• International Journal of Advanced Culture Technology
• /
• 제7권1호
• /
• pp.64-69
• /
• 2019

With the advent of the fourth industry, the need for office automation with automatic character recognition capabilities is increasing day by day. Therefore, in this paper, we study a character recognition algorithm that effectively recognizes a new experimental data character by using learning data characters. The proposed algorithm computes the degree of similarity that the structural regions of learning data characters match the corresponding regions of the experimental data character. It has been confirmed that satisfactory results can be obtained by selecting the learning data character with the highest degree of similarity in the matching process as the final recognition result for a given experimental data character.

• ETRI Journal
• /
• 제45권1호
• /
• pp.93-104
• /
• 2023

Coreference resolution is a task in discourse analysis that links several headwords used in any document object. We suggest pointer networks-based coreference resolution for Korean using multi-task learning (MTL) with an attention mechanism for a hierarchical structure. As Korean is a head-final language, the head can easily be found. Our model learns the distribution by referring to the same entity position and utilizes a pointer network to conduct coreference resolution depending on the input headword. As the input is a document, the input sequence is very long. Thus, the core idea is to learn the word- and sentence-level distributions in parallel with MTL, while using a shared representation to address the long sequence problem. The suggested technique is used to generate word representations for Korean based on contextual information using pre-trained language models for Korean. In the same experimental conditions, our model performed roughly 1.8% better on CoNLL F1 than previous research without hierarchical structure.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2021년도 춘계학술대회
• /
• pp.552-554
• /
• 2021

도플러 레이더 센서로 취득한 움직임 데이터를 딥러닝 알고리듬을 사용한 낙상 인식 시스템을 제안한다. 딥러닝 알고리듬중 시계열 데이터에 장점을 가지는 RNN을 사용하여 낙상 인식에 적용한다. 도플러 레이더 센서의 낙상데이터는 시계열 데이터로 시간적인 특성을 가지고 있으며 결과는 낙상인지 아닌지 만을 판단하기 때문에 RNN의 구조를 시퀀스 입력에 고정 크기를 출력하는 구조로 설계하였다.

• 로봇학회논문지
• /
• 제15권1호
• /
• pp.62-69
• /
• 2020

This paper proposes a pattern recognition and classification algorithm based on a circular structure that can reflect the characteristics of the sEMG (surface electromyogram) signal measured in the arm without putting the placement limitation of electrodes. In order to recognize the same pattern at all times despite the electrode locations, the data acquisition of the circular structure is proposed so that all sEMG channels can be connected to one another. For the performance verification of the sEMG pattern recognition and classification using the developed algorithm, several experiments are conducted. First, although there are no differences in the sEMG signals themselves, the similar patterns are much better identified in the case of the circular structure algorithm than that of conventional linear ones. Second, a comparative analysis is shown with the supervised learning schemes such as MLP, CNN, and LSTM. In the results, the classification recognition accuracy of the circular structure is above 98% in all postures. It is much higher than the results obtained when the linear structure is used. The recognition difference between the circular and linear structures was the biggest with about 4% when the MLP network was used.

• 한국초등과학교육학회지:초등과학교육
• /
• 제22권1호
• /
• pp.37-50
• /
• 2003

The purpose of this study was to develop the quality of the textbook and to find out reasonable selection and structure by examining and analyzing the cognition of teacher and students on the structure and contents of elementary science textbook in the 7th curriculum. The findings of this study were as follows: First, as a result of the students’ cognition, their interest level of the learning contents was high and the degree of the difficulty of the learning contents was low on the whole. Second, as a result of the teachers’ cognition of contents of the textbook, teachers who taught third graders understood that the third graders have relatively much contents to be studied and the level of the contents of the textbook was high. On the other hand, fourth graders’ teachers recognized that contents to be studied and the level of the contents were appropriate. And they understood that there were much work to be studied in the units which were difficult and there were difference between contents to be studied and the degree of the difficulty in some units such as life or the earth fold. Third, as a result of the teachers' cognition of structure of the textbook. teachers were very affirmative to reduce school hours. They understood that current numbers and scale of the unit were appropriate. Teachers were satisfied with the structure of elementary science textbook in the 7th curriculum on the whole.

• 한국지능시스템학회논문지
• /
• 제7권4호
• /
• pp.74-81
• /
• 1997

본 논문에서는 시그모이드 함수와 방사형 기저 함수 모두를 생성시킬 수 있는 특별한 은닉층 노드를 갖는 투영신경회로망에 대하여 알아롭고 그것을 훈련시키기 위한 진화 학습 기법을 제시한다. 제시된 기법은 신경회로망의 매개변수와 연결 가충치뿐만 아니라, 어떤 목적함수를 나타내기 위한 최적의 은닉층 노드개수 또한 구조 최적화를 위한 진화연산자를 통해 찾아낸다. 각각의 은닉층 노드의 역할은 진화를 거듭하면서 방사형 기저 함수를 나타낼지 시그모이드 함수를 나타낼지 결정된다. 알고리즘을 구현하기 위해서 투영신경회로망은 연결 고리 리스트 자료구조로 나타내었다. 모의 실험에서 기존으 오차역전파에 의한 학습과 구조 성장 방식보다 적은 노드로 투영신경회로망을 훈련시킬 수 있음을 볼수 있다.

• 전기학회논문지
• /
• 제66권7호
• /
• pp.1111-1116
• /
• 2017

In this paper, we developed bio-signal based eye tracking system using electrooculogram (EOG) and electromyogram (EMG) which measured simultaneously from same electrodes. In this system, eye gazing position can be estimated using EOG signal and we can use EMG signal at the same time for additional command control interface. For EOG signal processing, PLA algorithms are applied to reduce processing complexity but still it can guarantee less than 0.2 seconds of reaction delay time. Also, we developed dual machine learning structure and it showed robust and enhanced tracking performances. Compare to conventional EOG based eye tracking system, developed system requires relatively light hardware system specification with only two skin contact electrodes on both sides of temples and it has advantages on application to mobile equipments or wearable devices. Developed system can provide a different UX for consumers and especially it would be helpful to disabled persons with application to orthotics for those of quadriplegia or communication tools for those of intellectual disabilities.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권8호
• /
• pp.2473-2489
• /
• 2022

Urban development has brought about the increasing saturation of urban traffic demand, and traffic congestion has become the primary problem in transportation. Roads are in a state of waiting in line or even congestion, which seriously affects people's enthusiasm and efficiency of travel. This paper mainly studies the discrete domain path planning method based on the flow data. Taking the traffic flow data based on the highway network structure as the research object, this paper uses the deep learning theory technology to complete the path weight determination process, optimizes the path planning algorithm, realizes the vehicle path planning application for the expressway, and carries on the deployment operation in the highway company. The path topology is constructed to transform the actual road information into abstract space that the machine can understand. An appropriate data structure is used for storage, and a path topology based on the modeling background of expressway is constructed to realize the mutual mapping between the two. Experiments show that the proposed method can further reduce the interpolation error, and the interpolation error in the case of random missing is smaller than that in the other two missing modes. In order to improve the real-time performance of vehicle path planning, the association features are selected, the path weights are calculated comprehensively, and the traditional path planning algorithm structure is optimized. It is of great significance for the sustainable development of cities.

• 마케팅과학연구
• /
• 제10권
• /
• pp.139-164
• /
• 2002

90년 이후 현재까지 10여년 이상에 걸쳐 마케팅 개념의 구체적인 활동, 이른바 시장지향성과 관련개념에 대한 활발한 연구가 진행되어 오고 있다. 시장지향성은 고객지향에 초점을 둔 전사적 활동을 통하여 시장정보의 창출, 전파 및 활용과 경쟁자지향으로 요약될 수 있고 수익성에 직접적인 영향을 미치는 중요한 개념이다. 이러한 시장지향적 활동과 직접적인 관련이 있는 조직학습 및 조직몰입 역시 시장지향성과 함께 주요 연구대상이다. 본 연구에서는 조직몰입, 시장지향성, 조직학습의 문헌적 고찰을 통해 과거 연구들을 정리하고 이를 통해 아직까지 그 구조가 명확히 밝혀지지 않은 시장지향성 및 시장지향성과 조직학습간의 영향관계를 실증적으로 분석하였다. 분석 결과 시장지향성은 조직학습의 매우 밀접하게 관련있는 영향요인이며 조직몰입은 이러한 개념들의 선행적인 영향을 미치는 것으로 나타났다.

• 한국통신학회논문지
• /
• 제16권4호
• /
• pp.350-357
• /
• 1991

본 논문에서는 전화번호 서비스시 사용되고 있는 영(zero)에서 일까지의 2종류의 숫자음(한글발음의 셈수와 한자발음의 읽음수) 22개에 대하여 신경회로망을 이용한 음성인식 실험의 결과와 학습과정에서 나타난 제 현상에 관해 논하였다. 신경회로망은 입력단과 출력단만을 갖는 2단구조와 한 개의 은익단을 갖는 3단구조의 회로망으로 은익단의 뉴론(Neuron) 수를 11, 12 및 44개로 가변해 가면서 BP(Back-Propagation) 알고리즘에 의하여 학습하였고 학습과정에서는 학습팩터(Learning factor), 학습방법(예로써 Random or Cycle), 모멘텀(Momentum)등을 조정해 가면서 최적의 학습과정을 찾고자 하였다. 실험결과 2단구조에 의한 화자독립의 경우 최고 96%의 인식율을 나타냈고 학습과정이 너무 많을 경우 오히려 인식율이 낮아졌으며 이 현상은 3단구조의 회로망에서 더욱 두드러지게 나타났다.