• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.40-42
• /
• 2020

본 논문에서는 합성곱 기반의 얼굴 검출기 Dual Shot Face Detector (DSFD)에 대하여, 특징점 맵의 희소화와 채널 프루닝 목적 함수를 사용하여 네트웍 경량화를 수행하였다. 특징점 맵을 희소화하기 위해 L1 목적 함수를 사용했고, 특징점 맵의 채널 프루닝을 하기 위해 채널 최대값이 가장 낮은 채널들의 합을 최소화 시키는 목적함수를 적용했다. 기존의 신경망은 특징점 맵 희소화 비율이 45%였고 두 목적 함수를 적용했을 때 69.67% 로 희소화 비율이 높아진 것을 확인했다. 얼굴 검출 성능을 다양한 조명, 크기, 환경, 각도, 표정의 얼굴들을 포함하는 영상들로 이뤄진 Wider Face 데이터 셋으로 실험한 결과, average precision은 하락 했고 easy validation set에서 0.9257, hard validation set에서 0.8363 였다.

• The Journal of the Korea Contents Association
• /
• v.18 no.8
• /
• pp.166-177
• /
• 2018

The purpose of this study is to suggest a direction for developing the Korean language textbooks for the purpose of mission. For this, three kinds of Korean language textbooks published were analyzed and evaluated. Based on the results of the analysis, this study showed to which direction the Korean language textbooks for mission should be developed. They were as follows. Frist, the balance of everyday language and bible language is important. Second, the differentiated textbooks should be published according to the different patterns of learners. Third, professionals in Korean language education and in Christian education should cooperate in the textbook development. Fourth, the professionalism for textbook development should be enhanced. Fifth, textbook development should be closely connected with teacher education.

• Journal of the Korean Society for Library and Information Science
• /
• v.32 no.1
• /
• pp.155-167
• /
• 1998

In the climate of increasing calls for academic assessment, the author undertook a study to ascertain availability of original texts and their translations in the academic library. The object of this study is to compare the use frequency of original texts by academic major of users in the university libraries. To achieve this object, the author collected the data stored in 3 Korean university library online systems from September 10th to 30th, 1995 and tested the hypotheses by using the Minitab statistical package. Libraries with multilingual collection and automated systems will find the methodology Presented here Particularly valuable.

• 전자공학회논문지 IE
• /
• v.47 no.1
• /
• pp.48-55
• /
• 2010

The goal of this paper is to determine optimal hardware platform for specific applications. In order to develop an understanding of how select the optimal platform, we focus upon the real-time embedded vehicle system for processing forward image and sound. In this paper we propose to measure parameters such as instructions, execution cycle, required memory size for program and data by using ARMulator. We have measured three types of processor cores: ARM7, ARM9 and ARM10. The results of the study indicated that the proposed methods could measure the minimal requirements of hardware platform for specific applications. By defining lower limit of hardware specifications in embedded systems, we can minimize expenses with suitable system performance without implementing the system.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2010.05a
• /
• pp.441-446
• /
• 2010

이상기후에 의한 집중호우의 증가 및 강우량의 변동 등은 단순 기상현상의 변화에서 그치지 않고 우리가 삶을 영위하는 환경에 영향을 미치고 있으나 이상기후 영향을 정량적으로 표현하는 연구는 아직 미진하다. 이러한 환경적인 변화에 대응하여 피해를 감소시키기 위해서 구조적 대책과 비구조적 대책 등이 시행되고 있다. 비구조적 대책의 대표적인 예로 우리나라에서는 풍수해보험제도가 시행중이다. 풍수해보험제도를 활성화하기 위해서는 보험 목적물인 주택, 비닐하우스, 축사 등이 어떤 피해 패턴을 보이고 또한 통계청 및 소방방재청, 국토해양부 등 유관기관에서 발표하는 계량 지표와 어떠한 관계가 있는지를 검토하는 것이 필 요하다. 본 연구에서는 이러한 유관기관의 다양한 계량자료 중에서 시군구 단위의 수방기준준수여부, 보험목적물의 피해액, 자연재해위험지구의 수와 정비율, 그리그 국가하천과 지방하천에 대한 하천정비율을 이용하여 보험목적물과의 관계를 규명하였다. 분석결과 풍수해보험목적물의 피해액과 하천정비율, 자연재해위험지구 정비율은 뚜렷하게 음의 상관성을 갖는 것으로 나타났다. 분석된 자료를 현행 보험요율과 비교하여 할인 여부를 분석한 결과 수도권 및 광역도시의 할인율이 높게 나타났고 상대적으로 충청남도와 전라북도가 낮은 할인율로 나타났다.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.71-76
• /
• 2022

목적 지향 대화 시스템은 사용자가 원하는 목적을 달성하기 위해 사용하는 시스템으로 일상 대화와 다르게 시스템이 정보를 명확히 전달하는 것이 중요하다. 따라서 최근 연구에서 목적 지향 대화 시스템을 위한 자연어 생성 모델은 정해진 대화 정책에 따라 알맞은 응답을 생성할 수 있도록 의도와 슬롯 정보를 담은 대화 행위(Dialog Act)를 활용한다. 하지만 대화 행위는 생성하는 문장을 탁월하게 제어하는 반면에 대화의 흐름과 상황에 맞게 다양한 문장을 생성하기 어렵다는 문제점을 가지고 있다. 이러한 문제점을 해소하고자 본 논문에서는 목적에 부합하는 내용을 명확하게 자연어로 생성하기 위해 대화 행위를 사용하면서 동시에 일상 대화 생성 모델과 같이 문맥을 고려하여 대화 흐름에 어울리는 자연스러운 문장을 생성할 수 있는 문맥 기반의 제어 가능한 자연어 생성 모델을 제안한다. 실험에서는 KoGPT2 사전 학습 모델과 한국어 대화 데이터셋을 사용하였으며 실험을 통해 대화 행위 기반의 자연어 생성 모델과 본 연구에서 제안한 문맥 기반의 제어 가능한 자연어 생성 모델을 비교하였다. 결과적으로 대화 행위를 단독으로 학습한 모델보다 일정 문맥을 함께 학습한 모델이 유의미한 BLEU 점수 향상을 보인다는 점을 확인하였다.

• Journal of Korea Water Resources Association
• /
• v.53 no.1
• /
• pp.45-56
• /
• 2020

This study aims to quantify the uncertainty that can be induced by the objective function when calibrating SWAT parameters using SWAT-CUP. SWAT model was constructed to estimate runoff in Naesenong-cheon, which is the one of mid-watershed in Nakdong River basin, and then automatic calibration was performed using eight objective functions (R², bR², NS, MNS, KGE, PBIAS, RSR, and SSQR). The optimum parameter sets obtained from each objective function showed different ranges, and thus the corresponding hydrologic characteristics of simulated data were also derived differently. This is because each objective function is sensitive to specific hydrologic signatures and evaluates model performance in an unique way. In other words, one objective function might be sensitive to the residual of the extreme value, so that well produce the peak value, whereas ignores the average or low flow residuals. Therefore, the hydrological similarity between the simulated and measured values was evaluated in order to select the optimum objective function. The hydrologic signatures, which include not only the magnitude, but also the ratio of the inclining and declining time in hydrograph, were defined to consider the timing of the flow occurrence, the response of watershed, and the increasing and decreasing trend. The results of evaluation were quantified by scoring method, and hence the optimal objective functions for SWAT parameter calibration were determined as MNS (342.48) and SSQR (346.45) with the highest total scores.

• The Journal of the Korea Contents Association
• /
• v.18 no.5
• /
• pp.319-329
• /
• 2018

With rapid development of technology for strong AI chatbot, the role of chatbot has been extended from conducting simple tasks to being a friend or counsellor. For this newly emerging purpose of chatbot, endowment of personality is important to make the chatbot regarded as a human being. Nevertheless I found that there are few guides about it. Thus, this study identifies the proper personality of chatbot depending on the purpose of services and user types. The purposes of chatbot services are divided into three types such as leisure-time, counselling, and task. The DISC theory is used for categorizing personality, which consists of 4 types such as dominance(D), inducement(I), submission(S), and compliance(C). An interview and survey were conducted to investigate the preferred personality of chatbot and contents for leisure-time. As results, people tend to prefer people-oriented types such as I, S for their leisure time, task-oriented types such as D,C for their task, and slow types such as C,S for counselling. Women prone to prefer neutral gender except for counselling and men tend to prefer female in all chatbot services. Preferred chatbot age is either same or younger age for leisure-time, same or older for counselling, and 30's for tasks. Preferred contents for leisure-time are mostly recent information but many 20's want fun contents and 50-70's want emphatic conversation. 30-50's want honorific but 20's and 60-70's don't care. The research results useful guide on proper personality of AI chatbot for each purpose of its service.

• Journal of Legislation Research
• /
• no.44
• /
• pp.63-101
• /
• 2013

This article firstly explores into the concepts, components, and pictures of institutional realization of competition and diversity respectively on the premise that competition and diversity comprise the primary objectives to be pursued by the broadcasting-related laws which provide the concrete measures of media policy, and argues that while the competition objective has differentiation factors, there are also particularities in the diversity value in the broadcasting-related laws as sector-specific competition laws. Then assuming that special competition rules including structural regulatory measures particularly in the broadcasting market are required in order to realize values of competition and diversity harmoniously, this article suggests the following improvement directions for regulations aimed at protection of competition and diversity in the broadcasting-related laws. The first one is with the improvement method for regulations aimed at protection of competition. Regulation on share of audience as an ex ante regulation of status and regulation on prohibited activities as an ex post regulation of conduct may play important roles in substituting the causative regulation while seeking for diversity value. For this purpose, it is needed to develop a concrete method that incorporates diversity-related factors as consideration factors in the standard for determining illegality of prohibited activities by inference to methods of determining illegality in the competition law. The second one is with the improvement method for regulations aimed at protection of diversity. This could be considered from three viewpoints that are the setting of regulatory objectives, the identification of alternative regulatory measures, and the choice of regulatory measures and levels suitable for regulatory objectives. From these viewpoints, the regulatory framework should be improved mainly with institutional measures in which diversity value is used for tools of assessment and analysis, not just remaining as mere rhetorical devices, and whether or to what extent to maintain regulations seemingly unreasonable in terms of harmonization with economic objectives such as competition should be discreetly reviewed.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.15 no.1
• /
• pp.16-27
• /
• 2016

This study identified the factors which affect pedestrian behavior using the tobit model with a descriptive analysis. Individual characteristics, household characteristics, zone characteristics were selected as the independent variables. Walking frequency and walking time was adopted for the variables which represent walking behavior. The results showed that the walking time models were superior than the walking frequency models. The model for education purpose had the most lowest fitness and the model for others purpose had characteristics mixed up the various purposes. Futhermore, this research found that independent variables: individual, household and zone characteristic variables affected walking activities of elderly pedestrian by purpose. The walking activity decresed when a pedestrian grew older or had a vehicle or a driver's license. The Elderly facilities promoted the walking activity of seniors. As the floor area of neighborhood and business facilities was wide, the walking activities were animated. In addition, the compacted downtown developments decreased the elderly's walking activities, so it is needed to consider those results in terms of the urban planners.