http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
이완우 한국정서학습장애아교육학회 2002 정서ㆍ행동장애연구 Vol.18 No.1
The purpose of this study is to find out differences of the career development status visually impaired students and sighted(non-visually impaired) college students. The subjects of this study were 29 college students of the first semester in 1998. The instrument used for this study was the Career Development Inventory developed by Super in 1981. The main results showed: First, on the level of career plan, when compared visually impaired college students with sighted ones, upper-class students, male students and visually impaired students have been statistically higher. Second, on the level of career exploration, in comparison of visually impaired students with sighted ones, male students have been higher that female students. Third, on the level of career decision making, in comparison of visually impaired college students with sighted ones, female students have been higher than male students. Fourth, on the level of world of work information, in comparison of visually impaired students with sighted ones, female students have been higher that male students. Fifth, on the level of career development attitudes, in comparison of sighted college students with visually impaired ones, male students have been higher than female students, upper-class students have been higher than lower class ones. Sixth, on the level of career development knowledge and skills, in comparison of visually impaired college students with sighted ones, female students have been higher than male students. Seventh, on the level of career orientation total, in comparison of visually impaired college students with sighted ones, upper-class male students have been higher.
LRP(Layer-wise Relevance Propagation)를 이용한 딥러닝 데이터세트 건전성 확인에 대한 연구
김택규(Taek-Kyu Kim),김장열(Jang-Yeol Kim) 대한전자공학회 2020 대한전자공학회 학술대회 Vol.2020 No.8
In the nuclear field, deep learning is highly expected to improve not only safety, but also the economy. Deep learning is a class of machine learning based on learning data representation, as opposed to task-specific algorithms, and is based on statistical methods. It analyzes collected dataset to determine meaningful features and predicts future data. However, deep learning has a disadvantage that is depended on trained dataset. If the wrong data is included in dataset, deep learning can make wrong decisions. Thus, a method for examining dataset quality is required. We propose applying layer-wise relevance propagation (LRP) algorithm to check the dataset quality. We additionally implement LRP algorithm on our deep-learning model and check the LRP pattern for each input. According to this inspection, we find wrong data in dataset and improve the dataset quality.
이완우 한국서지학회 2012 서지학보 Vol.0 No.39
이 글은 최근 공개된 금속활자를 고려시대 석문․사경․대장경과 비교해본 것이다. 石文과의 비교에서 이 활자가 초당 구양순 서풍에서 나온 1125년 <대각국사비>에 가까웠다. 또 일부 자형은 중당 안진경 이후 유행된 것임도 알았다. 따라서 이 활자가 안진경 서풍이 유행한 12세기 이후 또는 <대각국사비>가 세워진 1125년 이후 주조되었을 것으로 추정했다. 또 충렬왕 초 1270~83년 사경과 비교하여 이 활자가 13세기 후반을 크게 넘지 않는다고 보았다. 또 대장경 판본과의 비교에서 재조본이나 寺刊本 불경과 유사한 성격이 있음을 알게 되었다. This study is to compare the metal types recently discovered with stelae, sutra inscriptions, and Tripitaka of Goryeo period. By comparing with the stele for the monk Daegak in 1125, which is come from the writing style of Ou Yang-xun, it could be found that the types and the stele are similar in character style. Also some characters of the types were originated from the style of Yan Zhen-qing. Therefore, it is possible to estimate that the types was produced after 12th century, when the style of Yan Zhen-qing was popular in Goryeo. And by comparing with sutra inscriptions in 1270~83, it could be found that the casting period of the types never exceed after the 13th century. Futhermore comparing with the Tripitaka woodblock-prints, it could be found that the types is very similar with the woodblock-prints which were published by temples before the Second Tripitaka.