韓國 綜合農協의 信用事業과 經濟事業의 費用構造 分析

정용규 중앙대학교 1993 국내박사

Individualized optimization of post-transplant hepatitis B prophylaxis with hepatitis B immunoglobulin using pharmacokinetic half-life simulation

정용규 울산대학교 2019 국내박사

Backgrounds: Prophylaxis for hepatitis B virus (HBV) recurrence is essential after liver transplantation (LT) in HBV-associated recipients. This study was intended to establish individualized HBV prophylaxis protocol to optimize hepatitis B immunoglobulin (HBIG) administration with application of simulative half-life (SHL). Methods: This study consisted of 5 parts: Part 1 was designed to develop SHL estimation method with 20 patients; Parts 2 and 3 were to assess variability of SHL and to develop a simulation model to apply SHL with 100 patients; Part 4 was to validate simulation model with 114 patients; and Part 5 was a cross-sectional study on current status of HBIG infusion intervals with 660 patients. Results: In Part 1, add-on rise of anti-HBs titer from infusion of 10,000 IU HBIG and means SHL were 5252.5±873.7 IU/L and 20.0±3.7 days respectively, which was 4.4% lower and 2.2% longer than those from actual measurement. In Part 2, the median values of intra-and inter-individual coefficient of variation were 13.5% and 18.5%, respectively. Pre-transplant HBV DNA load and post-transplant antiviral therapy did not affect SHL. In Part 3, a simulation model was developed to determine the interval of HBIG infusion by using SHL. In Part 4, all 114 patients were successfully managed with regular HBIG infusion intervals of ≥8 weeks and the interval was prolonged to ≥12 weeks in 89.4% with target trough anti-HBs titer ≥200 IU/L. In Part 5, 47.4% of our patients received HBIG excessively at a target trough titer of 500 IU/L. Conclusions: SHL estimation using clinically available parameters seems to be reliably accurate comparing with results from actual measurement. We believe that SHL estimation is helpful to establish personalized HBV prophylaxis protocol to optimize HBIG administration.

시각 디자인에 나타난 포스트모더니즘에 關한 硏究

정용규 漢陽大學校 1992 국내석사

음파반응 금 나노약물을 이용한 오토파지 항암치료 : 음파에 반응하는 금 나노입자를 함유한 오토파지 증가를 통한 항암치료를 연구

정용규 가천대학교 대학원 2022 국내석사

초음파역학 요법(sonodynamic therapy, SDT)와 음파반응물질(sonosensitizer)은 치료부위를 국소화 하고 독성 부작용을 최소화하는 장점을 가져 최근 다양한 유형의 종양치료에 사용되고 있는 첨단 기술이다. 그러나 초음파역학 요법(sonodynamic therapy, SDT)은 20kHz 이상의 초음파(ultrasound)에 의해 발생되며 가청주파수(audible frequency)를 질환치료에 이용하는 연구는 현재까지 보고된 바 없다. 본 연구에서 양전하를 띤 금 나노로드에 30Hz의 음파를 가해주면 서로 응집하여 입자의 크기가 커지는 것을 최초로 발견하였다. 이렇게 응집하는 현상을 이용하여 종양세포 내로 함입된 금 나노입자를 가청주파수에 반응시키고 궁극적으로 오토파지의 형성에 크게 영향을 줄 수 있음을 확인하였다. 음파에 의해 응집되는 나노 약물에 의한 오토파지의 증가는 종양세포에 나노 항암제의 증폭된 엔도좀(endosome)이입을 초래하여 효과적으로 세포사멸을 유도할 수 있다. 본 연구의 가청주파수 음파를 이용한다면 기존 초음파역학요법보다 환자들의 거부감이 적을 뿐만 아니라 복잡한 장비와 설비없이 좀 더 대중적으로 접근된 새로운 항암치료법을 제공할 수 있다.

베이지안 망에 기초한 보조 생식술의 임상 데이터 마이닝

정용규 경기대학교 대학원 2002 국내박사

불임증이란 피임을 하지 않고 부부가 정상적인 성관계를 1년 이상 하였음에도 불구하고 임신이 되지 않는 경우를 말한다. 일반적으로 부부가 모두 정상이면 자연 임신이 1년 이내에 이루어지는 확률이 80∼90%에 이른다. 하지만 점점 늘어나는 각종 공해와 이상 기후 및 사회생활에서의 각종 스트레스나 질병 등으로 인해 해마다 불임문제로 병원을 방문하는 환자는 늘고 있다. 그러나 요즈음 생명공학과 의료기술의 발전으로 정확한 검사에 의해 원인만 제대로 밝혀지면 불임시술의 성공률도 높다. 이와 같은 생식기능의 중단에 의한 불임증을 치료하는 데에 도움을 줄 수 있는 일체의 보조기술을 보조 생식술이라 한다. 불임문제 해결을 위해서는 검사결과를 가지고 각 원인별로 치료를 해 나가는 것이 보통이나 최근 들어서는 인위적으로 배란 촉진을 위한 약물 치료법과 함께, 생성된 난자와 정자를 체외에서 수정시키는 체외 수정술(In Vitro Fertilization, IVF)이 가장 많이 사용된다. 본 연구에서는 베이지안 망을 기초로 불임환자의 임상 데이터에 대한 다양한 실험을 전개하였다. 본 연구에서 사용한 데이터는 임상에서 얻은 실제의 불임환자들에 대한 검사기록 및 시술과정이 기록된 데이터이다. 이 실험을 통해 임신여부에 영향을 주는 요인들간의 상호 의존성을 분석하고, 또 제약조건이 다른 다양한 베이지안 망의 대표적 유형으로 나이브 베이지안 망, 베이지안 망으로 확장한 나이브 베이지안 망, 일반 베이지안 망 분류기들의 분류성능을 서로 비교 분석했다. 일반적으로 베이지안 망은 기존의 방법에 비해 불확실한 상황에서도 지식을 표현하고 결론을 추론하는데 유용한 것으로 알려져 있다. 본 논문에서는 대표적인 베이지안 망 분류기들을 제시하고, 동일 임상데이터에 대해 서로 다른 유형별 베이지안 망 분류기들을 학습하였다. 또 서로 다른 유형의 베이지안 망 분류기들 중에서 특징들 간의 상호 의존관계를 허용하는 좀 더 일반적인 베이지안 망으로 확장한 나이브 베이지안 망과 일반 베이지안 망이 나이브 베이지안 망에 비해 상대적으로 더 높은 분류 성능을 보여준다는 것을 확인하였다. 또한 기계학습의 다른 방법과도 성능을 비교하였다. 결정트리(decision tree)와 k-최근접 이웃(k-nearest neighbors, k-NN)과 같은 다른 분류기들과의 성능 비교를 통해, 임상 데이터의 특성상 확률적 표현과 추론에 기초한 베이지안 망 분류기들이 보다 높은 성능을 보여준다는 사실도 확인할 수 있었다. 베이지안 망을 적용할 때 변수의 수가 많아짐에 따라 베이지안 망의 구조를 학습하는데 탐색공간이 넓어져 시간의 요구량이 급격히 많아진다. 본 연구에서는 이런 탐색공간을 효율적으로 줄이기 위하여 클래스 노드의 Markov blanket에 속한 특징들로 집합을 축소하는 것을 제안하고, 실험을 통해 이 특징 축소 방법이 베이지안 망 분류기들의 성능을 높여 줄 수 있는지 알아보았다. 분류기들의 성능에서는 축소한 특징집합으로부터 얻은 베이지안 망으로 확장한 나이브 베이지안 망(NBN with Selected Features, NBNSF) 분류기가 가장 우수한 정확도를 가짐을 알 수 있었다. 본 연구를 통하여 의료 임상분야의 데이터를 이용하여 베이지안 망이 변수들간의 확률관계를 비교적 축약된 형태로 표현하는데 최적의 모델로서 의료영역과 같은 확률적 추론, 예측, 의사결정을 요하는 분야에 잘 적용될 수 있는 최적의 지식표현 방법이 베이지안 망임을 확인할 수 있었다. Infertility is defined as the state that normal couples, not using any form of contraceptive measures, fail to become pregnant over one year. For normal couples, probability of natural pregnancy is about 80∼90%. Due to recent environmental hazards, like the air and water pollution, climate, mental stress with social relations and other diseases, it has been reported that the number of infertile patients visiting gynecologists are increasing annually. For resolving infertility problem, identifying and treating the cause of the problems require extensive examination and testing for ovulatory, womb, uterocervical canal, laparoscopy and others. Recently, medically assisted conception methods like IVF-ET (In Vitro Fertilization) through insemination of oocyte and sperm are being widely used. Computationally there have been attempts from medical domain to analyze these factors. In our research with Bayesian Network, we experimented to identify and analyze the relations between infertility factors potentially contributing to medically assisted pregnancy. Bayesian Network is a powerful common knowledge representation and reasoning tool for partial beliefs under uncertainty. It combines graph theory and probability theory to provide a practical means for representing and updating probabilities about events of interest. The framework of Bayesian Network offers a compact, intuitive, and efficient graphical representation of dependence relations between features of infertility problem analysis. In this paper, we conducted various experiments with Bayesian networks in order to analyze clinical data of infertility patients. With these experiments, we tried to find out dependencies among important factors playing the key role in clinical pregnancy, and to compare different kinds of Bayesian network classifiers in terms of classification performance. The problem of learning Bayesian Network from data is to construct the structure and CP(Conditional Probabilities) tables from a given database of cases. It includes a number of complications learning the structure, the parameters given a structure, hidden features and missing features. For our experiments of learning Bayesian Network, root and leaf nodes are only checked. Three types of network are produced as a structure graphically and parameters of each CP table. It possible to learn both of these from data. However, learning structure is much harder than learning parameters. Also, learning when some of the nodes are hidden, or we have missing data, is much harder than when everything is observed. The goal of structure learning is to learn a DAG(directed acyclic graph) that best explains the data. So far, structure learning has meant finding the right connectivity between pre-existing nodes. Finally, we suggested a feature reduction method in which all features except only some ones within Markov blanket of the class node are removed, and investigated by experiments whether such feature reduction can increase the performance of Bayesian classifiers. Features reduction makes accuracy to improve high in NBN and BAN. A counteraction in GBN could be explained that removing features of ETD, ICT and IVF makes the network to clean all the relations with Stimulation which is deeply dependent upon clinical pregnancy. From the empirical evidence demonstrated in research on Bayesian algorithms, we conclude that a Bayesian Network induced from real world data retains potential utility to enhance or expand experts domain knowledge.

소형 터보홴 설계인자와 음질의 상관관계에 대한 연구

정용규 인하대학교 대학원 2005 국내석사

日月山의 維管束植物相

정용규 慶北大學校 1989 국내석사

Mt. Irwol-san(altitude 1218.5m, latitude 36?8'02"N and longitude 129?5'58"E) is located at the northern Kyungsangpookdo province which belongs to the temperate with annual averages of 1048.3mm in rainfall and of 10.7 in temperature, thus belongs to the central floral zone. The soils are derived from the weathering of bedrocks of which sandstones and conglomeratic sandstone vein run in north along this mountain while marl and mudstone in south. The survey was conducted 12 times from July, 1986 to July, 1988. The flora of the investigated area was 674 taxa which comprises 5 classes, 40 orders, 101 families, 338 genera, 556 species, 3 subspecies, 105 varieties and 10 forms, and this result is added 312 and 435 taxa to the previously reported 362 and 239 taxa respectively. The taxa of Mt. Irwol-san is summarized as follows; ◁표삽입▷(원문을 참조하세요)

형사불법의 내용과 기능에 관한 연구

정용규 경북대학교 2012 국내박사

Beim strafrechtlichen Delikts gilt es als der Grundsatz, das absichtliche und Vorsatzvollendung zu bestrafen. Aber ausnahmsweise werden auch das fahrlässige und Versuch bestraft. Beim gleichen Delikt werden das fahrlässige und Versuch, im Vergleich mit dem absichtlichen und Vorsatzvollendung, leicht bestraft. Was ist der Grund dafür?̀̀ Auch, wenn es der Grundsatz ist, das absichtliche und Vorsatzvollendung zu bestrafen, was ist die Grundlage dafür, das Vorsuch ohne Ergebnis zu bestrafen? Was ist gleichfalls die Grundlage dafür, das fahrlässige Verbrechen ohne Absicht zum Verbrechen zu bestrafen? Die Antwort auf diese Fragen geht gerade von dem Inhalt und Wesen des strafrechtlichen Unrechts aus. Diese Fragen haben, je nach dem Gesichtspunkt zum Wesen solches Unrechts, auch einen Einfluss auf die Straftatkonstruktion. Das Wesen des Unrechts ist in dem Ergebnis aus dem Eingreifen ins Rechtgut zu finden oder es kann nach dem Maßstab gesucht werden, dass Absicht und Zweck des Täters sowie Methode und Weise der Tat sozial erlaubt werden oder nicht. Mit anderen Worten lässt sich die Straftatkonstruktion danach bestimmen, ob als das Wesen des Delikts ein Schwerpunkt auf das Unrecht als Erfolgsunrecht oder auf das Unrecht als Handlungsunrecht oder auf die beiden gelegt wird. Die Aufkärung des Wesens des strafrechtlichen Unrechts ist nicht anders als die des Wesens er Deliktsystematik. In der Regel ist die Voraussetzung zur Realisierung des Haftungsprinzips als ein großer Grundsatz des Strafrechts die Absage und Zusage der Straftatkonstruktion und damit wird es gemeint, dass von der Haftung erst dann gesprochen werden kann, wenn ein Unrecht vorausgegangen ist. Dasjenige Unrecht, das dem Tatbestand entspricht und gesetzwidrige Tat ist, hat einen großen Einfluss auf die Entwicklung der Deliktsystematik in der Strafrechtwissenschaft gehabt, und ist besonders in Deutschland, das das kontinentale Recht vertritt, wurde zur Grunglage für die Entwicklung der Strafrechtwissenschaft. Daruaf basiert haben sich bis heute in Japan und auch Korea verschiedene Definitionstypen des Inhalts des Unrechts entwickelt. In dem Entwicklungsprozess der Unrechtslehre in Deutschland ist zuerst die klassische Unrechtslehre zu erklären. Diese Theorie betrachtet das Unrecht als Eingreifen in die objektive Sozialordnung. Das heißt, das Recht wird als Bewertungsnorm gesehen. Aus dem Gesichtspunkt der Rechtnorm aus gesehen, versteht die klassische Unrechtslehre das Recht als Bewertungsnorm und ist folglich der Meinung, dass das Unrecht eine Gegenwertigkeit des der Bewertungsnorm entsprechenden Gegenstands ist und nämlich, von dem Gesichtspunkt des Erfolgsunwert, ein Erfolgsunrecht in die objektive Ordnung ist. Deshalb ist hier das Unrecht von jedem subjektiv personalen Element frei, das lediglich der Schuld vorbehalten bleibt. Zunächst gibt es eine Theorie des persönlichen Unrechts(personale Unrechtslehre), und nach dem Verständnis über den Inhalt der Erfolgsunwert und Handlungsunwert teilt sich dieses Unrecht wiederum in eine dualistische Unrechtslehre und eine monologische Unrechtslehre. Die erstere versteht, unter Führung von Welzel, das Unrecht in erster Linie als eine persönliche Handlungsunwert, die auf eine Verbotnorm verletzende, zweckhafte Tat beruht, und in zweiter Linie als eine kausale Erfolgsunwert, die im Rahmen der Handlungsunwert auf eine Verursachung des Tatergebnises beruht. Hier teilt sich die Handlungsunwert, nach dem Verständnis über ihren Inhalt, in zwei Meinugen. Nämlich steht eine Meinung, Elemente der Handlungsunwert nur als Intentionsunwert und übrige Elemente als Erfolgsunwert zu sehen, einer Meinung, sowhl Intentionsunwert als auch objektive Elemente wie Art, Methode und Mittel der Tat sowie Stand der Täters in die Handlungsunwert einzuschließen. Ich meine, dass diese Meinung am richtigsten ist. Die letztere, die Bonner Schule genannt wird, hält Befehlscharakter der Rechtsnorm für wichtig und versteht das Unrecht als eine auf eine Intentionsunwert beschränkte Handlungsunwert und erkannt eine Erfolgsunwert nicht als eine Unrecht konstruierende Funktion an, weil sie nur eine objektive strafbare Bedingung ist. Schließlich gibt es eine soziale Unrechtslehre und diese Theorie hält, wie die dualistische persönliche Unrechtslehre, sowohl Erfolgsunwert, als auch Handlungsunwert für Gegenstand für die Unrechtbeurteilung. Aber die Handlungsunwert in der sozialen Unrechtslehre ist in dem Sinne anders, dass sie nicht auf der subjektiven Willenseinstellung der Täters beruht wie die der persönlichen Unrechtslehre, sondern dass sie sich, früher als dies, auf die objektive Pflichtenstellung stützt. Aber diese Meinung wird in dem Sinne kritisiert, dass sie allzu die subjektiven Elemente ignoriert. Einige Rechtwissenschaftler Koreas und Japans scheinen in der Regel der dualistischen persönlichen Unrechtslehre zuzustimmen. Nächstens, von dem Gesichtspunkt der Funktion des Unrecht aus gesehen, ist ein Fokus darauf gelegt, als was beim Versuch das Handlungsunrecht und beim fahrlässigen Verbrechen das Erfolgsunrecht gesehen werden sollen. Und in den subjektive Rechtfertigungselement wird der Fokus der Diskusion auf den rechtlichen Effekt beim Ausfallen des Handlungsunrecht gelegt, und in dem Erlaubnistatbestandsirrtum auf die Zuage und Absage der Anerkannung des Handlungsunrecht. Beim Teilnahme wird der Fokus, unter der Anerkennung des Unrechts die Teilnahme selbst, darauf, aus welcher Struktur das Unrecht besteht. So spielen der Inhalt und die Funktion eine wesentliche Rolle in der Konstruktion des Delikts, und zur Untersuchung darüber sind breites Denken und neue Bereitschaft erforderlich.

LED 램프를 이용한 PSALI 시스템의 制御에 따른 에너지 使用量 분석

鄭溶規 忠南大學校 大學院 2012 국내석사

수시로 변화하는 주광에 의한 실내조명환경은 창측과 실안측의 밝기 차에 의한 강도 높은 Glare, 불균일한 조도레벨에 의해 실루엣과 모델링 현상 등 재실자의 심리적?생리적 불쾌감을 유발 시키며, 외부 주광 조도에 대응이 가능한 Dimming CONTROL SYSTEM을 도입함으로써 상대적으로 어두운 실안측의 조도 수준을 개선, 재실자의 심리적? 생리적 쾌적성을 증가 시킬 수 있다. 또, LED의 기술적 장점인 자유로운 색온도 조절과 출력에 따른 짧은 반응속도, 신속한 조도 변경 등을 활용한다면, 수시로 변화하는 외부조도에 대응한 신속한 실내조도의 변경이 가능하다. 이에 본 연구에서는 이러한 LED광원을 설치한 실물대 모형과 조명 시뮬레이션을 이용하여 PSALI 조명시스템으로 부분적 Dimming제어를 통한 사무공간의 실내 조명환경과 사무공간의 창호의 특성과 계절에 따른 실내 조명환경을 각각 분석하였으며, 이를 바탕으로 최적의 Dimming 단계를 도출, 이때 사용된 에너지 사용량을 제어조건에 따라 비교? 분석하였다. 실험결과, Overcast sky 조건하에서 창호의 개구율과 계절에 따른 조명 시뮬레이션 분석을 실시한 결과, 춘분과 추분의 실내조명환경의 동일하게 나타났다. 또, 동지의 평균조도는 하지에 비교하여 47%의 변화를 보이는 것으로 나타났으며, 이는 창호의 개구율이 100%, 70%, 40%로 달라지더라도 일정하게 유지되는 것으로 나타났다. 시뮬레이션을 통한 적정 Dimming안을 도출한 결과, 창측 조명만을 사용할 경우 전체적인 평균조도는 높아졌지만, 주광의 직접적 영향을 받는 Dead Zone의 평균조도를 높임으로써 창측과 실안측의 조도차가 심화되는 것으로 나타났다. 또, 에너지 사용량을 분석하기 위하여 도출된 주광환경(Overcast sky) 적정 Dimming안의 조명기기 소비전력을 비교?분석한 결과, 조명기기 제어방식으로는 실안측 조명만을 조광할 때 두 조명기기를 교차 조광할 경우 대비 최대 70.94% 에너지 절감이 가능한 것으로 나타나 가장 효과적인 것으로 분석되었다. PSALI is referred to the supplementary lighting for the interior lighting under the daily lighting situation, and pursuant to the pertinent regulations in energy savings design standard and others in recent architecture works, the importance thereof has been increasing gradually coupled with the energy performance index (EPI), energy savings plan and the like as well as expansion of submittal and implementation policies. However, Indoor luminous environment using frequently changing daylight may produce psychological and physiological discomfort to occupants owing to high strength glare caused by difference in brightness between the window-side and the indoor-side, or silhouettes and modeling phenomena caused by uneven levels of illumination. It is possible to improve psychological and physiological comfort of occupants with a dimming control system that can cope with illumination using external daylight and improve the level of illumination in the dark indoor-side. In addition, when applying flexible control of color temperature, fast response speed based on output, and rapid change of illumination, which are the technological advantages of LED lighting, then it is possible to adjust indoor illumination without delay to cope with frequently changing outdoor illumination. To this end, the study analyzed indoor luminous environment of an office space through partial dimming control with a PSALI system based on a mock-up model with LED lighting and illumination simulation, as well as indoor luminous environment of the office space pursuant to characteristic of windows by season, respectively. Based on the simulation result, optimal dimming stages are deduced, of which energy consumption was compared and analyzed according to the . When analyzing the illumination simulation based on aperture ratio of windows and by season in overcast sky conditions, the results showed that indoor luminous environment was same at the spring equinox and the autumnal equinox. In addition, average illumination on the winter solstice day was changed by 47% compared to the summer solstice day. It appeared that such result was maintained when aperture ratio of a window is changed to 100%, 70%, and 40%. When estimating an optimal dimming plan based on the simulation, it showed that when using window-side lighting only, average illumination was raised generally, though the gap of illumination level between the window-side and the indoor-side was widened owing to the increase of average illumination at dead zone which is directly affected by daylight. Moreover, when comparatively analyzing power consumption of light fixtures specified in the obtained optimal dimming plan in a daylight environment (overcast sky) to analyze energy consumption, it showed that using two lamps crossed each other was the most effective as it would save more energy up to 70.94% than using indoor lighting only.

최대 엔트로피 모델을 이용한 한국어 품사 태깅

정용규 한림대학교 1998 국내석사

This paper presents a statistical model which trains from a corpus annotated with part-of-speech tags and assigns them to previously unseen text. The Model can be classified as a Maximum Entropy model and simultaneously uses many contextual "features" to predict part-of-speech tag. This paper demonstrates the use of specialized features to model difficult tagging decisions, discusses the corpus consistency problems discovered during the implementation of these features, and proposes a training strategy that mitigates these problems. We describe a method for statistical modeling based on maximum entropy. Given consistent statistical evidence, a unique ME solution is guaranteed to exist, and an iterative algorithm exists which is guaranteed to converge to it. The ME framework is extremely general: any phenomenon that can be described in terms of statistics of the text can be readily incorporated. We present a maximum likelihood approach for automatically constructing maximum entropy models and describe how to implement this approach efficiently, using as examples several problems in natural language processing. An stochastic tagger base on the maximum entropy model was trained on the KONAN corpus, showed about 94.2% accurate result of tagging on open test corpus.