• Proceedings of the Korea Water Resources Association Conference
• /
• 2019.05a
• /
• pp.21-21
• /
• 2019

본 연구에서는 국내 주요 종관기상관측장비(Automated Surface Observing System; ASOS)의 강수 자료에 CSEOF 분석(Cyclo-stationary Empirical Orthogonal Function Analysis)을 적용하여 주요 성분(principle component)을 추출한 후 이를 분석하여 국내 강수의 계절적 순환 특성을 평가고자 하였다. ASOS 자료로는 전국 131개의 ASOS 중에 40년 이상의 월 강수량 자료가 구축되어 있는 47개 지점의 자료를 이용하였다. 수집한 자료의 기간은 1978년부터 2018년까지이다. 강수 자료의 월별 공간적인 강수 분포 특성을 파악하기 위해 시간적인 순환성을 고려한 CSEOF 분석을 수행하였다. 강수자료의 주성분을 추출해본 결과, CSEOF 분석의 경우 첫 번째 CSEOF 외의 다른 CSEOF들의 원자료 설명 비율 또한 작지 않게 나타나 다양한 강수 변동 특성을 평가할 수 있음을 확인하였다. 8월의 2nd CSEOF는 한반도 전체의 강수가 감소하는 것으로 나타났으며, 이는 라니냐가 7-8월 한반도 강수에 미치는 영향과 유사하다. 아울러 9월의 2nd CSEOF 결과 또한 남부를 중심으로 전체적으로 감소하는 경향이 나타남. 이는 엘리뇨 발생 시 9월의 강수 패턴과 비슷한 것으로 확인되었다. 뿐만 아니라, 우리나라에 영향을 미친 주요 태풍과 CSEOF의 상관관계도 검증할 수 있었으며, 장마와의 관계도 발견할 수 있었다. 향후, CSEOF 분석 결과에 해석방법이 개발된다면, 보다 다각적인 측면에서의 강수 계절적 순환 특성 평가가 이루어 질 수 있을 것으로 기대한다.

• Korean Journal of Ichthyology
• /
• v.31 no.1
• /
• pp.1-6
• /
• 2019

The spermatozoa of Hemiculter eigenmanni is similar to other cyprinid by spherical head containing a nucleus with highly condensed chromatin, a short midpiece with mitochondria and a flagellum located tangentially to the head. The fine structure of cyprinid spermatozoa described classical characteristics of Cyprinidae spermatozoon comprising the absent of acrosome, the shallow nucleal fossa, postnuclear distribution of mitochondria and the lateral insertion of flagellum. However there were some structural differences for their morphology, in the mitochondria and the orientation of centrioles. The proxomal and distal centrioles are oriented approximately $145^{\circ}$ to each other. The mitochondria of 8 or 10 in number are arranged in two or three layers.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.18-22
• /
• 2019

FAQ 분류는 자주 묻는 질문을 범주화하고 사용자 질의에 대해 가장 유사한 클래스를 추론하는 방식으로 진행된다. FAQ 데이터셋은 클래스가 다수 존재하기 때문에 클래스 간 포함 및 연관 관계가 존재하고 특정 데이터가 서로 다른 클래스에 동시에 속할 수 있다는 특징이 있다. 그러나 최근 FAQ 분류는 다중 클래스 분류 방법론을 적용하는 데 그쳤고 FAQ 데이터셋의 특징을 모델에 반영하는 연구는 미미했다. 현 분류 방법론은 이러한 FAQ 데이터셋의 특징을 고려하지 못하기 때문에 정답으로 해석될 수 있는 예측도 오답으로 여기는 경우가 발생한다. 본 논문에서는 신뢰성이 부족한 FAQ 데이터셋에서도 분류를 잘 하기 위해 손실 함수를 조정하는 정규화 기법을 소개한다. 이 정규화 기법은 클래스 간 포함 및 연관 관계를 반영할 수 있도록 오답을 예측한 경우에도 예측 강도에 비례하여 손실을 줄인다. 이는 오답을 높은 확률로 예측할수록 데이터의 신뢰성이 낮을 가능성이 크다고 판단하여 학습을 강하게 하지 않게 하기 위함이다. 실험을 위해서는 다중 클래스 분류에서 가장 좋은 성능을 보이고 있는 모형인 BERT를 이용했으며, 비교 실험을 위한 정규화 방법으로는 통상적으로 사용되는 라벨 스무딩을 채택했다. 실험 결과, 본 연구에서 제안한 방법은 기존 방법보다 성능이 개선되고 보다 안정적으로 학습이 된다는 것을 확인했으며, 데이터의 신뢰성이 부족한 상황에서 효과적으로 분류를 수행함을 알 수 있었다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.514-517
• /
• 2023

본 연구는 인 컨택스트 러닝 (In-Context Learning)을 오디오-언어 작업에 적용하기 위한 멀티모달 (Multi-Modal) 딥러닝 모델을 다룬다. 해당 모델을 통해 학습 단계에서 오디오와 텍스트의 소통 가능한 형태의 표현 (Representation)을 학습하고 여러가지 오디오-텍스트 작업을 수행할 수 있는 멀티모달 딥러닝 모델을 개발하는 것이 본 연구의 목적이다. 모델은 오디오 인코더와 언어 인코더가 연결된 구조를 가지고 있으며, 언어 모델은 6.7B, 30B 의 파라미터 수를 가진 자동회귀 (Autoregressive) 대형 언어 모델 (Large Language Model)을 사용한다 오디오 인코더는 자기지도학습 (Self-Supervised Learning)을 기반으로 사전학습 된 오디오 특징 추출 모델이다. 언어모델이 상대적으로 대용량이기 언어모델의 파라미터를 고정하고 오디오 인코더의 파라미터만 업데이트하는 프로즌 (Frozen) 방법으로 학습한다. 학습을 위한 과제는 음성인식 (Automatic Speech Recognition)과 요약 (Abstractive Summarization) 이다. 학습을 마친 후 질의응답 (Question Answering) 작업으로 테스트를 진행했다. 그 결과, 정답 문장을 생성하기 위해서는 추가적인 학습이 필요한 것으로 보였으나, 음성인식으로 사전학습 한 모델의 경우 정답과 유사한 키워드를 사용하는 문법적으로 올바른 문장을 생성함을 확인했다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.05a
• /
• pp.227-230
• /
• 2008

본 연구에서는 대용량 궤적 데이터베이스에서 영역 질의를 효과적으로 처리하기 위한 인덱싱 기법에 대하여 논의한다. 먼저, 기존 인덱싱 기법의 문제점을 지적하고, 이러한 문제점을 해결하는 새로운 기법을 제안한다. 제안된 기법에서는 우선 시간 차원을 다수의 시간 구간으로 분할하고, 인덱싱의 대상이 되는 전체 라인 세그먼트들을 시간 구간별로 구분한다. 각 시간 구간에 속하는 라인 세그먼트들에 대하여 별도의 인덱스를 구축한다. 또한, 디스크에서 관리되는 과거 시간 구간에 대한 인덱스들과는 달리 최근 시간 구간에 대한 인덱스는 메인 메모리상에 관리함으로써 삽입과 검색의 성능을 크게 개선할 수 있다. 각 시간 구간에 속하는 라인 세그먼트들은 다음과 같은 방식으로 인덱스를 구축한다. 먼저, 2D-트리를 이용하여 전체 공간 차원을 유사한 수의 라인 세그먼트들이 배정되도록 다수의 셀들로 분할한다. 또한, 분할된 각 셀마다 시공간 차원 (x, y, t)에 대한 별도의 3차원 $R^*$-트리를 두어 보다 상세한 인덱싱을 지원한다. 실험 결과에 의하면, 기존 기법에 비하여 작은 인덱스 구조를 갖으면서도 검색 성능면에서 $300%{\sim}1000%$까지의 성능 향상 효과를 갖는 것으로 나타났다.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.6
• /
• pp.1065-1071
• /
• 2023

In this paper discuss with research developing an LSTM model for classifying the behavior of occupants within a residence. The multi-sensor consists of an IAQ (Indoor Air Quality) sensor that measures indoor air quality, a UWB radar that tracks occupancy detection and location, and a Piezo sensor to measure occupants' biometric information, and collects occupant behavior data such as going out, staying, cooking, cleaning, exercise, and sleep by constructed an experimental environment similar to the actual residential environment. After the data with removed outliers and missing, the LSTM model is used to calculate accuracy, sensitivity, specificity of the occupant behavior classification model, T1 score.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.8 no.3
• /
• pp.285-294
• /
• 2003

To select an appropriate sampler for the environmental monitoring survey in coastal waters of Saemangeum, Jeollabuk-do, a macrobenthic sampling was conducted in April 2002. Employed samplers were dredge (type Charcot), a semi-quantitative sampler and Smith-McIntyre (SM) and van Veen grab (VV) as quantitative ones. One haul was tried for dredge and 3 replicates (0.1 ㎡${\times}$3) for SM and W at each of 11 stations. Comparisons of sediment volume in sampler bucket and of precision of biological parameters (i.e., density, biomass, species number and diversity index, H') were made between SM and VV. Sediment volume was significantly different (SM > VV) at p-value of 0.0050 (paired t-test) and, in average, 3 replicate samples of SM and VV satisfied a precision level of 0.2 by applying 4th root transformation. Patterns of observed and expected species numbers and H' were compared. Dredge-VV samples showed higher affinity than any other pair. Several dominant species in the area were underestimated in dredge samples (e.g., polychaete Heteromastus filiformis. Aricidea assimilis etc.). Quantifying the agreement pattern of multi-species responses was accomplished by estimating correlations between similarity matrices. Correlation between dredge and VV was slightly higher, but near-per-fect matches were found in general. Different ranks and composition among principal species lists were presumably linked to the effect of penetration depth that differs among samplers. Lower level of some species' abundance in VV samples (ca. 50％ compared with those of SM) was explained in this context. It seem appropriate to regard the effect as a probable cause of relatively higher correlations in dredge-VV, Overall bio-logica1 features indicated that a better choice could be SM in situations of requiring high data quality. The others work well, however, on observing and defining faunal characteristics and their capability cannot be questionted if we do not expect a first-order quality.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.34 no.1
• /
• pp.74-80
• /
• 1989

Twentyseven rice cultivars were compared Quantitatively and classified by 15 agronomic traits. For this, 12 Tongil-type rice cultivars and 15 Japonica-type rice cultivars were transplanted at the experimental farm of the Yeongnam Crop Experiment Station on May 25 in 1987. Jangseongbyeo belonging to Tongil-type had the greatest dissimilarity total among 27 rice cultivars and followed by Chilseongbyeo, Gayabyeo, Hangangchalbyeo, Gihobyeo (Tongil-type) and Sangpoongbyeo (Japonica-type) while cultivars having greatest similarity total were Daecheongbyeo and Yeongsanbyeo and followed by Nagdongbyeo, Sinsunchalbyeo, Gwangmyeongbyeo, Yeongdeogbyeo, Palgongbyeo and Dongjinbyeo that were all belonging to Japonica-type cultivars. CuItivar pairs having the greatest dissimilarity coefficient (50-55％) were combinations of Gayabyeo and Sangpungbyeo, Gihobyeo, or Seonambyeo, combinations of Chilseongbyeo and Jangseongbyeo, Sangpung-byeo, Chucheongbyeo, Gihobyeo or Seonambyeo and combinations of }angseongbyeo and Gwangmyeong-byeo, Sangpungbyeo, Gihobyeo, Daechangbyeo or H waseongbyeo. On the other hand, cultivar pairs having the greatest similarity coefficient (above 85 ％) were combinations of Dongjinbyeo and Sinsunchalbyeo, Daecheongbyeo, Somjinbyeo or Yeongsanbyeo, combination of Sangpungbyeo and Gihobyeo, combinations of Yeongdeogbyeo and Yeongsanbyeo or Palgongbyeo, combination of Gwangmyeongbyeo and Yeongdeogbyeo, Combinations of Nagdongbyeo and Gwangmyeongbyeo, Yeongdeogbyeo, Daecheongbyeo, Hwaseongbyeo or Yeongsanbyeo, combinations of Sinsunchalbyeo and Daecheongbyeo or Yeongsanbyeo, combinations of Somjinbyeo and Daecheongbyeo or Yeongsanbyeo and combination of Daecheongbyeo nd Yeongsanbyeo. Two-dimensional ordination analysis method provided the information on cultivar classification between Japonica-type and Tongil-type cultivars except a few cultivars. Only two cultivars, Joongweonbyeo and Seomjinbyeo were located opposit area that implied Tongil-type cultivar Joongweonbyeo was rather close to Japonica-type while Japonica-type cultivar Seomjinbyeo was rather close to Tongil-type, respectively. Sinsunchalbyeo and Daecheongbyeo, on the other hand, were located on the line between two cultivar groups. The method of two-dimensional ordination analysis concluded as feasible approach to classify cultivars Quantitatively for the practical purpose.

• Journal of Intelligence and Information Systems
• /
• v.27 no.1
• /
• pp.129-149
• /
• 2021

Recently, interest in social robots that can socially interact with humans is increasing. Thanks to the development of ICT technology, social robots have become easier to provide personalized services and emotional connection to individuals, and the role of social robots is drawing attention as a means to solve modern social problems and the resulting decline in the quality of individual lives. Along with the interest in social robots, the spread of social robots is also increasing significantly. Many companies are introducing robot products to the market to target various target markets, but so far there is no clear trend leading the market. Accordingly, there are more and more attempts to differentiate robots through the design of social robots. In particular, anthropomorphism has been studied importantly in social robot design, and many approaches have been attempted to anthropomorphize social robots to produce positive effects. However, there is a lack of research that systematically describes the mechanism by which anthropomorphism for social robots is formed. Most of the existing studies have focused on verifying the positive effects of the anthropomorphism of social robots on consumers. In addition, the formation of anthropomorphism of social robots may vary depending on the individual's motivation or temperament, but there are not many studies examining this. A vague understanding of anthropomorphism makes it difficult to derive design optimal points for shaping the anthropomorphism of social robots. The purpose of this study is to verify the mechanism by which the anthropomorphism of social robots is formed. This study confirmed the effect of the human-likeness of social robots(Within-subjects) and the construal level of consumers(Between-subjects) on the formation of anthropomorphism through an experimental study of 3×2 mixed design. Research hypotheses on the mechanism by which anthropomorphism is formed were presented, and the hypotheses were verified by analyzing data from a sample of 206 people. The first hypothesis in this study is that the higher the human-likeness of the robot, the higher the level of anthropomorphism for the robot. Hypothesis 1 was supported by a one-way repeated measures ANOVA and a post hoc test. The second hypothesis in this study is that depending on the construal level of consumers, the effect of human-likeness on the level of anthropomorphism will be different. First, this study predicts that the difference in the level of anthropomorphism as human-likeness increases will be greater under high construal condition than under low construal condition.Second, If the robot has no human-likeness, there will be no difference in the level of anthropomorphism according to the construal level. Thirdly,If the robot has low human-likeness, the low construal level condition will make the robot more anthropomorphic than the high construal level condition. Finally, If the robot has high human-likeness, the high construal levelcondition will make the robot more anthropomorphic than the low construal level condition. We performed two-way repeated measures ANOVA to test these hypotheses, and confirmed that the interaction effect of human-likeness and construal level was significant. Further analysis to specifically confirm interaction effect has also provided results in support of our hypotheses. The analysis shows that the human-likeness of the robot increases the level of anthropomorphism of social robots, and the effect of human-likeness on anthropomorphism varies depending on the construal level of consumers. This study has implications in that it explains the mechanism by which anthropomorphism is formed by considering the human-likeness, which is the design attribute of social robots, and the construal level of consumers, which is the way of thinking of individuals. We expect to use the findings of this study as the basis for design optimization for the formation of anthropomorphism in social robots.

• Journal of Food Hygiene and Safety
• /
• v.32 no.1
• /
• pp.57-63
• /
• 2017

Foreign materials with a variety of types and sizes are found in food; thus, extraordinary efforts and various analytical methods are required to identify the types of foreign materials and to find out accurate causes of how they unintentionally enter food. In this study, human, cow, pig, mouse, duck, goose, dog, and cat were chosen as various types of animal hairs because they can be frequently incorporated into food during its production or consumption step. We morphologically analyzed them using stereoscopic, optical, SUMP method, and scanning electron microscopes, showing differences in each type. In addition, X-ray fluorescence spectrometer (XRF) was used to analysis chemical compositions ($^{11}Na{\sim}^{92}U$, Mass%) of samples. As a result, we observed that mammalian hairs were mainly composed of sulfur. Organic compounds of samples were further analyzed by fourier transform infrared spectroscopy (FT-IR) that can compare spectra of given materials; however, this method did not show significant differences in each sample. In this study, we suggest a rapid method for the identification of the causes and types of foreign materials in food.