• 대한의용생체공학회:의공학회지
• /
• 제18권3호
• /
• pp.233-241
• /
• 1997

The research on chromosomes is very significant in cytogenetics since genes of the chromosomes control revelation of the inheritance plasma. The human chromosome analysis is widely used to study leukemia, malignancy, radiation hazard, and mutagen dosimetry as well as various congenital anomalies such as Down's, Klinefelter's, Edward's, and Patau's syndrome. The framing and analysis of the chromosome karyogram, which requires specific cytogenetic knowledge is most important in this field. Many researches on automated chromosome karyotype analysis methods have been carried out, some of which produced commercial systems. However, there still remains much room to improve the accuracy of chromosome classification and to reduce the processing time in real clinic environments. In this paper, we proposed a hierarchical artificial neural network(HANN) to classify the chromosome karyotype. We extracted three or four chromosome morphological feature parameters such as centromeric index, relative length ratio, relative area ratio, and chromosome length by preprocessing from ten human chromosome images. The feature parameters of five human chromosome images were used to learn HANN and the rest of them were used to classify the chromosome images. The experiment results show that the chromosome classification error is reduced much more than that of the other researchers using less feature parameters.

• 한국안전학회지
• /
• 제39권1호
• /
• pp.82-95
• /
• 2024

Analyzing the aftermath of events at domestic nuclear power plants brings in the question: "Why do workers not comply with the prescribed procedures?" The current investigation of nuclear power plant events identifies their reasons considering the factors affecting the workers' behaviors. However, there are some complications to it: in addition to confirming the action such as an error or a violation, there is a limit to identifying the intention of the actor. To overcome this limitation, the study analyzed and examined the reasons for non-compliance identified in nuclear power plant events by Reason's rule-related behavior classification. For behavior analysis, I selected unit behaviors for events that are related to human and organizational factors and occurred at domestic nuclear power plants since 2017, and then I applied the rule-related behavior classification introduced by Reason (2008). This allowed me to identify the intentions by classifying unit behaviors according to quality and compliance with the rules. I also identified the factors that influenced unit behaviors. The analysis showed that most often, non-compliance only pursued personal goals and was based on inadequate risk appraisal. On the other hand, the analysis identified cases where it was caused by such factors as poorly written procedures or human system interfaces. Therefore, the probability of non-compliance can be reduced if these factors are properly addressed. Unlike event investigation techniques that struggle to identify the reasons for employee behavior, this study provides a new interpretation of non-compliance in nuclear power plant events by examining workers' intentions based on the concept of rule-related behavior classification.

• Journal of Biosystems Engineering
• /
• 제24권1호
• /
• pp.67-74
• /
• 1999

A prototype cut-flower sorter was developed and tested for its performance with five varieties of roses. Support plates driven by a chain mechanism transported the roses into an image inspection chamber. Color image processing algorithms were developed to evaluate the length, thickness, and straightness of stem and color, height, and maturity of bud. The average absolute errors of the system for the measurements of stem length, stem thickness, and height of bud were 19.7 mm, 0.5 mm, and 3.8 mm, respectively. The results of classification by the sorter were compared with those of a human inspector for straightness of stem and maturity of bud. The classification error for the straightness of stem was 8.6%, when both direct image and reflected image by a mirror were analyzed. The accuracy in classifying the maturity of bud varied among the varieties, the smallest for‘Nobless’(1.5%) and the largest for‘Rote Rose’(13.5%). The time required to process a rose averaged 2.06 seconds, equivalent to the capacity of 1,600 roses per hour.

• Safety and Health at Work
• /
• 제8권3호
• /
• pp.267-275
• /
• 2017

Background: The mining industry is known worldwide for its highly risky and hazardous working environment. Technological advancement in ore extraction techniques for proliferation of production levels has caused further concern for safety in this industry. Research so far in the area of safety has revealed that the majority of incidents in hazardous industry take place because of human error, the control of which would enhance safety levels in working sites to a considerable extent. Methods: The present work focuses upon the analysis of human factors such as unsafe acts, preconditions for unsafe acts, unsafe leadership, and organizational influences. A modified human factor analysis and classification system (HFACS) was adopted and an accident predictive fuzzy reasoning approach (FRA)-based system was developed to predict the likelihood of accidents for manganese mines in India, using analysis of factors such as age, experience of worker, shift of work, etc. Results: The outcome of the analysis indicated that skill-based errors are most critical and require immediate attention for mitigation. The FRA-based accident prediction system developed gives an outcome as an indicative risk score associated with the identified accident-prone situation, based upon which a suitable plan for mitigation can be developed. Conclusion: Unsafe acts of the worker are the most critical human factors identified to be controlled on priority basis. A significant association of factors (namely age, experience of the worker, and shift of work) with unsafe acts performed by the operator is identified based upon which the FRA-based accident prediction model is proposed.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2019년도 춘계학술대회
• /
• pp.395-397
• /
• 2019

조선기술과 항해장비 기술이 발전하고 있지만 여전히 해양사고는 80%이상이 인적과실에서 비롯되고 있다. 인적과실을 저감시켜 해양사고를 절감시키려는 노력은 항해사를 대상으로 면담이나 설문을 시행하는 등 정성적인 연구방식에 많이 의존하고 있어서 객관적인 인적과실의 실체를 규명하는데 제한이 있다. 본 연구에서는 이 같은 단점을 극복하기 위하여 항해사의 항해 업무 수행을 방해하지 않으며 공간적 제한을 극복할 수 있도록 웨어러블 센서를 활용하여 항해사의 동작을 실측하고 상황인지 여부가 항해 수행 동작에 어떤 영향을 미치는지 구분하고자 한다. Full mission ship handling simulator를 활용하여 항해사가 특정한 시나리오를 수행하는 중에 위험성을 가진 장애물을 발견하기 전과 후의 어떤 행동패턴 변화를 보이는지 측정하였다. 구분된 항해 동작 패턴은 항해 위험 상황에서 적절한 조치를 취하고 있는지 여부를 객관적으로 구분하여 인적과실을 절감하는데 활용될 것으로 기대된다.

• 한국컴퓨터정보학회논문지
• /
• 제26권3호
• /
• pp.19-28
• /
• 2021

위내시경 촬영은 조기에 위 병변을 진단하기 위해서 주로 사용한다. 하지만 위내시경을 했음에도 불구하고 위 내부를 자세히 관찰하지 못해서 10~20% 위 병변을 놓치는 경우가 생기는 것으로 보고되고 있다. 미국, 영국, 일본 등의 일부 국가와 세계내시경협회(Wold Endoscopy Organization)에서는 위내시경 시에 맹점 없는 관찰을 위해서 반드시 촬영해야 할 부위에 대한 촬영지침을 제안한 바 있다. 이에 본 논문에서는 수련의가 내시경을 하는 데 있어 위 내부를 자동으로 맹점 없이 관찰하는데 필요한 딥러닝 기술인 해부학적 분류모델을 제안한다. 제안한 모델은 위내시경 이미지에 적합한 전처리 모듈과 데이터 증강 기술들을 사용한다. 실험결과를 통해 최대 F1 점수 99.6% 분류 성능을 확인하였다. 또한, 실제 데이터를 통한 실험결과에서도 에러율이 4% 미만을 보였다. 이러한 성능을 바탕으로 설명 가능한 모델임을 보여 임상에서의 사용 가능성을 확인하였다.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2017년도 제29회 한글 및 한국어 정보처리 학술대회
• /
• pp.56-59
• /
• 2017

CNN(Convolutional Neural Network)을 이용하여 발화 주제 다중 분류 task를 multi-labeling 방법과, cluster 방법을 이용하여 수행하고, 각 방법론에 MSE(Mean Square Error), softmax cross-entropy, sigmoid cross-entropy를 적용하여 성능을 평가하였다. Network는 음절 단위로 tokenize하고, 품사정보를 각 token의 추가한 sequence와, Naver DB를 통하여 얻은 named entity 정보를 입력으로 사용한다. 실험결과 cluster 방법으로 문제를 변형하고, sigmoid를 output layer의 activation function으로 사용하고 cross entropy cost function을 이용하여 network를 학습시켰을 때 F1 0.9873으로 가장 좋은 성능을 보였다.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2015년도 제27회 한글 및 한국어 정보처리 학술대회
• /
• pp.157-162
• /
• 2015

자동통역(Speech-to-speech translation)의 최우선 단계인 음성인식과정에서 발생한 오류문장은 대부분 비문법적 구조를 갖거나 의미를 이해할 수 없는 문장들이다. 이러한 문장으로 자동번역을 할 경우 심각한 통역오류가 발생하게 되어 이에 대한 개선이 반드시 필요한 상황이다. 이에 본 논문에서는 음성인식 오류문장이 정상적인 인식문장에 비해 비문법적이거나 무의미하다는 특징을 이용하여 DNN(Deep Neural Network) 기반 음성인식오류 판별기를 구현하였으며 84.20%의 오류문장 분류성능결과를 얻었다.

• 전기학회논문지
• /
• 제61권9호
• /
• pp.1336-1339
• /
• 2012

In this paper, we focus on solving the classification problem by using semisupervised learning strategy. Traditional classifiers are constructed based on labeled data in supervised learning. Labeled data, however, are often difficult, expensive or time consuming to obtain, as they require the efforts of experienced human annotators. Unlabeled data are significantly easier to obtain without human efforts. Thus, we use AdaBoost algorithm with SVM-KNN classifier to apply semisupervised learning problem and improve the classifier performance. Experimental results on both artificial and UCI data sets show that the proposed methodology can reduce the error rate.

• Healthcare Informatics Research
• /
• 제24권4호
• /
• pp.309-316
• /
• 2018

Objectives: Both the valence and arousal components of affect are important considerations when managing mental healthcare because they are associated with affective and physiological responses. Research on arousal and valence analysis, which uses images, texts, and physiological signals that employ deep learning, is actively underway; research investigating how to improve the recognition rate is needed. The goal of this research was to design a deep learning framework and model to classify arousal and valence, indicating positive and negative degrees of emotion as high or low. Methods: The proposed arousal and valence classification model to analyze the affective state was tested using data from 40 channels provided by a dataset for emotion analysis using electrocardiography (EEG), physiological, and video signals (the DEAP dataset). Experiments were based on 10 selected featured central and peripheral nervous system data points, using long short-term memory (LSTM) as a deep learning method. Results: The arousal and valence were classified and visualized on a two-dimensional coordinate plane. Profiles were designed depending on the number of hidden layers, nodes, and hyperparameters according to the error rate. The experimental results show an arousal and valence classification model accuracy of 74.65 and 78%, respectively. The proposed model performed better than previous other models. Conclusions: The proposed model appears to be effective in analyzing arousal and valence; specifically, it is expected that affective analysis using physiological signals based on LSTM will be possible without manual feature extraction. In a future study, the classification model will be adopted in mental healthcare management systems.