거대 인용 자료를 이용한 문서 추천 방법

채민우,강민수,김용대,Chae, Minwoo,Kang, Minsoo,Kim, Yongdai 한국데이터정보과학회 2013 한국데이터정보과학회지 Vol.24 No.5

본 연구에서는 논문이나 특허 등의 문서들의 인용 정보를 활용하여 연관성이 높고 중요한 특허를 추천하는 방법을 제안한다. 문서 간의 연관성 지표인 공통피인용횟수와 중요도 지표인 HITS를 적절한 형태로 결합한 뉴먼 커널로부터 두 정보의 반영 정도를 조율하는 것이 핵심이다. 제안하는 방법은 미래의 인용에 대한 예측 오차를 최소화하는 것으로 이를 통해 뉴먼 커널의 조율모수 ${\gamma}$를 적절하게 선택할 수 있다. 또한, 거대 인용 자료를 분석하기 위해 필요한 계산 기술에 대해서 자세히 논의한다. 마지막으로, 미국 등록 특허 400만 건에 대한 실증적 자료 분석을 시행한다. In this research, we propose a document recommendation method which can find documents that are relatively important to a specific document based on citation information. The key idea is parameter tuning in the Neumann kernal which is an intermediate between a measure of importance (HITS) and of relatedness (co-citation). Our method properly selects the tuning parameter ${\gamma}$ in the Neumann kernal minimizing the prediction error in future citation. We also discuss some comutational issues needed for analysing large citation data. Finally, results of analyzing patents data from the US Patent Office are given.

A mixture of beta–Dirichlet processes prior for Bayesian analysis of event history data

채민우,김용대,Rafael Weißbach,조광현 한국통계학회 2013 Journal of the Korean Statistical Society Vol.42 No.3

In this paper, we propose a mixture of beta–Dirichlet processes as a nonparametric prior for the cumulative intensity functions of a Markov process. This family of priors is a natural extension of a mixture of Dirichlet processes or a mixture of beta processes which are devised to compromise advantages of parametric and nonparametric approaches. They give most of their prior mass to the small neighborhood of a specific parametric model. We show that a mixture of beta–Dirichlet processes prior is conjugate with Markov processes. Formulas for computing the posterior distribution are derived. Finally, results of analyzing credit history data are given.

부분방전 진단을 위한 초음파 계측 및 딥러닝기법의 응용

채민우(Min-Woo Chae),강봉수(Bong-Soo Kang) 제어로봇시스템학회 2021 제어·로봇·시스템학회 논문지 Vol.27 No.11

This paper presents an innovative diagnosis method for partial discharge which may damage the components of a high-voltage electrical equipment and cause power outbreak. Instead of using the conventional direct-attachment method, it is convenient to measure the arc noise produced by partial discharge from a safe distance. Then, the source signal is converted to a phase-resolved graph learned through the convolutional neural network. The proposed deep-learning scheme, after learning different phase-resolved graphs obtained from three types of partial discharges and environmental noises measured in the actual laboratory, exhibited a prediction accuracy of more than 94 % for the test samples. Therefore, if this technique is embedded on an ultrasonic camera, then a non-specialist like a gas inspector can easily diagnose the high-voltage facilities distributed over a broad area.

Statistical theory for deep learning

Minwoo Chae(채민우) 대한산업공학회 2022 대한산업공학회 추계학술대회논문집 Vol.2022 No.11

SCL-90-R을 사용한 사상체질별 심리정신 특성 연구

채한(Han Chae),이수진(Soo Jin Lee),박소정(So Jung Park),김병주(Byung Joo Kim),홍진우(Jin Woo Hong),황민우(Min Woo Hwang),이상남(Sang Nam Lee),한창현(Chang Hyun Han),권영규(Young Kyu Kwon) 한의병리학회 2010 동의생리병리학회지 Vol.24 No.4

실시간 반정량 PCR, JAK2 MutaScreenTM kit를 이용한 JAK2 V617F 유전자 변이 선별검사의 유용성

채효진1,이제훈,임지향,김명신,김용구,한경자,이종욱,민우성,정승원,조병식,조석구 대한진단검사의학회 2009 Annals of Laboratory Medicine Vol.29 No.3

Background : Real-time PCR for quantification of JAK2 V617F has recently been introduced and used to evaluate the importance of mutant allele burden in both diagnosis and disease progression in myeloproliferative diseases (MPDs). We evaluated the usefulness of JAK2 MutaScreenTM kit that uses a real-time semiquantitative PCR method and has been designed to screen JAK2 V617F mutant allele burden. Methods : Forty MPD patients were included in this study. We screened JAK2 V617F and determined the mutant allele burden using JAK2 MutaScreenTM kit. The mutant allele burden was estimated by six-scaled standards of JAK2 V617F mutant allele (2%, 5%, 12.5%, 31%, 50%, and 78%). For evaluation of test performance, an allele-specific PCR (AS-PCR) was carried out in all samples by using Seeplex JAK2 Genotyping kit. We assessed the clinical differences in distinct disease entities of MPDs according to JAK2 V617F mutant allele burden. Results : JAK2 V617F mutation was detected in 30 cases, including 10 of 11 cases (91%) of polycythemia vera (PV), 13 of 20 cases (65%) of essential thrombocythemia (ET), and 2 of 3 cases (67%) of chronic idiopathic myelofibrosis (CIMF). The concordance rate between the two tests was 95% (38/40). JAK2 V617F mutant allele burden was greater than 50% in 17 cases, and 10 of them (59%) were PV. In contrast, mutant allele burden was less than 50% in 13 cases and 11 of them (85%) were ET. Conclusions : JAK2 MutaScreenTM kit that utilizes a real-time semi-quantitative PCR method is a useful tool for diagnosing MPDs precisely. It can be used to assess the grade of mutant allele burden as well as to screen JAK2 V617F simultaneously. 배경 : 최근 JAK2 V617F 유전자 변이에 실시간 정량 PCR법 을 이용한 검사법이 소개되어 MPDs의 진단과 질환의 진행에 있어 유전자 변이량 측정의 중요성 평가에 사용되고 있다. 저자 들은 JAK2 V617F 유전자 변이 선별을 위해 새로이 고안된 실 시간 반정량 PCR법인 JAK2 MutaScreenTM kit의 유용성을 평 가하였다. 방법 : 40예의 MPDs 환자를 대상으로 JAK2 MutaScreenTM kit를 이용하여 JAK2 V617F 선별 및 유전자 변이량을 측정하 였다. JAK2 V617F 유전자 변이량 측정에는 6단계의 표준물질 을 사용하였다(2%, 5%, 12.5%, 25%, 31%, 50%, 78%). 검사 수 행능의 평가를 위하여 동일한 모든 검체에 대하여 Seeplex JAK2 Genotyping kit을 이용한 AS-PCR법으로 시행한 검사 결과와 비교하였다. 또한 각 MPDs의 질환군과 유전자 변이량에 따른 임상적 특성의 차이를 분석하였다. 결과 : 30예에서 JAK2 V617F 유전자 변이가 검출되었으며 각 질환별 양성률은 PV 91% (10/11), ET 65% (13/20), CMIF 67% (2/3)였다. 두 검사방법 간의 일치율은 95%였다(38/40). JAK2 V617F 유전자 변이량이 50% 이상인 17예 중 10예가 PV였으며, JAK2 유전자 변이량이 50% 이하인 13예 중에서는 11예가 ET였 다(85%). 결론 : 실시간 반정량 PCR법을 이용한 검사법인 JAK2 V617F MutaScreenTM kit는 유전자 변이 검출 및 유전자 변이량의 동 시 측정이 가능하므로 보다 정밀한 MPDs의 진단에 유용한 검 사방법으로 생각된다.

생존분석을 이용한 고부가 젖소 형질 개발

정구환,채민우,이슬기,조광현,김용대 한국자료분석학회 2013 Journal of the Korean Data Analysis Society Vol.15 No.5

우리나라의 낙농산업은 현재 젖소 1두당 유량은 우수하나, 젖소 1두당 우유 생산비가 세계에서 가장 높은 수준이기 때문에 순수익은 오히려 감소하고 있는 실정이다. 본 연구에서는 농촌진흥청에서 10여 년간 조사한 자료를 분석하여 고부가 젖소 형질을 개발하고, 이를 통해 젖소의 경제수명을 늘리는 방안을 모색하고자 한다. 젖소의 도태산차 또는 초산 이후 개월령을 경제수명으로 정의하고, 초산 직후 측정되는 다양한 변수들로 경제수명을 설명하는 모형을 개발하였다. 자료의 조사시점에 살아있는 젖소에 대해서는 경제수명이 중도절단된 자료로 취급하였고, 이를 처리하기 위해 생존분석 방법을 사용하였다. 또한 고부가 젖소 형질을 찾기 위해 AIC를 통한 변수선택을 하였으며 이 모형을 사용하여 젖소의 생존율에 비례하는 형질점수를 정의하였다. 형질점수가 상위 10% 이내인 젖소를 선별하면 경제수명이 약 20% 증가하는 결과를 얻었다. 이를 활용하여 우리나라 젖소의 경제수명을 낙농선진국의 수준으로 향상시킬 수 있을 것으로 사료된다. Dairy industry in Korea has excellent milk quantity per cattle. However, net income has decreased since Korea is among the highest cow milk production cost per cattle in the world. In this study, we develop high-value traits of dairy cattle by analyzing dairy cattle data from the Rural Development Administration. Using these traits, we are looking for a method to increase the economic life of dairy cattle. The economic life defined as calving number until death or month age after first childbirth is fitted by various traits measured right after first childbirth. The economic life of dairy cattle which is alive at the time of the survey is considered as censored data and we use survival analysis to treat these data. We choose traits that affect the economic life through the process of variable selection and we construct the best model using these traits. Using this model, we can define a trait score which represents the survival rate. Selecting dairy cattle in top 10% of trait scores results in about 20% increment of dairy cattle’s economic life. In this way, the dairy industry in Korea will reach the level of developed dairy countries.

Bayesian high‑dimensional semi‑parametric inference beyond sub‑Gaussian errors

이경재,채민우,Lin Lizhen 한국통계학회 2021 Journal of the Korean Statistical Society Vol.50 No.2

We consider a sparse linear regression model with unknown symmetric error under the high-dimensional setting. The true error distribution is assumed to belong to the locally β-Hölder class with an exponentially decreasing tail, which does not need to be sub-Gaussian. We obtain posterior convergence rates of the regression coefficient and the error density, which are nearly optimal and adaptive to the unknown sparsity level. Furthermore, we derive the semi-parametric Bernstein-von Mises (BvM) theorem to characterize asymptotic shape of the marginal posterior for regression coefficients. Under the sub-Gaussianity assumption on the true score function, strong model selection consistency for regression coefficients are also obtained, which eventually asserts the frequentist’s validity of credible sets.

프로브 검사 자료 분석을 통한 패키지 수율 개선 연구

김희진,김동근,채민우,배영목,심형섭,오영찬 한국자료분석학회 2021 한국자료분석학회 학술대회자료집 Vol.2021 No.1

Digestive system-related pathophysiological symptoms of Sasang typology: Systematic review

이미숙,손경우,김윤희,황민우,권영규,배나영,채한 한국한의학연구원 2013 Integrative Medicine Research Vol.2 No.2

The purpose of this study was to review clinical studies on digestive system-related pathophysiological symptoms of each Sasang type to obtain the generalizable typespecific clinical features, which are important for the diagnosis of the Sasang type and subsequent disease treatment. Sasang typology and digestive system symptom-related keywords were used to search through eight domestic and foreign databases up to March 2012. The results were organized and analyzed based on four categories [digestive function, appetite, eating pattern, and body mass index (BMI)] to elucidate type-specific symptoms. Sasang type-specific digestive system-related symptoms were identified by reviewing 30 related articles that were gathered by searching through the databases. The Tae-Eum (TE) type had the highest digestive functions and the So-Eum (SE) type had the lowest. The TE type appeared to have larger volume with fast eating speed compared with the SE type and individuals in the TE category preferred fatty or salty food, which is responsible for the high occurrence rates of organic digestive diseases such as gastritis. Moreover, BMI was higher in the TE type and lower in the SE type. We systematically reviewed previously published clinical reports on digestive functions, which can be used to meet the objective of Sasang-type differentiation and pathophysiological pattern identification.