• 대한음성학회지:말소리
• /
• 제56호
• /
• pp.207-223
• /
• 2005

The goal of language model adaptation is to improve the background language model with a relatively small adaptation corpus. This study presents a language model adaptation technique where additional text data for the adaptation do not exist. We propose the information retrieval (IR) technique with N-gram language modeling to collect the adaptation corpus from baseline text data. We also propose to use a dynamic language model interpolation coefficient to combine the background language model and the adapted language model. The interpolation coefficient is estimated from the word hypotheses obtained by segmenting the input speech data reserved for held-out validation data. This allows the final adapted model to improve the performance of the background model consistently The proposed approach reduces the word error rate by $13.6\%$ relative to baseline 4-gram for two-hour broadcast news speech recognition.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.120-125
• /
• 2004

In the rolling of steel or non-steel metal the most important quality aspect are thickness and flatness. In thickness, there are two important factors. One of them is getting close with accurate goal, nominal gauge, the other is minimize gauge bandwidth, the variation in gauge. In this thesis, we proposed the fuzzy model AGC to minimize gauge variation along the length, developed the rolling mill dynamic model using the math mode of the rolling mill process and the rolling model related with the variety character of the rolling material. We compared the gauge control efficiency of fuzzy model AGC and PI mass flow AGC. We have got a simulation result, that the exit gauge variation of PI mass flow AGC was 2 micron and fuzzy model AGC was 1.2 micron at 1200mpm of rolling speed when each controller was rolling 5 micron of material that is the entry gauge variation.

• 한국간호교육학회지
• /
• 제11권2호
• /
• pp.127-134
• /
• 2005

Purpose: The purpose of the study was to develop a cooperative clinical nursing education model in response to the challenge to provide quality clinical instruction for nursing students. Method: A hypothesized model was developed based on literature review and Cho's partnership model of preservice teacher. Final model was refined with cooperative committee meetings, workshop and post clinical education meetings. Results: A Cooperative clinical nursing education model was developed with three phases(collaboration, planning, redefinition of major roles) and ten constructs(organization of cooperative committee, goal setting, partnership contract, planning objectives, and operating manual). Conclusion: The Cooperative clinical nursing education model support the need for continued collaborative partnership between nursing college and hospitals to foster quality clinical instruction.

• 한국경영과학회:학술대회논문집
• /
• 한국경영과학회 2008년도 추계학술대회 및 정기총회
• /
• pp.8-23
• /
• 2008

The purpose of this study is to develop a Cross Impact System Alternatives Tree(CISAT) model necessary for selecting the interdependent R&D planning system alternatives. This model modifies System Alternatives Tree(SAT) model developed by Kwon et al.. The SAT model is composed of several functions necessary for the achievement of a final goal and several subsystems for satisfying each function. In case that the relationship of technology alternatives is interdependent, this model overlooked a relative importance derived from occurrence or nonoccurrence of the technology alternatives in the future time varience. However, a complex evaluation process within the ballot system is another disadvantage of the SAT. To solve such problems, the Cross Impact Analysis(CIA) model is applied in the SAT model so as to consider the cross impact among interdependent system alternatives.

• 공학교육연구
• /
• 제18권6호
• /
• pp.24-31
• /
• 2015

The purpose of this study is to develop the model of course evaluation design for CEA(Course-Embedded Assessment), which is focused on course level for assessment reliability. This study was conducted by the process of model development and validation according to Richey & Klein(2007)'s model development methodology. The model of this study was comprised with the three steps of learning objectives based on course outcome, assessment design and goal achievement analysis of course, which were explained by specific design principles. Also, this model was validated by two steps, which one was on the inner validation and the other was the outer validation(Richey & Klein, 2007). Considerations for designing the model of course evaluation have been suggested and a few implications were discussed.

• Journal of Electrical Engineering and Technology
• /
• 제9권1호
• /
• pp.154-161
• /
• 2014

To study and locate partial discharge(PD) and analyze the transient state of power transformer, there is a need for a high frequency model of transformer winding and calculation of its parameters. Due to the high frequency nature of partial discharge phenomenon, there is a need for an accurate model for this frequency range. To attain this goal, a Multi-Conductor Transmission Line (MTL) model is used in this paper for modeling this transformer winding. In order that the MTL model can properly simulate the transformer behavior within a frequency range it is required that its parameters be accurately calculated. In this paper, all the basic parameters of this model are calculated by the use of Finite Element Method (FEM) for a 20kV winding of a distribution transformer. The comparison of the results obtained from this model with the obtained shape of the waves by the application of PD pulse to the winding in laboratory environment shows the validity and accuracy of this model.

• 인터넷정보학회논문지
• /
• 제19권4호
• /
• pp.35-44
• /
• 2018

Monitoring through Synthesis Aperture Radar (SAR) is responsible for marine safety from floating icebergs. However, there are limits to distinguishing between icebergs and ships in SAR images. Convolutional Neural Network (CNN) is used to distinguish the iceberg from the ship. The goal of this paper is to increase the accuracy of identifying icebergs from SAR images. The metrics for performance evaluation uses the log loss. The two-layer CNN model proposed in research of C.Bentes et al.[1] is used as a benchmark model and compared with the four-layer CNN model using data augmentation. Finally, the performance of the final CNN model using the VGG-16 pre-trained model is compared with the previous model. This paper shows how to improve the benchmark model and propose the final CNN model.

• 한국의상디자인학회지
• /
• 제13권2호
• /
• pp.185-193
• /
• 2011

The purpose of this study was to examine the relationships among fashion orientation, shopping motivations, shopping flow experience, and purchases of fashion products. Two types of shopping motivations were employed in the study: goal-directed and experiential. A conceptual research model was developed using research variables and structural equational modeling was used to test hypotheses. Empirical results indicated that fashion orientation had a strong positive direct effect on both types of shopping motives. Experiential and goal-directed motivations both had significant influence on shopping flow experience by consumers. During this process the influence of experiential motivation on flow experience was stronger, which implies the importance of experiential motivation on shopping fashion products. In addition, shopping flow experience played an important role in fashion product purchasing. Results of this study suggest that retailers should take into consideration flow in the context of shopping.

• 한국정보시스템학회지:정보시스템연구
• /
• 제16권3호
• /
• pp.45-68
• /
• 2007

This study aims to explore factors of Assimilation Gap(AG) which is proposed and illustrated by Fichman and Kemerer(1997). AG is defined as the time delay between adoption and deployment and mainly studied at the organizational level. This study derives the factor affecting AG through a review of technology acceptance models and agency theory. The research model consists of three groups of independent variables(IT knowledge burden, Individual characteristics, and organization characteristics) and a moderator, which is goal incongruence. Using multiple regression analyses, four groups of hypotheses are tested with data of 221 knowledge workers from various organizations through e-mail survey. The result shows that radicalness(a dimension of knowledge burden), personal innovativeness, self-efficacy, and organizational support are statistically significant factors while complexity, experience, and perceived critical mass are not supported. The moderator effects of goal incongruence are found in personal innovativeness and organizational support. The results of this study demonstrate that agency theory is a useful perspective to deal with AG, especially in radical innovation. In addition, IT practitioners should consider not only user training but also incentives and possible organizational controls in implementing a new information technology.

• 전자공학회논문지CI
• /
• 제45권5호
• /
• pp.140-147
• /
• 2008

본 논문에서 우리는 감정 에이전트를 이용한 자동 이야기 생성 시스템을 제안한다. 제안한 시스템에서 감정 에이전트는 배우로서, 이야기 생성 시스템은 이러한 배우 및 감독 등을 이용하여 목표 지향적으로 이야기를 만들어 낸다. 자동 이야기 생성 시스템에서, 목표는 몇 가지의 상세한 계획으로 나뉘고 목표를 이루기 위한 계획을 수행함으로 이야기를 만들어 낸다. 이러한 이야기 생성 시스템은 감정 에이전트를 사용하기 때문에, 마치 감정을 소유한 인간이 연기를 하는 것처럼 효율적으로 이야기를 생성한다.