• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.716-721
• /
• 2023

Inverse Text Normalization(ITN) 모델은 음성 인식(STT) 엔진의 중요한 후처리 영역 중 하나이며, STT 인식 결과의 가독성을 개선한다. 최근 ITN 모델에 심층신경망을 활용한 연구가 진행되고 있다. 심층 신경망을 사용하는 대부분의 선행연구는 문장 내 변환이 필요한 부분에 토큰 태깅을 진행하는 방식이다. 그러나 이는 Out-of-vocabulary(OOV) 이슈가 있으며, 학습 데이터 구축 시 토큰 단위의 섬세한 태깅 작업이 필요하다는 한계점이 존재한다. 더불어 선행 연구에서는 STT 인식 결과를 그대로 사용하는데, 이는 띄어쓰기가 중요한 한국어 ITN 처리에 변환 성능을 보장할 수 없다. 본 연구에서는 BART 기반 생성 모델로 생성형 ITN 모델을 구축하였고, Bi-LSTM 기반 보조 신경망 모델을 결합하여 STT 인식 결과에 대한 고유명사 처리, 띄어쓰기 교정 기능을 보완한 모델을 제안한다. 또한 보조 신경망을 통해 생성 모델 처리 여부를 판단하여 평균 추론 속도를 개선하였다. 실험을 통해 두 모델의 각 정량 성능 지표에서 우수한 성능을 확인하였고 결과적으로 본 연구에서 제안하는 두 모델의 결합된 방법론의 효과성을 제시하였다.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.1
• /
• pp.271-279
• /
• 2024

This study was conducted to examine the policies and current status of career education at universities and to compile foundational data for improving career education through an analysis of undergraduate's career awareness and career education programs. The data was collected by restructuring career education-related items from a career awareness survey conducted on 1,322 enrolled students at G University. Analyzing students' career awareness based on their year and major, a cross-analysis was conducted, while the preference for career education programs was analyzed using descriptive statistics. Based on the research findings, the proposal was made to strengthen career exploration programs through self-understanding, expedite the timing of career decision-making, and highlight the necessity of developing career education programs tailored to individual career readiness levels.

• Journal of Korea Technology Innovation Society
• /
• v.17 no.4
• /
• pp.733-762
• /
• 2014

The government mandates that national research facilities & equipments through R&D business budget should be registered on the National Science and Technology Information Service (NTIS) for the purpose of the efficient use of the research facilities & equipments. This study is to contribute to the national policies on the efficient management of the research facilities & equipments by recognition methodology with the university's members and analysis of the impact factors of the universities' registration process improvement through the Define level and Measure level of the Six Sigma DAMIC. The survey and interview were conducted on research directors, professors joining university administration, graduate students, researchers, and staffs of A University. The findings are the lack of understanding specific steps and life-cycle management of research facilities & equipments. It is necessary to collect suggestions from universities and pursue policies considered the unique characteristics of the university for advanced operating and maximizing use of university's national research facilities & equipments. Research facilities & equipments enrollment compliance rate and registration accuracy were selected as CTQ-Y through the Six Sigma. 72 potential cause variables were derived through Process Map and C & E Diagram. 13 variables were determined as core potential factors through the X-Y Matrix and Pareto Chart. Research institutions should maximize utilization of research facilities & equipments through deriving a potential variables of the process improvements and designing a detail improvements based on the characteristics of each institutions.

• The KIPS Transactions:PartB
• /
• v.15B no.5
• /
• pp.477-486
• /
• 2008

Facial expression, which changes face geometry, usually has an adverse effect on the performance of a face recognition system. To improve the face recognition rate, we propose a normalization method of facial expression to diminish the difference of facial expression between probe and gallery faces. Two approaches are used to facial expression modeling and normalization from single still images using a generic facial muscle model without the need of large image databases. The first approach estimates the geometry parameters of linear muscle models to obtain a biologically inspired model of the facial expression which may be changed intuitively afterwards. The second approach uses RBF(Radial Basis Function) based interpolation and warping to normalize the facial muscle model as unexpressed face according to the given expression. As a preprocessing stage for face recognition, these approach could achieve significantly higher recognition rates than in the un-normalized case based on the eigenface approach, local binary patterns and a grey-scale correlation measure.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.8C
• /
• pp.731-740
• /
• 2012

To offer indoor location-aware services, the needs for efficient and accurate indoor localization system has been increased. In order to meet these requirement, we presented the BLIDx(Bidirectional Location ID exchange) protocol that is efficient localization system based on sensor network. The BLIDx protocol can cope with numerous mobile nodes simultaneously but the precision of the localization is too coarse because that uses cell based localization method. In this paper, in order to compensate for these disadvantage, we propose the fingerprint overlapping method by modifying a fingerprinting methods in WLAN, and localization system using proposed method was designed and implemented. Our experiments show that the proposed method is more accurate and robust to noise than fingerprinting method in WLAN. In this way, it was improved that low location precision of BLIDx protocol.

• Journal of Korean Academy of Nursing Administration
• /
• v.19 no.2
• /
• pp.254-264
• /
• 2013

연구목적: 본 연구는 간호사의 임상경력단계에 따른 경력개발제도에 대한 간호사의 인식을 파악하고, 경력개발제도에 대한 인식, 전문직 자아개념과 임파워먼트와의 관계를 파악하여 간호사 경력개발제도의 개선의 근거 마련하고 전문직 자아개념과 임파워먼트의 증대 방안 모색을 통한 인적자원관리에 기여하고자 시도 된 서술적 상관관계 연구이다. 연구방법: 서울시 소재 일 상급종합병원에서 근무하는 중환자실, 수술실, 응급실 간호사 162명을 대상으로 설문지를 이용하여 경력개발제도에 대한 인식, 전문직 자아개념, 임파워먼트를 측정하였다. 수집된 자료는 SPSS WIN 18.0 프로그램을 활용하여 서술적 통계, t-test, ANOVA, Pearson's Correlation Coefficient, Multiple linear Regression을 시행하였다. 연구결과: 간호사의 임상경력단계에 따른 경력개발제도에 대한 인식은 전임 2 간호사가 신입 간호사, 일반 간호사, 전임 1 간호사보다 높은 인식을 가지고 있었다. 경력개발제도에 대한 인식과 전문직 자아개념, 임파워먼트에는 통계적으로 유의한 양의 상관관계가 있었다. 다중회귀분석을 실시한 결과 경력개발제도에 대한 전반적 이해, 경력개발제도에 대한 기대효과, 최종학력, 임상경력단계가 전문직 자아개념의 42% 설명하는 것으로 나타났고, 경력개발제도에 대한 전반적 이해, 전문적 활동 참여에 대한 인식, 경력개발제도에 대한 기대효과, 임상경력단계가 임파워먼트를 42% 설명하였다. 결론: 전문직 자아개념과 임파워먼트에 영향을 미친 변수로 나타난 경력개발제도에 대한 인식을 향상시킬 수 있는 방안을 개발하여 적용을 통한 효과 검증이 요구되며, 간호 관리자들의 제도 운영과 관련된 장애요인의 파악 및 세심한 제도 개선이 필요하다.

• Journal of Digital Convergence
• /
• v.11 no.11
• /
• pp.367-372
• /
• 2013

In general, Levenshtein distance algorithm have a problem with not distinguish the consideration of vacabulary retrieve, because Levenshtein methode is used to vocabulary order are not defined. In this paper, we propose a improved Levenshtein methode, it effectively manage the vocabulary retrieve by frequency use of a vocabulary, and it gives the weight number which have a order between vocabularies. Therefore proposed methode have a advantage of solve the defect of perception rate in the case of increase the vocabulary, improve the recognition time become higher and it can be effectively retrieval space management.. System performance as a result of represent vocabulary dependence recognition rate of 97.81%, vocabulary independence recognition rate of 96.91% in indoor environment. Also, vocabulary dependence recognition rate of 91.11%, vocabulary independence recognition rate of 90.01% in outdoor environment.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.6
• /
• pp.774-784
• /
• 2021

The object detection and recognition process is a very important task in the field of computer vision, and related research is actively being conducted. However, in the actual object recognition process, the recognition accuracy is often degraded due to the resolution mismatch between the training image data and the test image data. To solve this problem, in this paper, we designed and developed an integrated object recognition and super-resolution framework by proposing an image super-resolution technique to improve object recognition accuracy. In detail, 11,231 license plate training images were built by ourselves through web-crawling and artificial-data-generation, and the image super-resolution artificial neural network was trained by defining an objective function to be robust to the image flip. To verify the performance of the proposed algorithm, we experimented with the trained image super-resolution and recognition on 1,999 test images, and it was confirmed that the proposed super-resolution technique has the effect of improving the accuracy of character recognition.

• Journal of Platform Technology
• /
• v.11 no.5
• /
• pp.16-22
• /
• 2023

There is a growing interest in facial age estimation because many applications require age estimation techniques from facial images. In order to estimate the exact age of a face, a technique for extracting aging features from a face image and classifying the age according to the extracted features is required. Recently, the performance of various CNN-based deep learning models has been greatly improved in the image recognition field, and various CNN-based deep learning models are being used to improve performance in the field of facial age estimation. In this paper, age estimation performance was compared by learning facial features based on various CNN-based models such as AlexNet, VGG-16, VGG-19, ResNet-18, ResNet-34, ResNet-50, ResNet-101, ResNet-152. As a result of experiment, it was confirmed that the performance of the facial age estimation models using ResNet-34 was the best.

• Journal of KIISE:Software and Applications
• /
• v.28 no.2
• /
• pp.132-140
• /
• 2001

야외에서 인지되는 자연 경치는 다양한 개체, 빛의 산란, 또는 변화를 주는 많은 요소들 때문에 컴퓨터 영상처리에서 인식하기가 쉽지 않다. 본 논문에서는 다층 인지 신경망을 이용하여 도로가 포함된 야외영상에 나타나는 개체들을 인식하는 방법을 연구하였다. 자연 영상을 영역화한 후, 각각의 영역들에 대하여 색상과 기하학적인 특성에 근거하여 특성벡터를 추출하고 이를 신경망에 입력하여 각 영역을 구분하는 2단계의 알고리듬을 제안한다. 먼저 야외 영상들을 개선된 영역 확장법과 병합과정에 의하여 개체별로 영역화하였다. 영역화된 연상은 자연 영상과 함께 영상 데이타베이스에 저장되고, 이 자료들을 이용하여 각 영역의 특성벡터를 계산하였다. 이 특성 벡터를 구성된 신경망의 입력층에 전달하면, 각 영역은 27개의 개체 중의 하나로 출력층에서 인식된다. 제안된 방법은 학습에 사용된 데이타, 학스베 사용되지 않은 새로운 데이타, 그리고 모두 합하여 놓은 데이타의 세가지 데이타 군에서 무작위로 선별하여 인식률을 측정하였다. 학습된 데이타에서는 99.4%까지의 인식률을 보여주었고, 학습되지 않은 데이타에 대해서도 최고 89.1%까지의 인식률을 나타내었다. 제안된 방법은 평균적으로 88.1%~97.9%의 인식률을 보여주어 자연 경치의 인식에 신뢰성이 있는 방법으로 사용될 수 있음을 증명하였다.