용된 오토인코더와 변이형 오토인코더는 기존 모델에 비해 최소 2.4 % 향상된 탐지 성능과 최소 9.2 % 높은 비정상 신호 추출 성능을 보였다.

• 한국응용곤충학회지
• /
• 제36권4호
• /
• pp.351-356
• /
• 1997

남부지방에서 분리한 곤충병원성곰팡이 Beauveria bassiana GY1-17균주를 이용하여 산림해충인 오리나무잎벌레(Agelastica coerulea), 밤나무혹나방(Meganola melancholica), 회양목명나방(Glyphodes perspectalis), 잔디해충인 등얼룩풍뎅이(Blitopertha orientalis), 채소해충인 배추좀나방(Plutella xylostella) 및 거세미나방(Agrotis segetum)의 생물적 방제 가능성을 알아보기 위하여 실험한 결과, 오리나무잎벌레와 배추좀나방유충은 7.0~2.0$\times$${10}^{7}$ conidia/ml 농도에서 처리 7일과 5일후 100%의 치사율을 나타내었다. 밤나무혹나방유충은 0.03875~3.1$\times$${10}^{7}$ conidia/ml 처리에서 66.7~100%의 높은 치사율을 보였으나 회양목명나방유충은 2.0$\times$${10}^{7}$~2.0$\times$$10^4$conidia/ml 처리에서도 전혀 치사되지 않았다. 등얼룩풍뎅이유충은 3.7$\times$${10}^{7}$ conidia/ml 농도에서 46.7%의 치사율을 보였고, 거세미나방유충은 2.5$\times$${10}^{7}$ conidia/ml 농도에서 63.3% 치사율을 나타내었다.

• Journal of Hospice and Palliative Care
• /
• 제11권3호
• /
• pp.140-146
• /
• 2008

목적: 말기 암 환자의 증가, 노년인구의 증가, 연명치료기술의 발달 등으로 완치될 수 없는 질환을 가진 말기 환자에 대한 치료 결정(End of life decision making)은 늘어나고 있다. 하지만 이에 대한 의료진의 태도나 환자 및 보호자의 인식은 낮은 상태에 있고 이로 인해서 말기 환자에게 말기진정과 같은 결정이 필요한 경우에 있어서도 그 시행에 많은 제약이 있으며 때로는 잘못된 시행으로 윤리적으로 어긋나는 경우가 발생되고 있다. 이에 의과대학 교육과정을 통하여 말기 환자 치료 결정에 대한 올바른 태도 변화를 가져올 수 있으며, 미래 의료계의 주역이 될 의과대학생들이 말기진정을 포함하여 능동적 안락사, 의사조력자살, 연명치료 유보 및 중단 등 말기 환자 치료결정에 대해 어떠한 태도를 가지고 있는가를 알아보기 위하여 본 연구를 시행하였다. 방법: 2007년 6월 25일부터 6월 29일 사이에 1개 의과대학 본과 1, 2학년 학생과 임상 실습중인 3학년 학생 총 388명을 대상으로 능동적 안락사, 의사조력자살, 연명치료 유보 및 중단, 말기 진정 등 말기 환자 치료결정에 대한 태도와 인구사회학적 자료를 설문 조사하였다. 응답이 완료된 267명을 대상으로 말기 환자 치료결정에 대한 태도와 각 인자들 간의 관련성을 분석하였다. 결과: 일개 의과대학생 267명을 대상으로 시행한 설문 조사에서 능동적 안락사, 의사조력자살, 연명치료의 유보 및 중단, 말기 진정의 시행에 찬성하는 비율은 각각 37.1%, 21.7%, 58.4%, 60.3%, 41.6%였다. 이 비율은 각 항목의 윤리적 타당성에 대한 설문 결과와 유사하였다. 1학년보다는 3학년에서 능동적 안락사와 의사조력자살은 더 반대하였고, 연명치료 유보 및 중단, 말기 진정에 대해서는 더 찬성하였다. 종교 활동 시간이 많을수록 각 항목에 대한 찬성이 적었으며 교육 경험 유무, 특히 임상실습경험이 있는 3학년 학생에서 말기 진정에 더 많이 찬성하였다. 연령, 임종 환자 경험 유무가 태도에 미치는 영향은 없었다. 결론: 말기 환자 치료 결정의 구체적 임상 행위에 대한 의과대학생의 태도에 이전 연구에서처럼 종교 또는 교육이 말기 환자 치료 결정에 영향을 미치는 것으로 나타났으며 특히, 임상실습을 통한 교육경험이 태도 변화에 중요하였다.