http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Fast offline transformer-based end-to-end automatic speech recognition for real-world applications
오유리,박기영,박전규 한국전자통신연구원 2022 ETRI Journal Vol.44 No.3
With the recent advances in technology, automatic speech recognition (ASR) has been widely used in real-world applications. The efficiency of converting large amounts of speech into text accurately with limited resources has become more vital than ever. In this study, we propose a method to rapidly recognize a large speech database via a transformer-based end-to-end model. Transformers have improved the state-of-the-art performance in many fields. However, they are not easy to use for long sequences. In this study, various techniques to accelerate the recognition of real-world speeches are proposed and tested, including decoding via multiple-utterance-batched beam search, detecting end of speech based on a connectionist temporal classification (CTC), restricting the CTC-prefix score, and splitting long speeches into short segments. Experiments are conducted with the Librispeech dataset and the real-world Korean ASR tasks to verify the proposed methods. From the experiments, the proposed system can convert 8 h of speeches spoken at real-world meetings into text in less than 3 min with a 10.73% character error rate, which is 27.1% relatively lower than that of conventional systems.
포스트 코로나 시대 호텔 비대면 서비스디자인 전략 제안- 카노모델과 서비스 블루프린트를 중심으로 -
오유리,구 유 리,강채민,김미경,조재희 사단법인 한국브랜드디자인학회 2023 브랜드디자인학연구 Vol.21 No.1
The purpose of this study is to propose effective services for each itinerary by analyzing the non-face-to-face services of the current hotel industry among the non-face-to-face services applied to the overall tourism industry after COVID-19. hrough prior research, five experience elements, including convenience, functionality, usability, security, and playfulness, were selected. Through pilot research and customer journey map analysis, 34 hotel non-face-to-face service quality attributes were derived, and the Kano model and Timko customer It was analyzed using the satisfaction coefficient. As a result of the analysis, 'functionality' was the most effective factor among experience factors, and specifically, quality attributes such as use of auxiliary facilities, provision of waiting status information, and carrying luggage by robot during check-in and check-out were included. There was a significant difference according to the age and gender of customers. Those in their 20s were positive about active services that grasp information such as facility congestion and usage status, while those in their 50s were not good at non-face-to-face services, so they were more satisfied with passive services. In addition, it was confirmed that women had twice as many items with a higher effect index than men. Therefore, the more the function-oriented non-face-to-face service was provided according to the age and gender of the customer, the more effective it was, and the service improvement and maintenance items were identified at a glance through the service blueprint.
오유리,정경아,이효진,정규열,박종문 한국생물공학회 2018 Biotechnology and Bioprocess Engineering Vol.23 No.1
Despite an increasing potential of red algal biomass as a feedstock, biological conversion of red algal biomass has been limited by lack of feasible microorganisms which can convert structured AHG, which is a main component of red algal carbohydrate, into a common metabolite. In the AHG uptake pathway, AHG dehydrogenase (AHGD) is known to be a key step, therefore it is important to find an efficient dehydrogenase to break down 3,6- anhydro-L-galactose (AHG) for practical use of red macroalgae biomass in biorefineries requires. In this study, we isolate a novel AHG dehydrogenase (AHGD) with high activity produced by a newly isolated bacteria strain, Raoultella ornithinolytica B6–JMP12. The stability and compatibility of the enzyme were evaluated under various conditions to achieve high enzyme production. The AHGD was partially purified using conventional protein purification techniques such as ammonium sulfate precipitation and ion exchange followed by gel filtration chromatography, 37.24 fold with a final specific activity of 5.47 U/mg of protein with 32% yield recovery. SDS-PAGE was used to determine the molecular weight of the partially purified AHGD and its molecular weight was found to be around ~34 kDa. The optimal pH and temperature for the partially purified AHGD were 7.0 and 35°C, respectively. The Km and Vmax for 3,6-anhydro-L-galactose are 0.63 mg/mL and 0.38 μM/mL/min, respectively.
학문과 연구에 있어서의 패러다임 변화 : 혁신적 대학 창조
유리오 소타마(Yrj? Sotamaa) 한국디자인산업연구센터 2008 IDCC Proceeding Vol.2008 No.5
“배움과 연구에서의 패러다임 쉬프트-혁신적인 대학 만들기” 점차 증가하는 세계적인 도전들은 우리가 풀어야 할 문제점의 복잡성을 증가시키고 혁신을 돌파하기 위한 요구 역시 점차 커지고 있다. 이것은 교육의 새로운 단계를 증진시키고 다 학제적인 프로세스와 리서치를 창출하기도 한다. 디자인과 디자인적인 사고는 이러한 프로세스에서 매우 중요한 역할을 한다. 유리오 소타마는 계획, 기술, 경영 그리고 디자인과의 합병을 통한 핀란드 정부와 기업, 혁신적인 대학으로부터의 지원을 통한 최근의 대규모의 계획에 대해서 설명할 것이다. 이것은 핀란드 경제의 경쟁력을 향상하기 위한 시도이며 Green Economy를 향한 전진이다. 유리오 소타마는 역시 어떻게 다 학제적인 배움의 과정이 창의력과 혁신을 양성하며 디자인의 활용도를 확장할 것인지에 대해서도 설명할 것이다. “Interdisciplinary education in design - fostering innovation and learning” at the University of Art and Design Helsinki Taik Taik has a long tradition in interdisciplinary design education in collaboration with the Helsinki University of Technology and the Helsinki Business School. This has also included training Korean Business executives in Kemba Program and training Finnish Business Executives, besides the postgraduate programs. The university has also experience of organizing interdisciplinary programs in international collaboration. Yrj? Sotamaa will discuss the learning processes and outcomes of the highly successful IDBM, HSCE and MEDES programs and?how new interdisciplinary learning processes foster creativity and innovation and expand the applications of design.
빅데이터 분석기법을 통한 N-스크린 서비스 활용에 영향을 미치는 요인 분석
오유리,김도훈 한국경영과학회 2015 한국경영과학회 학술대회논문집 Vol.2015 No.4
스마트 디바이스의 확산으로 인해 정보통신 서비스의 다양성과 양이 크게 증가하고 있다. 이에 따라 사용자의 니즈에 따른 디바이스의 선택의 폭도 넓어지고 디바이스 간 상호연동도 확장 되었다. 이러한 현상은 사용자의 콘텐츠 소비방식에 영향을 줄 것인데, 그 중심에 있는 것 중의 하나가 N-스크린 서비스이다. 본 연구에서는 KISDI에서 제공한 미디어패널 데이터를 바탕으로, N-스크린 서비스 이용에 영향을 미치는 핵심 요인들을 찾는다. 미디어 패널 데이터로부터 본 연구의 목적에서 사용가능한 7,128의 레코드를 추출하였으며, 종속변수를 포함하여 총 54개의 변수로 구성된 빅데이터를 구축하였다. 정확한 예측력을 위해 학습용 집합(train data set)과 비교용 집합(test data set)을 나누어 분석하여 비교한다. 그 전에 데이터를 분류하기 위해 K-fold Cross validation과 Bootstrap을 이용한다. 로지스틱 회귀분석(Logistic Regression, LR)을 적용한 결과, LR에서는 23개의 속성이 유의한 것으로 분석되었다. 따라서 향후 N-스크린 가입자 특성이나 행태적 요인 등을 규명하거나 이에 관련된 마케팅 전략이나 정책 개발에서 이들 요인을 적극적으로 고려할 필요가 있을 것이다.