• The Journal of the Acoustical Society of Korea
• /
• v.14 no.5
• /
• pp.44-50
• /
• 1995

This paper describes a large-vocabulary continuous speech recognition system using continuous hidden Markov models for the Korean language. To improve the performance of the system, we study on the selection of speech modeling units, inter-word modeling, search algorithm, and grammars. We used triphones as basic speech modeling units, generalized triphones and function word-dependent phones are used to improve the trainability of speech units and to reduce errors in function words. Silence between words is optionally inserted by using a silence model and a null transition. Word pair grammar and bigram model based oil word classes are used. Also we implement a search algorithm to find N-best candidate sentences. A postprocessor reorders the N-best sentences using word triple grammar, selects the most likely sentence as the final recognition result, and finally corrects trivial errors related with postpositions. In recognition tests using a 3,000-word continuous speech database, the system attained $93.1\%$ word recognition accuracy and $73.8\%$ sentence recognition accuracy using word triple grammar in postprocessing.

• Korean Journal of Food Science and Technology
• /
• v.39 no.3
• /
• pp.348-352
• /
• 2007

Inflammation is a complex process resulting from a variety of mechanisms. Combined inhibition of the activities of enzymes involved in the process may therefore be considered more important in anti-inflammatory property of plant extracts than any single contribution. In this study, the inhibitory effects of the ethanol extracts of thirty plant foods on the activities of secretory phospholipase $A_{2}$ ($sPLA_{2}$), cyclooxygenase-1 (COX-1), cyclooxygenase-2 (COX-2), and 12-lipoxygenase (12-LOX) were examined. Several legumes, mungbean sprout and some leaf vegetables inhibited the activity of $sPLA_2$, upstream enzyme of inflammation pathway. Only soybean sprout and mungbean sprout significantly inhibited 12-LOX activity. Although most of extracts inhibited the activities of both COX-1 and COX-2, water dropwort and amaranth showed selectivity for the inhibition of COX-2 over COX-1. Especially, mungbean showed anti-inflammatory property at both upstream and downstream of inflammation pathway with relatively low $IC_{50}$ values for $sPLA_{2}$ and COX-2 enzymes. Mungbean sprout exhibited inhibitory effects on all enzymes related to early and late inflammation and soybean sprout suppressed 12-LOX and COX-2 simultaneously, although the activities of these plants were showed at relatively high concentration. Therefore, mungbean, mungbean sprout, and soybean sprout appear to exhibit anti-inflammatory effects by combined inhibition of inflammatory enzymes.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.15 no.4
• /
• pp.531-539
• /
• 1996

The research for the classification of the artificial defects in welding parts is performed using the pattern recognition technology of ultrasonic signal. The signal pattern recognition package including the user defined function is developed to perform the digital signal processing, feature extraction, feature selection and classifier selection. The neural network classifier and the statistical classifiers such as the linear discriminant function classifier and the empirical Bayesian classifier are compared and discussed. The pattern recognition technique is applied to the classification of artificial defects such as notchs and a hole. If appropriately learned, the neural network classifier is concluded to be better than the statistical classifiers in the classification of the artificial defects.

• Annual Conference on Human and Language Technology
• /
• 2009.10a
• /
• pp.61-66
• /
• 2009

본 논문은 다양한 형태의 웹 문서에 적용하기 위해서, 언어의 통계정보 및 후처리 규칙에 기반 하여 개선된 문장경계 인식 기술을 제안한다. 제안한 방법은 구두점 생략 및 띄어쓰기 오류가 빈번한 웹 문서에 적용하기 위해서 문장경계로 사용될 수 있는 모든 음절을 대상으로 학습하여 문장경계 인식을 수행하였고, 문장경계인식 성능을 최대화 하기 위해서 다양한 실험을 통해 최적의 자질 및 학습데이터를 선정하였고, 다양한 기계학습 기반 분류 모델을 비교하여 최적의 분류모델을 선택하였으며, 학습데이터에 의존적인 통계모델의 오류를 규칙에 기반 해서 보정하였다. 성능 실험은 다양한 형태의 문서별 성능 측정을 위해서 문어체와 구어체가 복합적으로 사용된 신문기사와 블로그 문서(평가셋1), 문어체 위주로 구성된 세종말뭉치와 백과사전 본문(평가셋2), 구두점 생략 및 띄어쓰기 오류가 빈번한 웹 사이트의 게시판 글(평가셋3)을 대상으로 성능 측정을 하였다. 성능척도로는 F-measure를 사용하였으며, 구두점만을 대상으로 문장경계 인식 성능을 평가한 결과, 평가셋1에서는 96.5%, 평가셋2에서는 99.4%를 보였는데, 구어체의 문장경계인식이 더 어려움을 알 수 있었다. 평가셋1의 경우에도 규칙으로 후처리한 경우 정확률이 92.1%에서 99.4%로 올라갔으며, 이를 통해 후처리 규칙의 필요성을 알 수 있었다. 최종 성능평가로는 구두점만을 대상으로 학습된 기본 엔진과 모든 문장경계후보를 인식하도록 개선된 엔진을 평가셋3을 사용하여 비교 평가하였고, 기본 엔진(61.1%)에 비해서 개선된 엔진이 32.0% 성능 향상이 있음을 확인함으로써 제안한 방법이 웹 문서에 효과적임을 입증하였다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.24 no.4
• /
• pp.343-348
• /
• 2014

Face recognition system has advanctage to automatically recognize a person without causing repulsion at deteciton process. However, the face recognition system has a drawback to show lower perfomance according to illumination variation unlike the other biometric systems using fingerprint and iris. Therefore, this paper proposed a robust face recogntion method against illumination varition by slective fusion technique using both visible and infrared faces based on fuzzy linear disciment analysis(fuzzy-LDA). In the first step, both the visible image and infrared image are divided into four bands using wavelet transform. In the second step, Euclidean distance is calculated at each subband. In the third step, recognition rate is determined at each subband using the Euclidean distance calculated in the second step. And then, weights are determined by considering the recognition rate of each band. Finally, a fusion face recognition is performed and robust recognition results are obtained.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.7
• /
• pp.1531-1536
• /
• 2011

We propose an automatic swearword filter system for online game chatting by using Support Vector Machines(SVM). We collected chatting sentences from online games and tagged them as normal sentences or swearword included sentences. We use n-gram syllables and lexical-part of speech (POS) tags of a word as features and select useful features by chi square statistics. Each selected feature is represented as binary weight and used in training SVM. SVM classifies each chatting sentence as swearword included one or not. In experiment, we acquired overall 90.4% of F1 accuracy.

• Journal of the Korean Chemical Society
• /
• v.55 no.6
• /
• pp.1030-1041
• /
• 2011

We investigated the understanding of pre-service science teacher about the chemistry concept of middle school curriculum using some items in National Assessment of Educational Achievement and analyzed the result according to background variables of pre-service science teacher. The result was that there were some pre-service science teachers who select incorrect answer at all items, pre-service science teachers don't fully understand the concept needed to solve item. And the percentage of correct answer at some items was low regardless of selection of chemistry as an elective subject at CSAT(College Scholastic Ability Test). We found some facts through the depth interviews to find the cause of the result. First, the misconception acquired in middle school days is tend not to change until college student. Second, the formation of misconception is affected by the study habit with which solve problem by simple calculation and memory without essential understanding. Third, the study habit with which solve problem by simple calculation and memory without essential understanding could not replace misconceptions acquired in middle school days with scientific concept regardless of selection of chemistry as an elective subject at CSAT.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2006.06a
• /
• pp.1-7
• /
• 2006

최근 수년간 한국어를 위한 어휘의미망에 대한 관심은 꾸준히 높아지고 있지만, 그 결과물을 어떻게 평가하고 활용할 것인가에 대한 방안은 이루어지지 않고 있다. 본 논문에서는 단어클러스터링 시스템 개발을 통하여, 어휘의미망에 의해 확장되기 전후의 클러스터링을 수행하여 데이터를 서로 비교하였다. 단어클러스터링 시스템 개발을 위해 사용된 학습 데이터는 신문 말뭉치 기사로 총 68,455,856 어절 규모이며, 특성벡터와 벡터공간모델을 이용하여 시스템A를 완성하였다. 시스템B는 구축된 '[-하]동사류' 3,656개의 어휘의미를 포함하는 동사어휘의미망을 포함하여 확장된 것으로 확장대상정보를 선택하여 특성벡터를 재구성한다. 대상이 되는 실험 데이터는 '다국어 어휘의미망-코어넷'으로 클러스터링 결과 나타난 어휘들의 세 번째 층위까지의 노드 동일성 여부로 정확률 검수를 하였다. 같은 환경에서 시스템A와 시스템B를 비교한 결과 단어클러스터링의 정확률이 45.3%에서 46.6%로의 향상을 보였다. 향후 연구는 어휘의미망을 활용하여 좀 더 다양한 시스템에 체계적이고 폭넓은 평가를 통해 전산시스템의 향상은 물론, 연구되고 있는 많은 어휘의미망에 의미 있는 평가 방안을 확대시켜 나가야 할 것이다.

• Geophysics and Geophysical Exploration
• /
• v.7 no.2
• /
• pp.109-116
• /
• 2004

Petroleum reservoir characterization is a process for quantitatively describing various reservoir properties in spatial variability using all the available field data. Porosity and permeability are the two fundamental reservoir properties which relate to the amount of fluid contained in a reservoir and its ability to flow. These properties have a significant impact on petroleum fields operations and reservoir management. In un-cored intervals and well of heterogeneous formation, porosity and permeability estimation from conventional well logs has a difficult and complex problem to solve by conventional statistical methods. This paper suggests an intelligent technique using fuzzy logic and neural network to determine reservoir properties from well logs. Fuzzy curve analysis based on fuzzy logics is used for selecting the best related well logs with core porosity and permeability data. Neural network is used as a nonlinear regression method to develop transformation between the selected well logs and core analysis data. The intelligent technique is demonstrated with an application to the well data in offshore Korea. The results show that this technique can make more accurate and reliable properties estimation compared with previously used methods. The intelligent technique can be utilized a powerful tool for reservoir characterization from well logs in oil and natural gas development projects.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.40 no.1
• /
• pp.19-29
• /
• 2003

This paper presents a novel hybrid classifier for face recognition with artificially generated virtual training samples. We utilize both the nearest neighbor approach in feature angle space and a connectionist model to obtain a synergy effect by combining the results of two heterogeneous classifiers. First, a classifier called the nearest feature angle (NFA), based on angular information, finds the most similar feature to the query from a given training set. Second, a classifier has been developed based on the recall of stored frontal projection of the query feature. It uses a frontal recall network (FRN) that finds the most similar frontal one among the stored frontal feature set. For FRN, we used an ensemble neural network consisting of multiple multiplayer perceptrons (MLPs), each of which is trained independently to enhance generalization capability. Further, both classifiers used the virtual training set generated adaptively, according to the spatial distribution of each person's training samples. Finally, the results of the two classifiers are combined to comprise the best matching class, and a corresponding similarit measure is used to make the final decision. The proposed classifier achieved an average classification rate of 96.33% against a large group of different test sets of images, and its average error rate is 61.5% that of the nearest feature line (NFL) method, and achieves a more robust classification performance.