• 한국음향학회지
• /
• 제19권6호
• /
• pp.41-46
• /
• 2000

본 논문에서는 자바를 사용하여 연속분포 HMM 알고리즘과 Browser-embedded 모델로 음성인식시스템을 구현하였다. 이 시스템은 웹상에서 음성분석, 처리, 인식과정을 실행할 수 있도록 설계되었으며, 클라이언트에서는 자바애플릿을 이용하여 음성의 끝점검출과 MFCC와 에너지 그리고 델타계수들을 추출하여 소켓을 통해 서버로 전송하고, 서버는 HMM 인식기와 학습DB를 이용하여 인식을 수행하고 인식된 결과는 클라이언트에 전송되어 문자로 출력되어진다. 또한 이 시스템은 플랫폼에 독립적인 시스템으로 네트웍상에서 구축되었기 때문에 높은 에러율을 갖고 있지만 멀티미디어 분야에 접목시켰다는 의의와 향후에 새로운 정보통신 서비스가 될 가능성이 있음을 알 수 있었다.

• 한국산학기술학회논문지
• /
• 제14권12호
• /
• pp.6478-6483
• /
• 2013

본 연구에서는 MIMO 환경 하에서 모든 개별 채널을 추정하기 위해 다수의 사운딩 버스트 사이의 보간 기법을 이용한 아이디어를 제시한다. 제안된 기법의 성능은 $f_dT$ 값과 송수신단의 안테나 개수에 크게 의존하고 있으며 특히 본 연구를 통하여 제안된 기법은 $f_dT$ 값이 아주 크지 않거나 또는 안테나의 개수가 많지 않을 때 효과적임을 보인다. 또한 우리는 추정 채널의 개수와 안테나의 개수가 클 때는 사운딩 버스트들이 시간상에서 더 멀리 퍼져 있기 때문에 채널의 도플러 현상이 채널에 더 많은 오류를 유발할 수밖에 없으며 이로 인해 최상과 최악의 채널 추정 성능 차이가 더욱 커짐을 보인다.

• 제어로봇시스템학회논문지
• /
• 제21권5호
• /
• pp.471-476
• /
• 2015

This paper presents a maximum velocity trajectory planning algorithm for differential mobile robots with wheel velocity constraint to cope with physical limits in the joint space for two-wheeled mobile robots (TMR). In previous research, the convolution operator was able to generate a central velocity that deals with the physical constraints of a mobile robot while considering the heading angles along a smooth curve in terms of time-dependent parameter. However, the velocity could not track the predefined path. An algorithm is proposed to compensate an error that occurs between the actual and driven distance by the velocity of the center of a TMR within a sampling time. The velocity commands in Cartesian space are also converted to actuator commands to drive two wheels. In the case that the actuator commands exceed the maximum velocity the trajectory is redeveloped with the compensated center velocity. The new center velocity is obtained according to the curvature of the path to provide a maximum allowable velocity meaning a time-optimal trajectory. The effectiveness of the algorithm is shown through numerical examples.

• 한국의류학회지
• /
• 제42권1호
• /
• pp.133-147
• /
• 2018

This study examines the experience structure and success theme in the business experience of successful young entrepreneurs in the fashion field. The study was conducted as a qualitative case study. The participants were three CEOs with more than 10 years fashion businesses experience with small and medium enterprises. Data were collected from July 2016 to June 2017 through in-depth interviews in Seoul and Gwangju as well as literature collected from other journals, newspaper articles, and books. Collected data were first summarized in 284 meaning units and then reduced to 53 common meanings through description, analysis, and interpretation processes that produced 16 sub-themes. Finally, six successful themes were extracted. The study results are as follows. First, the experience structure for successful entrepreneurs is formed by the qualitative difference of experience according to time order. Second, there are six successful themes found in the fashion start-up process. 1. Developing talent as entrepreneurs with a sense of fashion and enthusiasm 2. Passing on trial and error as a chance to communicate with customers 3. Challenge the fashion retail market with global online start-up opportunities 4. Building a system that responds to market environment changes 5. Leading and distinguishing with expertise from entrepreneurs 6. Evolve into the coexistence of members.

• 스마트미디어저널
• /
• 제9권1호
• /
• pp.45-50
• /
• 2020

딥 러닝의 발전에 따라 추천시스템에서 딥 러닝 기반의 인공신경망을 활용한 연구가 활발히 진행되고 있다. 특히, RNN(Recurrent Neural Network)기반의 추천시스템은 데이터의 순차적 특성을 고려하기 때문에 추천시스템에서 좋은 성과를 보여주고 있다. 본 연구는 RNN기반의 알고리즘인 GRU(Gated Recurrent Unit)와 세션 기반 병렬 미니배치(Session Parallel mini-batch)기법을 활용한 여행경로 추천 시스템을 제안한다. 본 연구는 top1과 bpr(Bayesian personalized ranking) 오차함수의 앙상블을 통해 추천 성과를 향상시켰다. 또한, 데이터 내에 순차적인 특성을 고려한 RNN기반 추천 시스템은 여행경로에 내재된 여행지의 의미가 반영된 추천이 이루어진다는 것을 확인되었다.

• 한국콘크리트학회:학술대회논문집
• /
• 한국콘크리트학회 2008년도 춘계 학술발표회 제20권1호
• /
• pp.605-608
• /
• 2008

콘크리트용 혼화재로 사용량이 날로 증대되고 있는 고로슬래그 미분말의 품질 차이가 콘크리트의 중성화에 어떠한 영향을 미치는지를 통계적으로 분석해 보았다. 현행 KS F 2563에 의하면 고로슬래그 미분말을 1종, 2종, 3종으로 구분하는데 3종에 해당하는 분말도 $4000cm^2/g$급 고로슬래그 미분말을 3곳으로부터 입수하여 Type A, B, C로 하여 다른 모든 조건은 동일하게 하고 혼입하였다. 본 연구에서는 기존의 그래프간의 단순비교를 통해 공학적 판단을 하는 오류를 배제하고자 평균값과 분산을 동시에 고려하는 통계적인 기법을 적용하였다. 분석기법에는 A, B, C 3개 집단간의 F검정과 각각의 쌍체비교를 위해 T검정 기법을 적용하여 통계적인 의미를 고찰해 보았다. 그 결과 유의수준 5%에서 중성화에 미치는 영향이 통계적으로 유의미한 것으로 나타났다.

• 한국어교육
• /
• 제25권1호
• /
• pp.203-234
• /
• 2014

This paper demonstrates how to incorporate the teaching of communication strategies into a large class of English-speaking learners of the Korean language. The method proposed here was developed to overcome the difficulty of conducting language activities involving communicative interactions amongst students and also between teacher and students in a large classroom. As a way of compensating the minimal opportunities for interactions in the classroom, students are given the task of expressing in Korean the English translations of authentic Korean comics via Twitter, which was later replaced with the feedback feature on Moodle, and then their Korean expressions are collected and projected onto a big screen. These collected expressions by students naturally differ from one another, helping students to realize that it is possible for them to express the same message or meaning in many different ways. The results of two separately conducted questionnaires show that this method is an effective way of providing students with significantly increased chances of producing 'comprehensible output' that requires them to think of how to communicate with their limited knowledge of the Korean language. Many students also commented that the teachers' feedback on errors provides them with the opportunity to learn about common errors as well as their own errors.

• 한국의류학회지
• /
• 제33권7호
• /
• pp.1109-1120
• /
• 2009

This study develops a fitted torso type basic pattern for men by utilizing 3D body scan data. Recent fashion trends are reflected in the development of the pattern. The subjects were 15 men in their 20's, who wear size 95 (M size). Body scan data was obtained through a 3D whole body scanner (WB4, Cyberware, USA), and a body surface development figure for developing male fitted torso type basic pattern was attained through the use of Rapid Form 2006 as well as Auto CAD 2006 programs. The results are as follows: A body surface development figure through body surface segment method showed high exactitude in an error range of 100$\pm$1%. In addition, it occurred in an error range of 100:1:3% because of the hard scanning conditions in the incline of the shoulder and armpit areas. However, the body surface development figure as well as the direct measurement results can be used as basic data for the given patternmaking since the error range falls into 100$\pm$3%. Dart amounts obtained from the average cross section were center back 2.2cm (24.3%), back armpit point 3.8cm (41.8%), front armpit point 3.0cm (33.9%). As shown the jacket pattern, the biggest dart amount was portioned out at the back armpit point. The drafting equations for the development pattern acquired are as follows; Full width=C/2+5cm, back length=height/4-1cm, armhole depth=(C/10+12cm)+3cm, back width=2C/10+2cm, front width=2C/10. The development pattern was a fitted torso basic pattern that was composed of 3 pieces, so it would be very useful in developing shirt or jacket patterns. According to the results of the evaluation of the developed pattern appearance, it obtained higher scores of over 3.5 points in almost items, meaning that the developed pattern is appropriate for a male fitted torso type basic pattern. It suggests a possibility of patternmaking from a body surface development figure in 2-D to prototype.

• 정보처리학회논문지A
• /
• 제18A권3호
• /
• pp.109-114
• /
• 2011

IR(Intermediate Representation) 최적화 과정은 컴파일러 back-end의 중요한 부분으로서 sub-expression elimination, dead code elimination 등 최적화 기법들을 사용한다. 하지만 IR 최적화 단계에서 생기는 에러들을 검출하고 디버깅하는데 많은 어려움이 있다. 그 첫 번째 이유로는 컴파일 된 어셈블리 코드를 해독하여 에러를 체크하기 어렵고 두 번째로는 IR 최적화 단계에서 에러가 생겼는지 결정 짓기 어렵기 때문이다. 이런 이유들로 인하여, 우리는 C 레벨에서 IR 코드변환 무결점 여부를 체크하기 위한 기법들에 관한 연구를 진행하여 왔다. 우리는 MeCC(Memory Comparison-based Clone) 탐색기를 기반으로 하여, 최적화하기 전 IR코드와 최적화 한 후의 IR코드를 각각 C코드로 다시 변환한 뒤, 이 두 개의 C코드를 MeCC의 입력으로 주고, 결과의 일치 여부를 확인하는 방법을 사용한다. 하지만 MeCC가 완벽한 결과를 알려주지 않기 때문에, 우리는 각 IR 최적화 기법마다의 특징에 대한 정보를 사전에 처리해서 그 결과의 정확도를 높였다. 이 논문에서는 dead code elimination, instruction scheduling 및 common sub-expression elimination 등 최적화 기법들을 이용한 변환 코드들을 예시로 실험하여 최종적으로 MeCC에서의 C 레벨 코드의 정확한 에러 체크 동작여부를 보여준다.

• 한국환경과학회지
• /
• 제10권1호
• /
• pp.21-25
• /
• 2001

The Keum river has been utilized for drinking water supply of several city including Kunsan city and is deepening pollution state due to numerous municipal and industrial discharges. The concentration BOD in river is affected by the organic loading from a tributary and the algae biomass that largely happen to under eutrophication state. In the eutrophic water mass such as the Keum river, the autochthonous BOD was very important part for making a decision of water quality management, because it was accounted for majority of the total BOD. The predict of water quality has important meaning for management of water quality pollution of the Keum river. The purpose of this study will manage and predict water quality of the Keum river using QUAL-2E model considering the autochthonous BOD. The estimation of autochthonous BOD represented that the relationship between BOD and chlorophyll a. The regression equation was shown to be autochthonous BOD=$\beta$(sub)5$\times$chlorophyll a. The results of this study may be summarized as followed; The QUAL-2E model was calibrated with the data surveyed in the field of the study area in June, 1998. The calculated value by QUAL-2E model are in good agree to measured value within relative error of 7.80~20.33%. Especially, in the case of the considering autochthonous BOD, the calculated value of BOD were fairly good coincided with the observed values within relative error of 15%. But the case of not considering autochthonous BOD, relative error of BOD was shown to be 43.2%. In order to attain II grade of water quality standard in Puyo station which has a intake facility of water supply, we reduced to the pollutants loading of tributaries. In the case of removed 100% BOD of tributaries, the BOD of Puyo station was 4.07mg/$\ell$, belong to III grade of water quality standard. But in the case of removed 88% nutrient of tributaries, it was satisfied to II grade of water quality standard as below 3mg/$\ell$ of BOD. For estimation of autochthonous BOD in Keum river, we are performed simulating in accordance with reduction of nutrient load(50~100%) under conditions removal 90% organic load. Occupancy of autochthonous BOD according to nutrient loading reductions were varied from 25.97~79.51%. Occupancy of autochthonous BOD was shown to be a tendency to increasing in accordance with reduction of nutrient loading. Showing the above results, the nutrient that one of the growing factor of algae was important role in decision of BOD in the Keum river. For the water quality management of the Keum river, therefore, it is necessary to considering autochthonous BOD and to construction of advanced sewage treatment plant for nutrient removal.