• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제14권4호
• /
• pp.240-248
• /
• 2014

We present a robust speaker identification algorithm that uses novel features based on soft bag-of-word representation and a simple Naive Bayes classifier. The bag-of-words (BoW) based histogram feature descriptor is typically constructed by summarizing and identifying representative prototypes from low-level spectral features extracted from training data. In this paper, we define a generalization of the standard BoW. In particular, we define three types of BoW that are based on crisp voting, fuzzy memberships, and possibilistic memberships. We analyze our mapping with three common classifiers: Naive Bayes classifier (NB); K-nearest neighbor classifier (KNN); and support vector machines (SVM). The proposed algorithms are evaluated using large datasets that simulate medical crises. We show that the proposed soft bag-of-words feature representation approach achieves a significant improvement when compared to the state-of-art methods.

• 한국실내디자인학회논문집
• /
• 제21권2호
• /
• pp.197-206
• /
• 2012

This study focuses on comparative study of the representation of 'traditionality' in Asian hotel guest rooms. Hotel can be a concentration of the country's culture and tradition and provide unique experience to guests through its space, decoration, and material. However, most hotels in Korea are lack of a strong identity based on the Korean culture and tradition, due to adoptation of the western hotel styles without any criticism. The purpose of this study was to investigate the expression methods of the traditionality in Asian hotel guest rooms to provide design guideline to enhance identity of hotel guest rooms in Korea. Through analysis of the previous researches, criteria of the three design application methods and the five design elements were defined to analyze the representation of the traditonality. Design application methods were categorized as "Original form", "Partial adoption", and "Metaphor". Five design elements include "Shape", "Material", "Color", "Object", and "Pattern". Thirty nine Asian hotels containing the traditional design elements were explored in the study. In result, design application methods in Korea used all three methods equally, while other Asian countries used mainly the Partial adoption and Metaphor methods to express their traditions rather than the Original form method. All five deign elements were mostly used in case of the Original form methods, and two or three elements among five elements were used for the Partial adoption and Metaphor methods. The traditional representation of hotel guest rooms in Korea, reflecting current thinking, living pattern and culture, will be a solution for the new hotel design as well as elevation of Korea's status to a higher level.

• 한국수학교육학회지시리즈C:초등수학교육
• /
• 제7권1호
• /
• pp.15-29
• /
• 2003

The present study examined the 7th national elementary school mathematics curriculum from a perspective of mathematical creativity. The study investigated to what extent the activities in the Pattern and Function lessons in the national elementary school mathematics textbooks promoted the development of mathematical creativity. The results indicated that the current elementary school mathematics curriculum was limited in many ways to promote the development of mathematical creativity. Regarding the activities in Pattern lessons, for example, most activities presented closed tasks involving finding and extending patterns. The lesson provided little opportunities to explore the relationships among various patterns, apply patterns to different situations, or create ones own patterns. In regard to the Function lessons, the majority of activities were about computing the rate. This showed that the function was taught from an operational perspective, not a relational perspective. It was unlikely that students would develop the basic understanding of function through the activities involving the computing the rate. Further, the lessons had students use exclusively the numbers in representing the function. Students were provided little opportunities to use various representation methods involving pictures or graphs, explore the strengths and limitations of various representation methods, or to choose more effective representation methods in particular contexts. In conclusion, the lesson activities in the current elementary school mathematics textbooks were unlikely to promote the development of mathematical creativity.

• 한국지능시스템학회논문지
• /
• 제11권9호
• /
• pp.817-821
• /
• 2001

본 논문은 분류 문제의 훈련 패턴으로부터 형성되는 커널 공간의 저밀도 표현을 가능하게 하는 커널 방법에 대한 새로운 학습방법론을 제안한다. 선형 판별 함수에 대한 기존의 학습법 중에서 이완 절차가 SVM(Support Vector Machine) 분류기와 동등하게 선형분리 가능 패턴분류 문제의 최대 마진 분리 초평면을 얻을 수 있다. 기존의 이완 절차는 지원 백터에 대한 필요 조건을 만족한다. 본 논문에서는 학습 중 지원 벡터를 확인하기 위한 충분 조건을 제시한다. 순차적 학습을 위하여 기존의 SVM을 확장하고 커널 판별함수를 정의한 후에 체계적인 학습방법을 제시한다. 실험 결과는 새 방법이 기존의 방법과 동등하거나 우수한 분류 성능을 갖고있음을 보여준다.

• 인지과학
• /
• 제19권1호
• /
• pp.75-88
• /
• 2008

시각 장면은 중복적인 정보가 많이 포함되어 있다. 우리의 시각체계는 다양하고 중복적인 정보를 처리하기 위해 뇌 용적을 늘이기보다는 들어오는 외부 정보를 요약한다. 유사한 형태의 다양한 정보가 시각체계에 주어지면 시각체계는 정보의 통계적 특성을 추출해 낸다. 이런 통계적 표상의 대표적 형태가 바로 평균 표상이다. 평균 표상의 한 예로 시각 체계에서 계산해 내는 유사한 여러 크기들의 평균 크기를 들 수 있다. 평균 표상은 빠르고 정확하며 비교적 오랜 시간 지속되는 표상이고 평균 표상의 처리과정 또한 병렬적인 처리과정이다. 하지만 지금까지의 통계 표상에 관한 연구는 행동측정방법에 의한 연구였다. 따라서 본 연구는 기능적 자기 공명 영상 기법을 사용하여 통계 표상에 관한 신경기제를 찾고자 하였다. 사전 연구 결과들에 따르면 특정 자극을 연속하여 제시하였을 때 특정 자극을 담당하는 영역에서 자기 공명 영상 신호가 감소함을 알 수 있다. 본 연구에서는 이 반복 감소 현상을 사용하여 원들의 평균이 동일한 자극을 제시하였을 때 우측 후두 영역에서 유의미하게 자기 공명 영상 신호가 감소하는 것을 발견하였다. 이것은 우측 후두 영역이 시각자극에 대한 평균 표상을 처리하는 영역일 수 있음을 시사한다.

• 대한수학교육학회지:학교수학
• /
• 제18권1호
• /
• pp.127-141
• /
• 2016

본 연구는 Chauvat의 그래프 사용 모드에 근거하여 고등학교 1학년 학생의 이차함수 문제해결에서 나타나는 그래프 표상의 사용 모드를 분석하고자 한다. 이 분석으로부터 Bannister (2014)의 표상의 유연성을 통해 연구 참여 학생들의 이차함수 이해 정도를 조사하였다. 그 결과 고등학교 1학년 학생들이 주로 사용하는 그래프 표상 모드는 계산 도표학적 모드이며, 조작적 모드를 사용할 경우에는 오류를 발생하는 것을 알 수 있었다. 그리고 함수의 이해를 대상과 과정 관점에서 표상의 사용으로 분류한 Bannister(2014)의 유연성의 분류에서는 과정 관점으로 함수를 이해하고 두 표상 사이에 조작이 일어나지 않는 경직된 형태를 보이는 것으로 나타났다. 이러한 결과를 바탕으로 교실에서 학생들을 위한 그래프 표상 사용에 대한 교육 및 다양한 관점으로 함수를 이해할 수 있는 교수 -학습 방법에 대한 연구가 필요할 것으로 보인다.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2007년도 학술발표회 논문집
• /
• pp.909-910
• /
• 2007

The volatile organic compounds exposure is governed by the source distance and dispersion of the pollutant into air and groundwater. The purpose of this study was to validate suggested models for the prediction of concentration distributions. The study design was organized into different methods to simulate industry site. The distribution models generally showed a fair agreement with measured data. For graphical representation of concentration of volatile hydrocarbon, it has to obtain a continuous representation of the contamination of the site. Therefore, the used interpolative methods examined for this project are the IDW(inverse Distance Weighting) and kriging method. In the results, in summary, all two different methods can be used to quantify exposures at a particular source area, and thus provide, a solid foundation for making risk-based decisions. All the calculations can be performed using Excel's built-in functions, and the capabilities of geospatial analysis allow the results to be displayed visually. However, anyone who uses these methods should understand all of the assumptions and limitation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권7호
• /
• pp.2390-2406
• /
• 2022

Scene text recognition has important application value and attracted the interest of plenty of researchers. At present, many methods have achieved good results, but most of the existing approaches attempt to improve the performance of scene text recognition from the image level. They have a good effect on reading regular scene texts. However, there are still many obstacles to recognizing text on low-quality images such as curved, occlusion, and blur. This exacerbates the difficulty of feature extraction because the image quality is uneven. In addition, the results of model testing are highly dependent on training data, so there is still room for improvement in scene text recognition methods. In this work, we present a natural scene text recognizer to improve the recognition performance from the feature level, which contains feature representation and feature enhancement. In terms of feature representation, we propose an efficient feature extractor combined with Representative Batch Normalization and ResNet. It reduces the dependence of the model on training data and improves the feature representation ability of different instances. In terms of feature enhancement, we use a feature enhancement network to expand the receptive field of feature maps, so that feature maps contain rich feature information. Enhanced feature representation capability helps to improve the recognition performance of the model. We conducted experiments on 7 benchmarks, which shows that this method is highly competitive in recognizing both regular and irregular texts. The method achieved top1 recognition accuracy on four benchmarks of IC03, IC13, IC15, and SVTP.

• 아동학회지
• /
• 제38권1호
• /
• pp.17-31
• /
• 2017

Objective: The purpose of this research was to examine the effects of parents' childhood attachment representations and parenting behavior in developing early childhood self-regulation, a developmental skill. Methods: This research was conducted with 171 preschoolers, 171 parent couples, and 22 teachers of 5-year-old classes in kindergartens and children's houses in Seoul. Results: First, there was significant correlation among parental childhood attachment representations, parenting behavior, and child self-regulation. Second, parental attachment representations and parenting behavior were shown to affect self-monitoring, a subvariable of self-regulation, and were influenced by maternal independence-oriented parenting behavior, maternal attachment representation, and parental attachment representation. As factors affecting self-control, a subvariable of self-regulation, they were influenced by maternal attachment representation, and maternal and paternal affectionate parenting behavior. Lastly, as factors affecting self-control, they were influenced by attachment representation to parents of origin, maternal affectionate parenting behavior, and maternal independence-oriented parenting behavior. Conclusion: This research revealed that parental childhood attachment representations and parenting behavior are important variables affecting the development of self-regulation in preschoolers. This finding can be used as basic data for parent education content to help preschoolers grow healthier and happier and as basic data for a program to improve parent-child attachment.

• International journal of advanced smart convergence
• /
• 제8권1호
• /
• pp.141-146
• /
• 2019

These days, there are many applications using neural networks as parts of their system. On the other hand, adversarial examples have become an important issue concerining the security of neural networks. A classifier in neural networks can be fooled and make it miss-classified by adversarial examples. There are many research to encounter adversarial examples by using denoising methods. Some of them using GAN (Generative Adversarial Network) in order to remove adversarial noise from input images. By producing an image from generator network that is close enough to the original clean image, the adversarial examples effects can be reduced. However, there is a chance when adversarial noise can survive the approximation process because it is not like a normal noise. In this chance, we propose a research that utilizes high-level representation in the classifier by combining GAN network with a trained U-Net network. This approach focuses on minimizing the loss function on high representation terms, in order to minimize the difference between the high representation level of the clean data and the approximated output of the noisy data in the training dataset. Furthermore, the generated output is checked whether it shows minimum error compared to true label or not. U-Net network is trained with true label to make sure the generated output gives minimum error in the end. At last, the remaining adversarial noise that still exist after low-level approximation can be removed with the U-Net, because of the minimization on high representation terms.