• Journal of Intelligence and Information Systems
• /
• v.28 no.4
• /
• pp.251-273
• /
• 2022

Recently, research on applying text analysis to deep learning has steadily continued. In particular, researches have been actively conducted to understand the meaning of words and perform tasks such as summarization and sentiment classification through a pre-trained language model that learns large datasets. However, existing pre-trained language models show limitations in that they do not understand specific domains well. Therefore, in recent years, the flow of research has shifted toward creating a language model specialized for a particular domain. Domain-specific pre-trained language models allow the model to understand the knowledge of a particular domain better and reveal performance improvements on various tasks in the field. However, domain-specific further pre-training is expensive to acquire corpus data of the target domain. Furthermore, many cases have reported that performance improvement after further pre-training is insignificant in some domains. As such, it is difficult to decide to develop a domain-specific pre-trained language model, while it is not clear whether the performance will be improved dramatically. In this paper, we present a way to proactively check the expected performance improvement by further pre-training in a domain before actually performing further pre-training. Specifically, after selecting three domains, we measured the increase in classification accuracy through further pre-training in each domain. We also developed and presented new indicators to estimate the specificity of the domain based on the normalized frequency of the keywords used in each domain. Finally, we conducted classification using a pre-trained language model and a domain-specific pre-trained language model of three domains. As a result, we confirmed that the higher the domain specificity index, the higher the performance improvement through further pre-training.

• Proceedings of the Korean Society of Fisheries Technology Conference
• /
• 2002.10a
• /
• pp.133-134
• /
• 2002

제주해협에 접해 있는 한국남해연안역은 대체로 50m미만의 천해로써 제주해협에 비해 계절별로 하계에 저온ㆍ고염분수, 동계에 저온ㆍ저염분수가 출현해 해협내 연중 전선을 형성하는 해역으로서 특히, 한국 남해연안역에 위치해 있는 추자도 주변해역은 지형적 특성상 대마난류수, 한국 남해연안수와 중국대륙연안수, 황해저층냉수 등 이러한 이질수괴들이 시기와 계절별로 서로 상접하여 복잡한 해황을 형성하는 해역이다(Rho, 1985, 최, 1989, 김ㆍ노, 1994, Yoon, 1986, Rhoㆍ평야, 1983). (중략)

• Journal of Bio-Environment Control
• /
• v.27 no.4
• /
• pp.356-362
• /
• 2018

Hot pepper is a kind of seasoning vegetables, which is a major item in the Korean vegetable market. Since the use of hot pepper is processed into pepper powder, which is a powder form of dried hot pepper, improvement of quality and productivity of dried hot pepper is important. Therefore, this experiment was conducted to suggest proper drying method by comparing the changes of hot pepper powder ingredients considering production cost according to the drying method. As a drying method, we used sun drying and heat drying which are widely used in practice. We also compared the productivity and quality of dried hot pepper by applying a dehumidifying drying method using a dehumidifier. Drying rate of hot pepper was highest of 85.1% at heat drying. Accordingly moisture content of hot pepper powder was lowest of 13.5% at heat drying. The American Spice Trade Association (ASTA) color value, which influenced the coloring of red pepper, showed higher in heat drying and dehumidified drying treatment than the sun drying treatment. The content of capsaicinoids was higher at sun drying treatment than that of at both heat drying and dehumidified drying treatments. The content of sugar was higher at heat drying and dehumidified drying treatment where drying time was relatively short than that of sun drying treatment. Also, there was no significant difference in sugar content between the two treatments. The production cost of dried hot pepper with dehumidified drying was 9.9% more efficient than heat drying. Through this study, it was found that heat and dehumidified drying method were effective in increasing sugar content and coloring of hot pepper powder. In order to improve the capsaicinoid content of red pepper, it is considered that appropriate drying temperature and drying time should be added in the process of heat drying and dehumidified drying.

• 동물약계
• /
• no.89
• /
• pp.4-6
• /
• 2003

[ $\cdot$ ]긴급 이사회 개최 $\cdot$동물약사감시 행정처분 결과 홍보 $\cdot$동물약품 마케팅반 교육 실시 $\cdot$돼지콜레라 예방약 추가 공급 $\cdot$농림부 및 검역원 인사 $\cdot$캐나다산 동물약품 BSE 증명서 제출 $\cdot$동물약품 부가가치세 영세율 적용 건의 $\cdot$2003 한국국제축산박람회 개최 안내 $\cdot$수입 원료 동물용의약품 관리 철저 $\cdot$동물용의약품등취급규칙 검토회의 참석 $\cdot$조합 2003년도 하반기 알찬거래선 선정 $\cdot$동물약품제조용 양허관세적용 유당 추가 배정 $\cdot$2002년도 세계 동물약품 시장 현황

• Journal of Internet Computing and Services
• /
• v.3 no.3
• /
• pp.31-38
• /
• 2002

A method of DID generating step is suggested for applying the XML technology to the information data extracted from the Legacy databases. The IDREF-ID attribute reference modeling is used for representing the complex relationship between tables and excluding the prearranged step of ID insertion. ID Insertion procedure is performed in parallel with investigating the relationship between the tables and the frequent search direction between the table data. As a result, ID insertion procedure can be performed simultaneously with understanding of the IDREF-ID relationship between tables, and DID are also generated.

• Computational Structural Engineering
• /
• v.11 no.4
• /
• pp.361-370
• /
• 1998

본 논문에서는 역전파학습에 의한 신경망기법을 사용하여 구조물의 미지계수를 추정하는 기법을 연구하였다. 대형구조물의 경우 계측 또는 추정하여야 하는 자유도의 수가 많으므로 인하여 구조계수를 추정하는 데에는 많은 어려움이 존재한다. 이러한 어려움을 극복하기 위하여 부구조추정법과 부행렬계수를 사용하여 추정하고자 하는 미지계수의 수를 효율적으로 줄일 수 있도록 하였다. 구조물의 고유주파수 및 모드형상 등의 모드계수를 신경망의 입력자료로 사용하였으며, 추정하고자 하는 부재의 부행렬계수를 신경방의 출력자료로 사용하였다. 입력자료로 사용되는 모드계수에 포함되어 있는 계측오차 및 신호처리오차의 영향을 줄이기 위하여, 신경망의 학습과정에서 노이즈를 첨가하는 기법을 사용하였다. 일반적인 형태의 자켓구조물을 대상으로 수치해석을 수행함으로써 제안기법의 대형구조계에 대한 적용성을 검증하였다.

• Journal of the Korean Statistical Society
• /
• v.6 no.2
• /
• pp.167-177
• /
• 1977

유전적 질환이 있는 가계에서 출생하는 자녀중에서 유전적인 질환을 보유할 수 있는 확률을 예측하는 방법의 하나로서 우도추정치(likelihood estimator)를 사용하는 것은 분리분석(segregation analysis)에서 중요한 역할을 하고 있다. Elston과 Stewart(1971)는 이러한 분석방법의 일반적인 통계모형을 정립하였으며 필자(1974)와 Morton 등 (1974)은 complex segregation이 될 때에 분석되는 4가지의 통계모형을 주장하였다. 본 연구의 목적은 multiple ascertainment $\pi$가 존재하는 경우 분리확률모수(segregation parameter) $\theta$의 우도추정치를 구하고 둘째로 oligogenic case에 대한 이론적인 배경을 구명하고자 한다.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.11 no.2
• /
• pp.151-162
• /
• 2009

This study proposes a new methodology for quantitative rock classification in unsampled rock zone, which occupies the most of tunnel design area. This methodology is to train an ANN (artificial neural network) by using results from a drilling investigation combined with electric resistivity survey in sampled zone, and then apply the trained ANN to making a prediction of grade of rock classification in unsampled zone. The prediction is made at the center point of a shifting window by using a number of electric resistivity values within the window as input reference information. The ANN training in this study was carried out by the RPROP (Resilient backpropagation) training algorithm and Early-Stopping method for achieving a generalized training. The proposed methodology is then applied to generate a rock grade distribution on a real tunnel site where drilling investigation and resistivity survey were undertaken. The result from the ANN based prediction is compared with one from a conventional kriging method. In the comparison, the proposed ANN method shows a better agreement with the electric resistivity distribution obtained by field survey. And it is also seen that the proposed method produces a more realistic and more understandable rock grade distribution.

• Monthly Duck's Village
• /
• s.196
• /
• pp.42-45
• /
• 2019

• Monthly Duck's Village
• /
• s.235
• /
• pp.38-41
• /
• 2023