http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
남녀 대학생의 식행동에 관한 연구(Ⅱ) : 경제적 환경을 중심으로
김명선 한국조리과학회 2005 한국식품조리과학회지 Vol.21 No.2
The purposes of this study were to determine what kinds of food university students had, how often they were eating out and how much they consumed. A questionnaire survey was conducted on 597 university students (male : 40.9%, female : 59.1%). The students who spent under 200 thousand won per month had regular meals at breakfast, lunch and dinner. The students with personal monthly spending over 300 thousand won, both male and female, ate out more than others. The female students had more protein foods than the male students. Consumption of milk and milk products increased with rising monthly spending for female students. Regardless of monthly spending, the male students ate seaweeds low in 1-2 time per a week. Consumption of rice decreased with rising monthly spending for male and female students.
CPU-GPU 협력 스케줄링을 통한 임베디드 시스템 내 다중 DNN 모델들의 성능 향상
김명선 대한전자공학회 2021 전자공학회논문지 Vol.58 No.1
The use of deep neural network (DNN) models has improved the performance of many applications. To get the better performance, research on the DNN model itself is actively progressing, but research on a system that can perform this efficiently is relatively insufficient. In order to satisfy the high recognition success rate and the requirements of various applications, various types of DNN models are used inside embedded systems such as robotics and autonomous vehicles, and the number itself is increasing. Therefore, when a DNN accelerator such as a GPU is shared and used by multiple applications, a performance bottleneck due to the GPU is bound to appear. In this paper, to solve this problem, we propose a scheduling framework that can utilize both CPU and GPU for DNN model computation. This technique uses a method of minimizing data transmission overhead between cores of different types by selecting a core type suitable for the computational characteristics of each layer of DNN models. As a result of experimenting with the proposed technique on an actual commercial board, it is up to 71.1% higher than before applied. DNN(deep neural network) 모델의 활용은 여러 응용 분야의 성능을 향상시켜왔다. 성능 향상을 위하여 DNN 모델 자체에 대한 연구는 활발하게 진행되어 오고 있지만 이를 효율적으로 수행할 수 있는 시스템에 대한 연구는 상대적으로 미비하다. 높은 인식 성공률과 다양한 응용의 요구사항들을 만족하기 위하여 로보틱스 및 자율주행 자동차와 같은 임베디드 시스템 내부에서도 여러 종류의 DNN 모델들이 활용되고 그 숫자 자체도 증가되고 있다. 따라서 GPU와 같은 DNN 가속 장치를 여러 응용들이 공유해서 사용할 때 GPU로 인한 성능 병목 현상이 나타날 수밖에 없다. 본 논문에서는 이러한 문제를 풀고자 CPU와 GPU를 동시에 DNN 모델 연산에 활용할 수 있는 스케줄링 프레임워크를 제안한다. 이 기법은 DNN 모델들의 각각의 레이어의 연산 특성에 알맞은 코어 타입을 선택하여 수행을 맡기고 서로 다른 타입의 코어 간에 데이터 전송 오버헤드를 최소화하는 방법을 사용한다. 제안된 기법을 실제 상용 보드에서 실험한 결과 적용 전 대비 최대 71.1% 향상되었다.
교사의 직무만족도와 교수효능감이 유아의 정서능력, 인성특성, 사회적 능력에 미치는 영향
김명선,황선영,오재연 한국생태유아교육학회 2014 생태유아교육연구 Vol.13 No.3
The purpose of this study is to examine the effects of teachers' job-satisfaction and teaching efficacy on children's emotional competence, personality character, and social competence. The subjects of this study were 197 teachers randomly sampled, selected (1 child per 1 teacher) in kindergartens and daycare centers in Gwangju, Jeonnam, and Jeonbuk. This study shows that there is relationship between teachers’ job satisfaction and teaching efficacy, and children’s emotional competence, personality character, and social competence, and that teachers’ job satisfaction and teaching efficacy have very high effects on children’s emotional competence, personality character, and social competence. Such results suggest that multilateral exploration and political supports from educational institutes, society, and the nation are necessary for teacher related variables like teachers’ job satisfaction and teaching efficacy that have influence on young children. 본 연구는 광주․전남․전북의 유치원과 어린이집에 재원하고 있는 유아와 유아 1명당 담임교사 1인을 기준으로 197명의 교사를 대상으로 유아의 정서능력, 인성특성, 사회적능력에 영향을 미치는 직무만족도와 교수효능감 변인을 분석하였다. 연구결과, 직무만족도와 교수효능감이 유아의 정서능력, 인성특성, 사회적능력에 관련성이 있었으며, 높은 영향력을 미쳤다. 특히, 교수효능감이 직무만족도보다 유아의 정서능력, 인성특성, 사회적능력에 영향을 미치는 것으로 나타나 교수효능감이 높을 때 유아의 정서능력, 인성특성, 사회적능력이 증가함을 알 수 있었다. 이러한 결과를 통해서 유아에게 영향을 미치는 직무만족도와 교수효능감 뿐만 아니라 교사관련 변인 탐색을 위한 다각적인 노력과 지원이 필요하며, 교사가 유아의 행동에 영향을 미칠 수 있는 능력을 가지고 있다고 믿는 교수효능감을 높일 수 있는 교사교육이 필요함을 시사한다.
김명선,김주리,김휘주,손유진,신하늘,이나영,정다영,채유미,최지연,김미영,김태연 이화여자대학교간호학회 2020 이화간호학회지 Vol.- No.54
Purpose: The purpose of this study was to explore what nursing students have experienced in carrying out team projects and what factors should be considered in the following team projects. Methods: The data were collected from June 2019 to January 2020 by nine researchers conducting a semi-structured and face-to-face interview with 10 participants. A total of three interviews per participant were conducted respectively, and additional datas were supplemented by In-person or telephone interview. This study was applied qualitative content analysis method by Elo and Kyngas in data analysis process. Results: Participants' team project experience were emerged into six categories. These were ‘method of team organization’, ‘the number of team members’, ‘team leader’, ‘method of communication’, ‘free rider’, and ‘peer evaluation’. Conclusion: This study is meaningful in that it focused on nursing students and explored their team project experience. It is also important that this study suggested that the number of team members, which has not been noted before in other studies, affects the quality of team projects.
김명선,김재진,권준수 한국임상심리학회 2004 Korean Journal of Clinical Psychology Vol.23 No.2
The temporal stages of the cognitive processes and the neurophysiological mechanisms underlying the working memory were investigated using the 2-back task and event-related potential (ERP). Geometric figures such as a circle, triangle and square were used as stimuli. The target stimulus was a circle, while the triangle and square served as nontargets. The target stimulus was further divided into the match target and nonmatch target. If the current stimulus was a circle, and the circle was presented two trials earlier (match circle: MC), the subjects were required to respond. If the current stimulus was a circle, but the circle was either not presented two trials earlier (nonmatch circle: NC), or the current stimulus was either a triangle or a square (nontarget stimulus: NS), subjects were asked not to respond. The target stimulus, i. e., MC and NC produced a significantly higher N100 amplitude than the nontarget stimulus, which indicates that more attention was directed to the target stimulus than to the nontarget stimulus. In terms of P300, the MC than either the NC or NS elicited a greater amplitude and earlier latency. In addition, the P300 was at a maximum at the parietal sites in the MC, while it was at its highest at the frontal sites in the NC or NS. All three types of stimuli produced a positive slow potential with similar amplitudes and latencies. The amplitude of the slow potential was at a maximum at the frontal sites. These results indicate that such temporal stages of cognitive functions underlying the working memory as attention allocation, updating the working memory contents and the retention of sequential information can be reflected toward the ERP components. The results also suggest that the prefrontal and parietal cortex, which are known to be the neural networks of working memory, are associated with the retention process and updating working memory contents, respectively.
1950·60년대 김한섭 건축에 나타난 입면구성의 표현특성에 관한 연구
김명선,김용춘 대한건축학회지회연합회 2019 대한건축학회연합논문집 Vol.21 No.6
The aim of this study is to examine the characteristics of facade expression designed by Kim. Han-sup, a key architect in the regional modern architecture movement in Gwang-ju and Jeonnam from the 1950’s to 1960’s. The study begins with the architectural background and thoughts of Kim. Han-sup through literature reviews and introduces an interview with Han. chong-un, a former college student and employee of Kim, which explains the aesthetic architecture that appeared in his design of facades that was reasonably applied to materials, technology, structure and function in line with the economically tough period in Korea. Furthermore, in order to investigate the characteristics of his facade expression, detailed analysis items will be derived from precedent studies. This study will attempt to conclude with statements that his aesthetic architecture can be defined as an emphasis on formative mass by contrasting between basic rectangular forms, implementation with detailed three-dimensional facade by extruding differently vertical and horizontal structural lines and with the pursuit of decoration through compartmental faces on his 12 representative architectural works.
원유회수증진을 위한 microseismic monitoring 기법 연구
김명선,변중무,설순지 한국자원공학회 2010 한국자원공학회지 Vol.47 No.6
In this study, a module, which can detect the location of the microseismic from the recorded seismogram, was developed by improving Simulps 14 program used in earthquake seismology. The use of the difference between P and S first arrivals provides the more exact locations than P first arrival only. In addition, the results of examining the sensitivity to the errors in the velocity model used in estimating the locations, show that about 10% error in the velocity model was not significantly affected in the calculation of the location for the model used in this paper. On the other hand, the algorithm which can obtain the moment magnitude and the source parameters of microseism by spectral analysis in frequency domain was developed. Finally, the direction and the aperture of the fractures can be revealed by describing the locations and the magnitudes of microseismics in three dimensions. 이 연구에서는 microseismic 신호의 도달시간을 이용하여 microseismic의 발생위치를 찾는 모듈을 지진에서 사용되는 Simulps 14 프로그램을 개선하여 적용하였다. 개선된 알고리듬을 이용하여 P파의 도달시간을 이용한 경우와 S파와 P파의 도달시간의 차이를 이용한 경우를 비교한 결과 S파와 P파의 도달시간의 차이를 이용하였을 경우 보다 정확한 결과를 얻을 수 있음을 확인하였다. 또한 위치 계산에 사용되는 속도모델의 오차에 대한 개발된 모듈의 민감도를 조사한 결과 이 연구에서 사용된 수치모형의 경우 속도모델이 10% 정도의 오차를 가지고 있을 경우 microseismic 발생 위치 계산에 크게 영향을 미치지 않았다. 한편, 획득한 microseismic 자료를 주파수 영역에서 진폭분석을 통해 다양한 송신변수를 계산해내는 알고리듬을 개발하였다. 마지막으로 획득한 microseismic의 발생 위치와 크기를 3차원으로 도시함으로써 균열의 방향과 크기를 평가할 수 있었다.