• Journal of KIISE:Software and Applications
• /
• v.26 no.1
• /
• pp.139-139
• /
• 1999

In most application programs, loops usually comprise most of the computation in a program and the most important source of parallelism. When the data dependency relation is uniformin terms of distance, several compile time parallelization methods were introduced. On the otherhand,when the data dependency relation is non-uniform in distance, the compile time extraction ofparallelism is much complicated. In this paper, a general method the extracting parallelism in nestedloops is presented. This algorithm can be applicable where the dependency relation is both uniform andnon-uniform in distance. According to execution repeatedly the statements in nested loops, thealgorithm which effectively removes these kind of data dependencies is developed in order to presentthe total parallelization of nested loops.

• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.137-142
• /
• 2017

원격 지도 학습은 자동으로 매우 큰 코퍼스와 지식베이스 간의 주석 데이터를 생성하여 기계 학습에 필요한 학습 데이터를 사람의 손을 빌리지 않고 저렴한 비용으로 만들 수 있어, 많은 연구들이 관계 추출 문제를 해결하기 위해 원격 지도 학습 방법을 적용하고 있다. 그러나 기존 연구들에서는 모델 학습의 입력으로 사용되는 단어 임베딩에서 단어의 동형이의어 성질을 반영하지 못한다는 단점이 있다. 때문에 서로 다른 의미를 가진 동형이의어가 하나의 임베딩 값을 가지다 보니, 단어의 의미를 정확히 파악하지 못한채 관계 추출 모델을 학습한다고 볼 수 있다. 본 논문에서는 원격 지도 학습 기반 관계 추출 모델에 다중-어의 단어 임베딩을 적용한 모델을 제안한다. 다중-어의 단어 임베딩 학습을 위해 어의 중의성 해소 모듈을 활용하였으며, 관계 추출 모델은 문장 내 주요 특징을 효율적으로 파악하는 모델인 CNN과 PCNN을 활용하였다. 본 논문에서 제안하는 다중-어의 단어 임베딩 적용 관계추출 모델의 성능을 평가하기 위해 추가적으로 2가지 방식의 단어 임베딩을 학습하여 비교 평가를 수행하였고, 그 결과 어의 중의성 해소 모듈을 활용한 단어 임베딩을 활용하였을 때 관계추출 모델의 성능이 향상된 결과를 보였다.

• 한국어정보학회:학술대회논문집
• /
• 2017.10a
• /
• pp.137-142
• /
• 2017

원격 지도 학습은 자동으로 매우 큰 코퍼스와 지식베이스 간의 주석 데이터를 생성하여 기계 학습에 필요한 학습 데이터를 사람의 손을 빌리지 않고 저렴한 비용으로 만들 수 있어, 많은 연구들이 관계 추출 문제를 해결하기 위해 원격 지도 학습 방법을 적용하고 있다. 그러나 기존 연구들에서는 모델 학습의 입력으로 사용되는 단어 임베딩에서 단어의 동형이의어 성질을 반영하지 못한다는 단점이 있다. 때문에 서로 다른 의미를 가진 동형이의어가 하나의 임베딩 값을 가지다 보니, 단어의 의미를 정확히 파악하지 못한 채 관계 추출 모델을 학습한다고 볼 수 있다. 본 논문에서는 원격 지도 학습 기반 관계 추출 모델에 다중-어의 단어 임베딩을 적용한 모델을 제안한다. 다중-어의 단어 임베딩 학습을 위해 어의 중의성 해소 모듈을 활용하였으며, 관계 추출 모델은 문장 내 주요 특징을 효율적으로 파악하는 모델인 CNN과 PCNN을 활용하였다. 본 논문에서 제안하는 다중-어의 단어 임베딩 적용 관계추출 모델의 성능을 평가하기 위해 추가적으로 2가지 방식의 단어 임베딩을 학습하여 비교 평가를 수행하였고, 그 결과 어의 중의성 해소 모듈을 활용한 단어 임베딩을 활용하였을 때 관계추출 모델의 성능이 향상된 결과를 보였다.

• Journal of Institute of Control, Robotics and Systems
• /
• v.13 no.9
• /
• pp.868-874
• /
• 2007

This paper proposes an efficient partial matching system and representation by using a region-based method for 2D image, and we applied to an extraction of the ROI(Region of Interest) according to its matching score. The matching templates consist of the global pattern and the local one. The global pattern can make it by using region-based relation between center region and its rest regions in an object. And, the local pattern can be obtained appling to the same method as global, except relation between objects. As the templates can be normalized, we use this templates for extraction of ROI with invariant to size and position. And, our system operates only one try to match, due to normalizing of region size. To use our system for searching and examining if it's the ROI by evaluating the matching function, at first, we are searching to find candidate regions with the global template. Then, we try to find the ROI among the candidates, and it works this time by using the local template. We experimented to the binary and the color image respectively, they showed that the proposed system can be used efficiently for representing of the template and the useful applications, such as partially retrievals of 2D image.

• Journal of Information Management
• /
• v.39 no.3
• /
• pp.163-177
• /
• 2008

In this paper, we analyzed the relationship between user query keywords that is used to search NDSL and technical terms extracted from NDSL journals. For the analysis, we extracted about 833,000 query keywords from NDSL search logs during nearly 17 months and approximately 41,000,000 technical terms from NDSL, INSPEC, FSTA journals. And we used only the English noun phrase in extracted those and then we did an experiment on analysis of equality, relationship analysis and frequency analysis.

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• v.32 no.2
• /
• pp.161-167
• /
• 2006

Objectives: Pericoronitis was the most common indication for mandibular third molar surgery and there are no universally acceptable predictive criteria for pericoronitis occurrence. This study was designed to analyze the correlation of the pericoronitis and the eruption state of the mandibular third molar using panoramic radiographs statistically. Materials and Methods: 218 patients whose chief complaint was the extraction of the mandibular third molar were examined. The presence and absence of pericoronitis, age, sex, position of extraction site, angulation, impaction degree, position to the anterior border of mandibular ramus, distance between distal cementoenamel junction of second molar and mesial cementoenamel junction of the mandibular third molar were assessed. Then the correlation of pericoronitis and the eruption state of the mandibular third molar were analyzed by Student's t-test and chi-square test. Results: There was no correlation between Pericoronitis and age, sex, position of the mandibular third molar. The angulation(P=0.005), impaction degree(P=0.043), relation with anterior border of mandibular ramus(P=0.003), distance between distal cementoenamel junction of second molar and mesial cementoenamel junction of the mandibular third molar(P<0.05) were correlated with pericoronitis. Conclusions: The occurrence of the pericoronitis can be predicted by the eruption state of the mandibular third molar such as angulation, impaction degree, relation with anterior border of mandibular ramus, distance between distal cementoenamel junction of second molar and mesial cementoenamel junction of third molar.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.185-189
• /
• 2019

관계추출기의 학습을 위해서는 많은 학습 데이터가 필요한데, 사람이 모으게 되면 많은 비용이 필요하여 원격 지도 학습을 이용한 데이터 수집이 많은 연구에서 사용되고 있다. 원격 지도 학습은 지식베이스를 기반으로 학습 데이터를 자동으로 만들어 내는 방식이기에 비용이 거의 들지 않지만, 지식베이스의 질과 양에 영향을 받는다. 본 연구는 원격 지도 학습을 기본으로 관계추출기의 성능을 향상 시키고, 지식베이스를 확장하는 방안으로 반복학습을 제안한다. 실험을 적은 비용으로 빠르게 진행하기 위해 반복학습을 자동화 하는 시스템을 설계하여 실험을 하였고, 이 시스템으로 관계추출기의 성능이 향상 될 수 있는 가능성을 보였으며, 반복학습을 통한 지식베이스의 확장 방안을 제시한다.

• Bioinformatics and Biosystems
• /
• v.1 no.1
• /
• pp.51-60
• /
• 2006

The paradigm in biology is currently changing from that of conducting hypothesis-driven individual experiments to that of utilizing the results of a massive data analysis with appropriate computational tools. We present LayMap, an implemented visualization system that helps the user to deal with a high volume of the biomedical literature such as MEDLINE, through the layered maps that are constructed on the results of an information extraction system. LayMap also utilizes filtering and granularity for an enhanced view of the results. Since a biomedical information extraction system gives rise to a focused and effective way of slicing up the data space, the combined use of LayMap with such an information extraction system can help the user to navigate the data space in a speedy and guided manner. As a case study, we have applied the system to datasets of journal abstracts on 'MAPK pathway' and 'bufalin' from MEDLINE. With the proposed visualization, we have successfully rediscovered pathway maps of a reasonable quality for ERK, p38 and JNK. Furthermore, with respect to bufalin, we were able to identify the potentially interesting relation between the Chinese medicine Chan su and apoptosis with a high level of detail.

• Proceedings of the Korea Contents Association Conference
• /
• 2010.05a
• /
• pp.425-427
• /
• 2010

Large-scaled information extraction consists of named-entity recognition, terminology extraction and relation extraction. Since all the elementary technologies have been studied independently so far, test collections for related machine learning models also have been constructed independently. As a result, it is difficult to handle scientific documents to extract both named-entities and technical terms at once. In this study, we integrate named-entities and terminologies with PLOT(Person, Location, Organization, Terminology) in a biomedical domain and construct a test collection of PLOT and relations between PLOTs.

• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2004.03a
• /
• pp.241-246
• /
• 2004

The ocean signal that reaches the detector of an imaging system after multiple interactions with the atmospheric molecules and aerosols was retrieved from the total signal recorded at the top of the atmosphere (TOA). A simple method referred to as 'Path Extraction' applied to the Landsat-TM ocean imagery of turbid coastal water was compared with the conventional dark-pixel subtraction technique. The shape of the path-extracted water-leaving radiance spectrum resembled the radiance spectrum measured in-situ. The path-extraction was also extended to the SeaWiFS ocean color imagery and compared with the standard SeaWiFS atmospheric correction algorithm, which relays on the assumption of zero water leaving radiance at the two NIR wavebands (765 and 865nm). The path-extracted water-leaving radiance was good agreement with the measured radiance spectrum. In contrast, the standard SeaWiFS atmospheric correction algorithm led to essential underestimation of the water-leaving radiance in the blue-green part of the spectrum. The reason is that the assumption of zero water-leaving radiance at 755 and 865nm fails due to backscattering by suspended mineral particles. Therefore, the near infrared channels 765 and 865nm used fur deriving the aerosol information are no longer valid for turbid coastal waters. The path-extraction is identified as a simple and efficient method of extracting the path radiance largely introduced due to light interaction through the complex atmosphere carried several aerosol and gaseous components and at the air-sea interface.interface.