• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.414-418
• /
• 2021

다국어 음성인식은 단일언어 음성인식에 비해 높은 난이도를 보인다. 하나의 단일 모델로 다국어 음성인식을 수행하기 위해선 다양한 언어가 공유하는 음성적 특성을 모델이 학습할 수 있도록 하여 음성인식 성능을 향상시킬 수 있다. 본 연구는 딥러닝 음성인식 모델인 Wav2Vec2.0 구조를 변경하여 한국어와 영어 음성을 하나의 모델로 학습하는 방법을 제시한다. CTC(Connectionist Temporal Classification) 손실함수를 이용하는 Wav2Vec2.0 모델의 구조에서 각 언어마다 별도의 CTC 출력 계층을 두고 각 언어별 사전(Lexicon)을 적용하여 음성 입력을 다른 언어로 혼동되는 경우를 원천적으로 방지한다. 제시한 Wav2Vec2.0 구조를 사용하여 한국어와 영어를 잘못 분류하여 음성인식률이 낮아지는 문제를 해결하고 더불어 제시된 한국어 음성 데이터셋(KsponSpeech)에서 한국어와 영어를 동시에 학습한 모델이 한국어만을 이용한 모델보다 향상된 음성 인식률을 보임을 확인하였다. 마지막으로 Prefix 디코딩을 활용하여 언어모델을 이용한 음성인식 성능 개선을 수행하였다.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2003.11a
• /
• pp.200-203
• /
• 2003

최근 아날로그 TV는 디지털 TV로 교체되고 있는 추세이다. 이는 오디오/비디오 스트림 뿐만이 아닌 데이터를 수신기에 전송할 수 있는 디지털 TV의 장점에 기인한다. 그러나 그러한 데이터를 프로세싱 할 수 있는 규격이나 규약이 존재하지 않는다. 그래서 대부분의 회사와 개발자들은 자신들의 각각의 방식으로 그 데이터를 프로세싱 하는 방법을 택하고 있다. 이러한 규격과 규약의 부재는 많은 혼동과 시간, 비용, 인력의 낭비를 초래한다. 본 논문에서는 표준적인 측면이 아닌 방법적인 측면에서 이를 해결하고자 한다. 그 해결책은 방송 콘텐츠를 생성하고 표현하는데 XML을 이용하는 것이다. 이 해결책은 많은 인력과 비용을 절약해고, 값비싼 방송 콘텐츠의 재 사용률을 높여 줄 것이다.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2004.05a
• /
• pp.459-462
• /
• 2004

최근 아날로그 TV는 디지털 TV로 교체되고 있는 추세이다. 이는 오디오/비디오 스트링 뿐만이 아닌데 데이터를 수신기에 전송할 수 있는 디지털 TV의 장점에 기인한다. 그러나 그러한 데이터를 프로세싱할 수 있는 규격이나 규약이 존재하지 않는다. 그래서 대부분의 회사와 개발자들은 자신들의 각각의 방식으로 그 데이터를 프로세싱하는 방법을 택하고 있다. 이러한 규격과 규약의 부재는 많은 혼동과 시간, 비용, 인력의 낭비를 초래한다. 본 논문에서는 표준적인 측면이 아닌 방법적인 측면에서 이를 해결하고자 한다 그 해결책은 방송 콘텐츠를 생성하고 표현하는데 XML을 이용하는 것이다. 이 해결책은 많은 인력과 비용을 절약해 주고. 값비싼 방송 콘텐츠의 재사용률을 높여줄 것이다.

• Journal of Science Education
• /
• v.33 no.1
• /
• pp.111-121
• /
• 2009

The purpose of this research is to survey elementary pre-service teachers' in understand the legal Units, focusing on seven basic unit such a 'm', 'm2', 'L', 'kg', 'K', 'cd', 's'. This study specifically investigates whether the students understand the legal units. The subjects were 1096 students from the University of Education in Jinju, Gyeongnam. Data was collected through a questionnaire which was designed by this research and checked by authority, and the frequency and percentage of responses to each question were obtained and analysed. The survey was the legal units on interesting, using the experience of confusing and understanding of elementary pre-service teachers. The Korea Government is regulating using traditional measures such as 'pyeong' or 'don' in commercial transactions change to adopt the metric system for as a subsidiary the first of July, 2007. The interesting of the legal units dose not exceed a positive answer to the question 52.1%. Their were answered that the experience of the confused of 60.1% in the life. How to do efforts for the settle down of the legal units that answered broadcasting>in class>a campaign>study and training by an academic year in oder. Findings show regardless of academic year, gender and from the department of liberal arts or the science department all the students knew very well that 'm' '$m^2$', 'L', 'kg' are included in the legal units, compared to the others low percentage of 'K', 'cd' and 's' the legal units. In case of time(s), women has correct answered 2.7 times than man. In case of academic year, except for the third-year students was not to exceed 50%. In case of from the department of liberal arts or the science department contrary to one's expectations increase of 50% or more correct answer while half the students scored in science. The elementary pre-service teachers are seems to thinking separate the legal units with their in university life. Also elementary pre-service teachers are the lack of interest on society. Their should be for settle down of the legal units through learning to class in university, newspapers, strengthen publicity activities of broadcast media's further more by maintenance efforts of the government.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.1
• /
• pp.6-14
• /
• 1996

In this paper we present an algorithm to discriminate an underwater target under track against an acoustic counter-measure(ACM) source, based on a sequential hypothesis test. The ACM source is separated from the target under track and generates, while drifting, measurements with false range and Doppler information. The purpose of the ACM is to mislead the target-tracking and to help the true target evade a pursuer. The algorithm uses as a test statistic a function of the innovation sequences from extended Kalman filters to estimate the target dynamics and the drifting position of the ACM source. Numerical experiments on various scenarios show that the proposed algorithm discriminates the target against an ACM source very fast with a high probability of success.

• Journal of Korea Multimedia Society
• /
• v.7 no.11
• /
• pp.1564-1570
• /
• 2004

Recently, The Analog TV has been being replaced by digital TV. This dues to the advantage of digital TV that sends data as well as audio/video stream to the receiver. But there are no standards or rules for processing the data. So, most companies or developers process the data in separate methods. This absence of standards or rules for processing the data causes many confusions and waste of time, money, man power. This Thesis proposes the solution of the problem not based on standardized respect but methodological respect. This solution is using XML in producing and playing the broadcasted multimedia contents and data. We believe that this solution saves much man power, time, money and increase reusability of expensive broadcasting contents.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.7
• /
• pp.1881-1892
• /
• 1997

This paper proposes the method for recognition of printed chinese characters by probabilistic VQ networks and multi-stage recognizer has hierarchical structure. We use modular neural networks, because it is difficult to construct a large-scale neural network. Problems in this procedure are replaced by probabilistic neural network model. And, Confused Characters which have significant ratio of miss-classification are reclassified using the entropy theory. The experimental object consists of 4,619 chinese characters within the KSC5601 code except the same shape but different code. We have 99.33% recognition rate to the training data, and 92.83% to the test data. And, the recognition speed of system is 4-5 characters per second. Then, these results demonstrate the usefulness of our work.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.8
• /
• pp.59-66
• /
• 2023

The Golden Cross is commonly seen as a buy signal in financial markets, but its reliability for predicting stock price movements is limited due to market volatility. This paper introduces a time-invariant approach that considers the Golden Cross as a singular event. Utilizing LSTM neural networks, we forecast significant stock price changes following a Golden Cross occurrence. By comparing our approach with traditional time series analysis and using a confusion matrix for classification, we demonstrate its effectiveness in predicting post-event stock price trends. To conclude, this study proposes a model with a precision of 83%. By utilizing the model, investors can alleviate potential losses, rather than making buy decisions under all circumstances following a Golden Cross event.

• Journal of The Korean Association For Science Education
• /
• v.22 no.2
• /
• pp.336-344
• /
• 2002

The purpose of this study was to analyse achievement of 'Scientific Inquiry and the Nature of Science' in the Third International Mathematics and Science Study-Repeat (TIMSS-R), which was performed in 1999 with 38 nations participating. Korean 8th grade students' achievement of 'Scientific Inquiry and the Nature of Science' was compared to that of other countries and other content areas in science. Average percent correct of items in each subcategory - Scientific Method, Experimental Design, Scientific Measurements, Describing and Interpreting Data - was also analysed. Although 'Scientific Inquiry and the Nature of Science' topics were not included in intended curriculum in Korea, Korean students' average scale score of 'Scientific Inquiry and the Nature of Science' was significantly higher than international average and, in comparison with other science content areas, achievement of that area was relatively high. The reasons could be that the most students studied topics related to 'Scientific Inquiry and the Nature of Science' through the implemented curriculum and that the Korean teachers recognized the importance of inquiry. According to the results to analyze subcategories, the average percent correct of Korea were higher than 50% except the 'Scientific Measurements' subcategory. However, the international average percent correct were lower than 50%. Especially, the average percent correct of Korea was the highest in 'Describing and Interpreting Data' subcategory despite there were many students who were confused at observation, hypothesis and conclusion.

• Education of Primary School Mathematics
• /
• v.27 no.3
• /
• pp.303-320
• /
• 2024

This study presents the development and performance evaluation of a custom GPT-based chatbot tailored to provide solutions following Polya's problem-solving stages. A beta version of the chatbot was initially deployed to assess its mathematical capabilities, followed by iterative error identification and correction, leading to the final version. The completed chatbot demonstrated an accuracy rate of approximately 89.0%, correctly solving an average of 57.8 out of 65 image-based problems from a 6th-grade elementary mathematics textbook, reflecting a 4 percentage point improvement over the beta version. For a subset of 50 problems, where images were not critical for problem resolution, the chatbot achieved an accuracy rate of approximately 91.0%, solving an average of 45.5 problems correctly. Predominant errors included problem recognition issues, particularly with complex or poorly recognizable images, along with concept confusion and comprehension errors. The custom chatbot exhibited superior mathematical performance compared to the general-purpose ChatGPT. Additionally, its solution process can be adapted to various grade levels, facilitating personalized student instruction. The ease of chatbot creation and customization underscores its potential for diverse applications in mathematics education, such as individualized teacher support and personalized student guidance.