• 방송공학회논문지
• /
• 제25권5호
• /
• pp.742-749
• /
• 2020

본 논문에서는 하나의 공용 신경망을 사용하여 음소와 감정을 모두 인식하는 방법과 공용 신경망 학습을 위한 다중 학습 방법을 제안한다. 공용 신경망은 동일한 동작을 수행하여 두 정보를 모두 인식하며, 이는 인간이 하나의 청각기관으로 여러 정보를 동시에 인식하는 구조에 해당한다. 다중 학습은 여러 정보를 위한 공통 모델링을 진행하므로 여러 정보에 대한 일반화된 학습을 진행시켜 기존의 정보별 개별 학습에서 나타나는 과적합을 감소시키고 인식 성능을 향상시킨다. 또한, 다중 학습에서 음소 인식에 가중치를 부여하여 음소 인식 성능을 추가 향상시키는 방법을 제안한다. 동일한 특성벡터와 신경망을 사용할 때, 제안한 다중 학습이 적용된 공용 신경망의 성능이 각 정보별로 학습시킨 개별 신경망에 비하여 우수한 것을 확인하였다.

• 한국응급구조학회지
• /
• 제25권2호
• /
• pp.55-69
• /
• 2021

Purpose: This study aimed to develop 360° virtual reality videos and common videos for cardiac arrest recognition. Methods: A sequential mixed methods study was conducted, using qualitative interviews (with a total of seven experts) and a quantitative survey. Results: First, the cardiac arrest situation should be developed within 2-3 minutes after witnessing the patient collapse, for each of the four cardiac arrest videos about apnea, gasping, seizure, gasping, and seizure. Second, the education program should be designed so that the education on cardiac arrest recognition can proceed before CPR practice begins. Conclusion: 360° virtual reality videos and common videos for cardiac arrest recognition can play an important role in the education of the general public about cardiac arrest experience.

• 한국통신학회논문지
• /
• 제33권5C호
• /
• pp.371-377
• /
• 2008

본 논문은 비전에 기반한 사람의 양팔 제스처의 모델링과 인식에 관한 연구이다. 우리는 양팔 제스처 인식을 위한 특징점의 추출에서부터 제스처의 분류에 이르는 전체적 틀을 제안하였다. 먼저, 양팔 제스처의 모델링을 위해 색채 기반의 양손 추적 방법을 제안하였고, 추출된 양손의 궤적 정보를 효과적으로 선택하게 하는 제스처 구(Phrase) 분석법을 제시하였다. 선택된 특징 점들의 시퀀스(sequence) 들로 이루어진 훈련 데이터들의 최대 공통부열(Longest Common Subsequence) 정보를 이용하여 제스처를 모델링하고 이에 따른 유사도 척도를 제안하였다. 제안된 방법론을 공항 등에서 이용하는 항공기 유도 수신호에 적용하였고, 실험을 통해 제안된 방법론의 효율성과 인식성능을 보였다.

• 한국식품조리과학회지
• /
• 제26권5호
• /
• pp.521-529
• /
• 2010

The purpose of this study was to research healthy food behavior and food recognition for each Asian country after subjects had visited Asian restaurants. The subjects of the study were university students from Griffith university and Queensland university, Australia. The survey was conducted from June 1 to 28, 2010. The summary of the analysis is as follows. Firstly, for dietary behavior related to healthy food, 'average' was the most common answer at 41.0% (102 respondents). Regarding the standard of selecting healthy food, 'if it is good for health' was the most common answer, regarding the reasons to like healthy food, 'because it is good for health', was the most common, and for information about healthy food, 'obtain from TV or media' was the most common. Regarding eating healthy food at home or dining out, most respondents answered 'once or twice a week', whereas regarding thinking of eating healthy food while dining out, 'average' was the most common answer. Secondly, the recognition of six Asian cuisines were ranked in the order of Chinese, Japanese, Korean, Thai, Indian, and Vietnamese. Representative well-being food by country, Bibimbap of Korea, Sushi of Japan, Shark's Fin of China, Tom Yum Kung of Thailand, Curry of India and Goi Cuon of Vietnam were selected. Thirdly, regarding recognition of well-being food, disease effect factor, health-oriented factor, nutrition factor and vegetarian diet factor were extracted. We found that disease effect factor and nutrition factor had positive (+) effects on visiting Asian restaurants due to recognition of well-being foods. Therefore, it is expected that more local people will eat at Asian restaurants if the public relations for Asian restaurants emphasizes harmony between well-being food and Asian food.

• 한국중재학회지:중재연구
• /
• 제31권3호
• /
• pp.3-23
• /
• 2021

England is a significant base for international trade in Europe, and dispute resolution through arbitration is active. Therefore, due to the geographical relationship with the European continent, the settlement of trade transactions and disputes with European countries is one of the most essential tasks. In this regard, arbitration procedures in England have been actively used for a long time. In England, dispute resolution methods through arbitration have been developed centered on merchant groups such as guilds from the 16th century and have been actively used until today. However, the arbitration procedure also had the characteristics of the common law because there was no legislation related to arbitration. Therefore, arbitration based on common law was carried out until the first half of the 19th century. In the 'Arbitration Act 1889', two types of arbitration systems, 'common law arbitration' and 'statutory arbitration' coexisted. However, in the arbitration procedure, according to the newly enacted 'Arbitration Act 1889', the arbitration agreement was binding from the time the arbitration agreement was reached. There was a way to select an arbitrator even if it was not explicitly stipulated in the arbitration agreement, and the arbitration award was quickly enforced. Arbitration under contract was preferred over common law arbitration, where withdrawal and revocation of awards were possible. However, in response to these provisions, the England courts considered the arbitration system to deprive the courts of jurisdiction, while a strengthened judicial review of arbitration procedures was done. In particular, England unified the arbitration-related laws, which had been scattered for a long time, adopted the model law, and enacted the 'Arbitration Act 1996'. Under the recognition and enforcement of arbitral awards in 'Arbitration Act 1996', Section 66 deals with the recognition and enforcement of arbitral awards and foreign arbitral awards. Section 2 of the 'Arbitration Act 1950' is inherited and used as it is. Second, it deals with the execution of arbitral awards under the New York Convention: Article 100 (New York Convention), Section 101 (Approval and Enforcement of Awards), Section 102 (Evidence Presented by a Party Seeking Recognition and Enforcement), and Section 103 (Provides Matters Concerning Rejection Recognition and Enforcement).

• 반도체디스플레이기술학회지
• /
• 제18권4호
• /
• pp.57-61
• /
• 2019

Nowadays, portable digital cameras such as smart phone cameras are being popularly used for entertainment and visual information recording. Given a database of geo-tagged images, a visual location recognition system can determine the place depicted in a query photo. One of the most common visual location recognition approaches is the bag-of-words method where local image features are clustered into visual words. In this paper, we propose a new bag-of-words-based visual location recognition algorithm using time-series streetview database. The proposed algorithm selects only a small subset of image features which will be used in image retrieval process. By reducing the number of features to be used, the proposed algorithm can reduce the memory requirement of the image database and accelerate the retrieval process.

• 대한음성학회지:말소리
• /
• 제46호
• /
• pp.145-157
• /
• 2003

In this paper, we'd like to briefly introduce Korean common speech database, which project has been started to construct a large scaled speech database since 2002. The project aims at supporting the R&D environment of the speech technology for industries. It encourages domestic speech industries and activates speech technology domestic market. In the first year, the resulting common speech database consists of 25 kinds of databases considering various recording conditions such as telephone, PC, VoIP etc. The speech database will be widely used for speech recognition, speech synthesis, and speaker identification. On the other hand, although the database was originally corrected by manual, still it retains unknown errors and human errors. So, in order to minimize the errors in the database, we tried to find the errors based on the recognition errors and classify several kinds of errors. To be more effective than typical recognition technique, we will develop the automatic error detection method. In the future, we will try to construct new databases reflecting the needs of companies and universities.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 1996년도 10월 학술대회지
• /
• pp.464-467
• /
• 1996

This paper aims to exploit inter/intra-speaker phoneme sub-class variations as criteria for adaptation in a phoneme recognition system based on a novel neural network architecture. Using a subcluster neural network design based on the One-Class-in-One-Network (OCON) feed forward subnets, similar to those proposed by Kung (2) and Jou (1), joined by a common front-end layer. the idea is to adapt only the neurons within the common front-end layer of the network. Consequently resulting in an adaptation which can be concentrated primarily on the speakers vocal characteristics. Since the adaptation occurs in an area common to all classes, convergence on a single class will improve the recognition of the remaining classes in the network. Results show that adaptation towards a phoneme, in the vowel sub-class, for speakers MDABO and MWBTO Improve the recognition of remaining vowel sub-class phonemes from the same speaker

• 대한전자공학회논문지TE
• /
• 제42권1호
• /
• pp.13-20
• /
• 2005

본 논문에서는 M. Bilginer 등이 제안한 CVEM(common vector extraction method)을 이용하여 한국어 화자독립 고립단어 인식실험을 수행하였다. CVEM은 학습용 음성신호들로부터 공통된 특징의 추출이 비교적 간단하고, 많은 계산 량을 필요로 하지 않을 뿐만 아니라 높은 인식 결과를 보여주는 알고리즘이다. 그러나 학습 음성의 개수를 일정 한도 이상으로 늘릴 수 없고, 추출된 공통벡터들 간의 구별정보(discriminant information)를 가지고 있지 않다는 문제점을 가지고 있다. 임의의 음성군으로부터 최적의 공통벡터를 추출하기 위해서는 다양한 음성들을 학습에 사용해야만 하는데 CVEM은 학습용 음성 개수에 제한이 있으므로 지속적인 인식률 향상을 기대하기 어렵다. 또한 공통벡터들 간의 구별정보 부재는 단어 결정에 있어서 치명적인 오류의 원인이 될 수 있다. 본 논문에서는 CVEM이 가지고 있는 이러한 문제점들을 보완하면서 인식률을 향상시킬 수 있는 새로운 방법인 KSCM(k-clustering subspace method)과 DCVEM(discriminant common vector extraction method)을 제안하였고 이 방법을 사용하여 고립단어를 인식하였다. 그리고 제안한 방법들의 우수성을 입증하기 위해 ETRI에서 제작한 음성 데이터베이스를 사용, 다양한 방법으로 실험을 수행하였다. 실험 결과 기존 방법의 문제점들을 모두 극복할 수 있었을 뿐 아니라 기존에 비해 계산량의 큰 증가 없이 향상된 결과를 얻을 수 있었다.

• 대한음성학회지:말소리
• /
• 제43호
• /
• pp.89-99
• /
• 2002

Recently, researches into ways of improving large vocabulary continuous speech recognition and speech synthesis are being carried out intensively as the field of speech information technology is progressing rapidly. In the field of speech recognition, developments of stochastic methods such as HMM require large amount of speech data for training, and also in the field of speech synthesis, recent practices show that synthesis of better quality can be produced by selecting and connecting only the variable size of speech data from the large amount of speech data. In this paper we design and discuss linguistic contents for speech copora for speech recognition and synthesis to be shared in common.