• Proceedings of the Korea Contents Association Conference
• /
• 2019.05a
• /
• pp.425-426
• /
• 2019

전통적인 마이닝 기법은 다양한 디지털 매체와 센서 등에서 생산되는 빅데이터를 처리하기 어려울 뿐 아니라 신규 데이터 누적시 전체 데이터를 재분석 해야하는 비효율성과 대용량의 문서를 학습함에 있어 메모리부족 문제, 학습 소요시간 문제 등이 있다. 이러한 문제를 해결하기 위하여 본 논문에서는 자질축소 기법에 의존하지 않고 대량의 문서를 자유롭게 학습하고 부분적인 자질 추가 변경 시에 변경요소만을 추가 반영할 수 있는 범용적이고 일반적인 분류기의 구조설계 방법을 설계 및 구현하였다. 점진적 학습 모듈은 일반적인 학습 방법이 데이터의 추가 및 변동시마다 모든 데이터를 재학습하는 데 반해, 기존의 학습 결과에 증분된 데이터만 재처리 없이 추가적으로 학습한다. 재학습을 위해 사용자는 작업 수행 중 자원 관리를 통해 기존에 처리된 데이터를 자유롭게 가져와서 새로운 데이터와 병합이 가능하다. 이러한 점직적 학습 효율성은 빅데이터 기반 데이터 처리에 주요한 특성인 데이터 생산 속도를 극복하기 위한 좋은 대안이 될 수 있음을 확인하였다.

• Annual Conference on Human and Language Technology
• /
• 1997.10a
• /
• pp.360-363
• /
• 1997

코퍼스 속의 언어적 규칙을 직접적으로 사용하여 한국어 의존 구조를 분석하기 위해, 본 한국어 의존 구조 분석기는 의존 구조가 나타나 있는 코퍼스로부터 변환 규칙 학습기로 규칙을 자동적으로 학습하고 그 규칙을 적용함으로써 한국어 의존 구조를 분석한다. 이를 위해 기존의 연구된 구구조 문법의 규칙 틀과는 다른 한국어 의존 구조에 맞는 규칙 틀을 연구하였고 또 의존 구조에서 발생할 수 있는 교차구조(Crossing structure)를 방지하는 연산을 고안하였다.

• 한국어정보학회:학술대회논문집
• /
• 2016.10a
• /
• pp.257-262
• /
• 2016

본 논문은 비구조적인 자연어 문장으로부터 두 개체 사이의 관계를 표현하는 구조적인 트리플을 밝히는 관계추출에 관한 연구를 기술한다. 사람이 직접 언어적 분석을 통해 트리플이 표현되는 형식을 입력하여 관계를 추출하는 규칙 기반 접근법에 비해 기계가 데이터로부터 표현 형식을 학습하는 기계학습 기반 접근법은 더 다양한 표현 형식을 확보할 수 있다. 기계학습을 이용하려면 모델을 훈련하기 위한 학습 데이터가 필요한데 학습 데이터가 수집되는 방식에 따라 지도 학습, 원격지도 학습 등으로 구분할 수 있다. 지도 학습은 사람이 학습 데이터를 만들어야하므로 사람의 노력이 많이 필요한 단점이 있지만 양질의 데이터를 사용하는 만큼 고성능의 관계추출 모델을 만들기 용이하다. 원격지도 학습은 사람의 노력을 필요로 하지 않고 학습 데이터를 만들 수 있지만 데이터의 질이 떨어지는 만큼 높은 관계추출 모델의 성능을 기대하기 어렵다. 본 연구는 기계학습을 통해 관계추출 모델을 훈련하는데 있어 지도 학습과 원격지도 학습이 가지는 단점을 서로 보완하여 타협점을 제시하는 학습 방법을 제안한다.

• Annual Conference on Human and Language Technology
• /
• 2016.10a
• /
• pp.257-262
• /
• 2016

본 논문은 비구조적인 자연어 문장으로부터 두 개체 사이의 관계를 표현하는 구조적인 트리플을 밝히는 관계추출에 관한 연구를 기술한다. 사람이 직접 언어적 분석을 통해 트리플이 표현되는 형식을 입력하여 관계를 추출하는 규칙 기반 접근법에 비해 기계가 데이터로부터 표현 형식을 학습하는 기계학습 기반 접근법은 더 다양한 표현 형식을 확보할 수 있다. 기계학습을 이용하려면 모델을 훈련하기 위한 학습 데이터가 필요한데 학습 데이터가 수집되는 방식에 따라 지도 학습, 원격지도 학습 등으로 구분할 수 있다. 지도 학습은 사람이 학습 데이터를 만들어야하므로 사람의 노력이 많이 필요한 단점이 있지만 양질의 데이터를 사용하는 만큼 고성능의 관계추출 모델을 만들기 용이하다. 원격지도 학습은 사람의 노력을 필요로 하지 않고 학습 데이터를 만들 수 있지만 데이터의 질이 떨어지는 만큼 높은 관계추출 모델의 성능을 기대하기 어렵다. 본 연구는 기계학습을 통해 관계추출 모델을 훈련하는데 있어 지도 학습과 원격지도 학습이 가지는 단점을 서로 보완하여 타협점을 제시하는 학습 방법을 제안한다.

• The Journal of the Korea Contents Association
• /
• v.21 no.2
• /
• pp.395-405
• /
• 2021

This study explored the structural relationship between achievement goal orientation, learning motivation, and level of interest on the level of flow in distance PBL classes. A structural model was constructed through analysis of previous studies. Data for the verification of the structural model were collected from 59 students in the teaching courses. The study took PBL lessons remotely. After the PBL class was ended, flow, achievement goal orientation, learning motivation, and level of interest were measured. The collected data were analyzed using the structural equation model analysis method. As a result, the structural model proposed in this study was verified to be valid. It was verified that the level of interest mediated the influence of achievement goal orientation and learning motivation to influence commitment. Next, it was verified that the path through which the level of interest mediates the effect of achievement goal orientation has the greatest influence. Therefore, it is possible to suggest that the level of flow can be adjusted if the level of interest is adjusted in consideration of the achievement goal orientation of learners when operating distance PBL classes.

• Journal of Gifted/Talented Education
• /
• v.11 no.3
• /
• pp.45-68
• /
• 2001

The Gifted need the constructivist loaming environments that reflect his or her cognitive and affective characteristics and needs to exert their potential fully. 'Structural Communication'was designed to encourage creative thinking in learners, allowing them to create an understanding of a topic, not simply memorize facts. It is considered in line with constructivist philosophy and cognitive paradigms. The major purpose of this study is to explore 'Web-Based Structural Communication'program to embody the collaborative loaming based on constructivism. It was applied on high school $2\times2$ matrix teaming for the gifted students. Recently developed computer technology, emerging network facilities, and internet enable us to extend the usefulness and efficiency of 'Structural Communication’Especially web provides not only the discussion environment that is free from space and time constraints and characteristics of leasers, but also the experiences of knowledge construction through the collaborative learning. Through the 'Web-based Structural Communication', the gifted will be able to argue, persuade and share their unique ideas and gradually elaborated ill-structured ideas. The gifted will escape from the tunnel vision of the early time and have multiple perspectives that are more objective and logical. As the result, the gifted are expected to acquire the effect of 'the Zone of Proximal Development'that Vygotsky advocated.

• Journal of Elementary Mathematics Education in Korea
• /
• v.12 no.2
• /
• pp.101-124
• /
• 2008

The purpose of this study is to test if problem posing based on structural approach cooperative learning has a positive effect on mathematical achievement and mathematical disposition. For this purpose, this study carried out tasks as follows: First, we design a problem posing teaching learning program based on structural approach cooperative learning. Second, we analyze how problem posing based on structural approach cooperative learning affects students' mathematical achievement. Third, we analyze how problem posing based on structural approach cooperative learning affects students' mathematical disposition. The results of this study are as follows: First, in the aspect of mathematical achievement, the experimental group who participated in the problem posing program based on structural approach cooperative teaming showed significantly higher improvement in mathematical achievement than the control group. Second, in the aspect of mathematical disposition, the experimental group who participated in the problem posing program based on structural approach cooperative teaming showed positive changes in their mathematical disposition. Summing up the results, through problem posing based on structural approach cooperative learning, students made active efforts to solve problems rather than fearing mathematics and, as a result, their mathematical achievement was improved. Furthermore, through mathematics classes enjoyable with classmates, their mathematical disposition was also changed in a positive way.

• Journal of the Korea Convergence Society
• /
• v.10 no.11
• /
• pp.181-188
• /
• 2019

The purpose of this study is to implement the convergence thinking in problem-based learning (PBL) on paramedic education. PBL scenario course was conducted for 78 students in the third year of emergency medical technology during the first semester of 15 weeks in 2017. After 15 weeks, data of 73 students were analyzed. Among questions about learning interest in PBL, 'neutral' was the most frequent response with 38% for "PBL scenario classes were more effective in learning and acquiring knowledge than lecture class". For "The lessons learned in the class helped to improve the ability to come up with appropriate solutions for problem solving", 57.5% responded 'agree', and for "The lessons learned in the class helped with confidence in the emergency scene", 50.7% responded 'agree'. PBL will be an effective and efficient way of teaching as a learning curriculum for understanding the field situation.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.15 no.1
• /
• pp.137-146
• /
• 2002

The predictions of the loading characteristics was performed by the neural networks which use the results through structural analysis. The momentum backperpagtion which can be modified the teaming rate and momentum coefficient, was developed. Input patterns of the neural networks are the 9 strains which positioned at the side of the shell and output layers is the loading characteristics. Hidden layers were increased from 1 layers to 3 layers. Developed program which were trained by 9 strains predict the loading characteristics under 0.5%. Inverse engineering can be applicable to the composite laminated cylindrical shells with developed neural networks.

• Journal of the Korea Computer Graphics Society
• /
• v.27 no.5
• /
• pp.73-79
• /
• 2021

Recently, in the field of computer animation, a method for generating motion using deep learning has been studied away from conventional finite-state machines or graph-based methods. The expressiveness of the network required for learning motions is more influenced by the diversity of motion contained in it than by the simple length of motion to be learned. This study aims to find an efficient network structure when the types of motions to be learned are diverse. In this paper, we train and compare three types of networks: basic fully-connected structure, mixture of experts structure that uses multiple fully-connected layers in parallel, recurrent neural network which is widely used to deal with seq2seq, and transformer structure used for sequence-type data processing in the natural language processing field.