• 제목/요약/키워드: leave-one-out

검색결과 120건 처리시간 0.022초

Tracing history of the episodic accretion process in protostars

  • Kim, Jaeyeong;Lee, Jeong-Eun;Kim, Chul-Hwan;Hsieh, Tien-Hao;Yang, Yao-Lun;Murillo, Nadia;Aikawa, Yuri;Jeong, Woong-Seob
    • 천문학회보
    • /
    • 제46권2호
    • /
    • pp.66.3-67
    • /
    • 2021
  • Low-mass stars form by the gravitational collapse of dense molecular cores. Observations and theories of low-mass protostars both suggest that accretion bursts happen in timescales of ~100 years with high accretion rates, so called episodic accretion. One mechanism that triggers accretion bursts is infalling fragments from the outer disk. Such fragmentation happens when the disk is massive enough, preferentially activated during the embedded phase of star formation (Class 0 and I). Most observations and models focus on the gas structure of the protostars undergoing episodic accretion. However, the dust and ice composition are poorly understood, but crucial to the chemical evolution through thermal and energetic processing via accretion burst. During the burst phase, the surrounding material is heated up, and the chemical compositions of gas and ice in the disk and envelope are altered by sublimation of icy molecules from grain surfaces. Such alterations leave imprints in the ice composition even when the temperature returns to the pre-burst level. Thus, chemical compositions of gas and ice retain the history of past bursts. Infrared spectral observations of the Spitzer and AKARI revealed a signature caused by substantial heating, toward many embedded protostars at the quiescent phase. We present the AKARI IRC 2.5-5.0 ㎛ spectra for embedded protostars to trace down the characteristics of accretion burst across the evolutionary stages. The ice compositions obtained from the absorption features therein are used as a clock to measure the timescale after the burst event, comparing the analyses of the gas component that traced the burst frequency using the different refreeze-out timescales. We discuss ice abundances, whose chemical change has been carved in the icy mantle, during the different timescales after the burst ends.

  • PDF

서울 지역 지상 NO2 농도 공간 분포 분석을 위한 회귀 모델 및 기계학습 기법 비교 (Comparative Assessment of Linear Regression and Machine Learning for Analyzing the Spatial Distribution of Ground-level NO2 Concentrations: A Case Study for Seoul, Korea)

  • 강은진;유철희;신예지;조동진;임정호
    • 대한원격탐사학회지
    • /
    • 제37권6_1호
    • /
    • pp.1739-1756
    • /
    • 2021
  • 대기 중 이산화질소(NO2)는 주로 인위적인 배출요인으로 발생하며 화학 반응을 통해 이차오염 물질 및 오존 형성에 매개 역할을 하는 인체 건강에 악영향을 미치는 물질이다. 우리나라는 지상 관측소에 의한 실시간 NO2 모니터링을 수행하고 있지만, 이는 점 기반의 관측 값으로써 미관측 지역의 공간 분포 분석이 어렵다는 한계점을 지닌다. 본 연구에서는 선형 회귀 기반 모델인 다중 선형 회귀와 회귀 크리깅, 기계학습 알고리즘인 Random Forest (RF), Support Vector Regression (SVR)을 적용한 공간 내삽 모델링을 통해 서울 지역의 지상 NO2 농도 지도를 제작하였고, 일별 Leave-One-Out Cross Validation (LOOCV) 교차 검증을 시행하였다. 2020년 연구기간 내 일별 LOOCV에서 MLR, RK, SVR 모델의 일별 평균 Index of agreement (IOA)는 약 0.57로 유사한 성능을 보였으며, RF (0.50)보다 높은 성능이 확인되었다. RK의 일별 평균 nRMSE는 0.9483%으로 MLR (0.9501%)보다 상대적으로 낮은 오차를 나타냈다. MLR과 RK, RF 모델의 계절별 공간 분포는 비슷한 양상을 보였으며, RF는 다른 모델에 비해 좁은 NO2 농도 범위가 확인되었다. 본 연구에서 제안된 선형 회귀 기반 공간 내삽은 지상 NO2 뿐 아니라 다른 대기 오염 물질의 도시 지역 공간 내삽을 위해 활용 가능성이 높을 것으로 기대된다.

다종 위성자료와 인공지능 기법을 이용한 한반도 주변 해역의 고해상도 해수면온도 자료 생산 (Generation of Daily High-resolution Sea Surface Temperature for the Seas around the Korean Peninsula Using Multi-satellite Data and Artificial Intelligence)

  • 정시훈;추민기;임정호;조동진
    • 대한원격탐사학회지
    • /
    • 제38권5_2호
    • /
    • pp.707-723
    • /
    • 2022
  • 위성기반 해수면온도는 광역 모니터링이 가능한 장점이 있지만, 다양한 환경적 그리고 기계적 이유로 인한 시공간적 자료공백이 발생한다. 자료공백으로 인한 활용성의 한계가 있으므로, 공백이 없는 자료 생산이 필수적이다. 따라서 본 연구에서는 한반도 주변 해역에 대해 극궤도와 정지궤도 위성에서 생산되는 해수면온도 자료를 두 단계의 기계학습을 통해 융합하여 4 km의 공간해상도를 가지는 일별 해수면온도 합성장을 만들었다. 첫번째 복원 단계에서는 Data INterpolate Convolutional AutoEncoder (DINCAE) 모델을 이용하여 다종 위성기반 해수면온도 자료를 합성하여 복원하였고, 두번째 보정 단계에서는 복원된 해수면온도 자료를 현장관측자료에 맞춰 Light Gradient Boosting Machine (LGBM) 모델로 학습시켜 최종적인 일별 해수면온도 합성장을 만들었다. 개발된 모델의 검증을 위해 복원 단계에서 무작위 50일의 자료 중 일부분을 제거하여 복원한 뒤 제거된 영역에 대해 검증하였으며, 보정 단계에서는 Leave One Year Out Cross Validation (LOYOCV) 기법을 이용하여 현장자료와의 정확도를 검증하였다. DINCAE 모델의 해수면온도 복원 결과는 상당히 높은 정확도(R2=0.98, bias=0.27℃, RMSE=0.97℃, MAE=0.73℃)를 보였다. 두번째 단계의 LGBM 보정 모델의 정확도 개선은 표층 뜰개 부이와 계류형 부이 현장자료와의 비교에서 모두 상당한 향상(RMSE=∆0.21-0.29℃, rRMSE=∆0.91-1.65%, MAE=∆0.17-0.24℃)을 보여주었다. 특히, 모든 현장 자료를 이용한 보정 모델의 표층 뜰개 부이와의 정확도는 동일한 현장 자료가 동화된 기존 해수면온도 합성장보다 나은 정확도를 보였다. 또한 LGBM 보정 모델은 랜덤포레스트(random forest)를 사용한 선행연구에서 보고된 과적합의 문제를 상당부분 해결하였다. 보정된 해수면온도는 기존의 초고해상도 해수면온도 합성장들과 유사한 수준으로 수온 전선과 와동 등의 중규모 해양현상을 뚜렷하게 모의하였다. 본 연구는 다종위성 자료와 기계학습 기법을 사용해 시공간적 공백 없는 고해상도 해수면온도 합성장 제작 방법을 제시하였다는 점에서 가치가 있다.

On the "Virtual and Real" and Blankness in Chinese Landscape Painting

  • Dongqi, Liu
    • International Journal of Advanced Culture Technology
    • /
    • 제10권3호
    • /
    • pp.174-183
    • /
    • 2022
  • The abstract should summarize the contents of the paper and written below the author information. Use the word "Abstract" as the title, in 12-point Times New Roman, boldface type, italicized, centered relative to the column, initially capitalized, fixed-spacing at 13 pt., 12 pt. spacing before the text and 6 pt. after. The abstract content is to be in 11-point, italicized, single spaced type. Leave one blank line after the abstract, and then begin the keywords. All manuscripts must be in English. When it comes to the issue of "virtual and real" in traditional Chinese painting, the first impression is to describe the problems of painting strokes and ink, layout of pictures, etc., but it runs through the initial conception of the work, creation in the middle and aesthetic appreciation of the work. It exists in the whole process of artistic creation and appreciation. In essence, it is a problem of aesthetic thinking and philosophical thinking. Because the traditional Chinese painting theory is influenced by Taoism, when the concept of "virtual and real" is implemented in the specific picture of Chinese painting, it is contained in the specific shape of "physics", that is, the painting theory research of "blank space" in the picture. Based on the traditional Taoist philosophy of China, this paper takes the "virtual and real" view in Lao Zhuang's thought as the research object, deeply analyzes and compares its relationship with the "virtual and real" in Chinese landscape painting, and finds out their artistic spirit, essential characteristics and how to present them. This paper mainly discusses the internal relationship between Taoist philosophy and "virtual and real" in Chinese landscape painting from the following aspects. The introduction expounds the origin, purpose, significance, innovation and research methods of the topic. This paper analyzes the philosophical thoughts about landscape in the philosophical thoughts represented by Lao Tzu and Zhuangzi. The development of Chinese traditional aesthetics theory is closely related to Taoist philosophy, which has laid the foundation and pointed out the direction for the development of Chinese painting theory since ancient times. It also discusses the influence of the Taoist philosophy of "the combination of the virtual and real" on the emergence and development of the artistic conception of landscape painting. Firstly, through the analysis of the artistic conception of landscape painting and its constituent factors, it is pointed out that the artistic conception is affected by the personality and the painting artistic conception. Secondly, through the Taoist thought of "the combination of the virtual and real" in landscape painting, so as to reflect that it is the source of the artistic conception of Chinese landscape painting. It is the unique spiritual concept of "Yin and Yang" and "virtual and real" that creates the unique "blank space" aesthetic realm of Chinese painting in the composition of the picture. Finally, it focuses on the "nothingness" in Taoist philosophy and the "blank space" in Chinese landscape painting. The connotation of the "blank space" in Chinese painting exceeds its own expressive significance, which makes the picture form the aesthetic principle of emotional blending, virtual and real combination and dynamic and static integration. Through the "blank space", it deepens the artistic characteristics of the picture and sublimates the expression of "form" in Chinese painting.

류마티스 관절염 환자의 치료행위 모형 검증을 위한 횡문화적 비교연구 (A Transcultural Study for Testing Models of the Treatment-seeking Behaviors in Patients with Rheumatoid Arthritis)

  • 이인숙;이은옥;은영
    • 근관절건강학회지
    • /
    • 제6권2호
    • /
    • pp.253-277
    • /
    • 1999
  • Patients with chronic disease have various treatment patterns because it shows a progressive degenerative feature. Especially various physical and emotional problems of the rheumatoid arthritis patients leave them shopping around various types of treatment. According to previous studies, over 70% of patients with arthritis experienced the traditional oriental medicine or folk remedies simultaneously with medical treatment within one year after the onset of disease. The purposes of this study are 1) to compare the patterns of treatment-seeking behaviors between Korean arthritis patients and Americans ; and 2) test two models of treatment-seeking behaviors by path analysis, one for early treatment-seeking behavior model(ETBM) and the other is chronic treatment-seeking behavior model (CTBM) in Korean sample. The interview survey was performed to 133 RA patients with structured questionnaire at out-patient clinic or public health center. Patients characteristics such as age, duration of disease were similar in two countries except higher educational background in Americans. There were no patients using only alternative therapies or no medical treatment in the US. Most of the American patients have chosen both medical treatment and alternative therapy, while the Koreans less than American. In Korea, combined treatment group usually consists of the people who are younger, more educated and higher economic status than the characters of other groups in early or chronic stages. In early stage, they tend to have strong belief of curing from the disease, satisfy the relationship with their physicians and comply with direction of the medical professional. The paths of two models were explained by 70% in ETBM and 33% in CTBM. When the models were modified, almost all paths of the CTBM were the same as the previous one, but direct determinant factor was changed from the relationship with physicians to the lay referral system in chronic model. These two models' explanation powers became 94% and 88%, respectively. The attitude or perception of disease, lay referral system and the relationship with medical personnel are the main determinants of treatment-seeking behaviors.

  • PDF

Development of an Imaging Based Gang Protection System

  • Grimm, M.;Pelz, M.
    • International Journal of Railway
    • /
    • 제1권4호
    • /
    • pp.149-156
    • /
    • 2008
  • During maintenance or construction works in or at the tracks of railways, high risks for passengers and railway staff, especially for the workers on the construction site exist. The high risks result out of the movement of rail vehicles, like trains or construction vehicles, which must be faced by using any available technical and operational technologies for securing them against the environment. Therefore, it is necessary to evaluate the level of protection continuously and to identify new and innovative methods and technologies for the protection of the gang (construction worker, machines and material). Especially on construction sites at line sections with two or more parallel tracks but also with single tracks, there are still a lot of incidents and accidents mostly with seriously injured persons or fatalities. These were mainly gang members that breach the railway-loading gage. By using proper warning or protection systems, the avoidance of such accidents must be achieved. The latest developments. in gang protection systems concern on the one hand fixed barriers in the middle between the construction site and the operated track and on the other hand construction vehicles equipped with automatic warning systems. The disadvantage of such protection methods is that the gang can be warned against an approaching train but a monitoring of the gang members cannot be performed. Only one part of a potential dangerous situation will be detected. If the gang members will overhear the acoustic warning signal of the security staff and the workers will not leave the danger zone in the track, the driver of the approaching train had no chance to react to the dangerous situation. An accident is often inevitable. While the detection of acoustic warning signals by the gang members working on a construction site is very difficult, the acoustical planning of an automatic warning system has to be designed for an acoustic short range level of one meter besides the construction vehicle. The decision about the use of today's technical warning system (fixed systems, automatic warning systems, etc.) must be geared to the technical feasibility and the level of safety which is needed. Criteria for decision guidance to block a track should be developed by danger estimation and economical variables. To realize the actual jurisdiction and to minimize the hazards of railway operations by the use of construction vehicles near the tracks further developments are needed. This means, that the warning systems have to be enhanced to systems for protection, which monitor the realization of the warning signal as a precondition for giving a movement authority to a train. This method can protect against accidents caused by predictable wrongdoing. The actual state of the art technique of using a collective warning combined with additional security staff is no longer acceptable. Therefore, the Institute of Transportation System of the German Aerospace Center in Braunschweig (Germany) will develop a gang warning and protection system based upon imaging methods, with optical sensors such as video in visible and invisible ranges, radar, laser, and other. The advantage of such a system based on the possibility to monitor both the gang itself and the railway-loading gauge either of the parallel track or of the same track still in use. By monitoring both situations, the system will be able to generate a warning message for the approaching train, that there are obstacles in the track, so that the train can be stopped to prevent an accident. And also the gang workers will be warned, while they breach their area.

  • PDF

감염 근관에서 혐기성 배양법과 간접 면역 형광법 및 DNA 프로브법에 의한 Porphyromonas endodontalis의 검출에 관한 비교 연구 (COMPARATIVE STUDY ON THE DETECTION OF PORPHYROMONAS ENDODONTALIS BY ANAEROBIC CULTURE, IIF AND DNA PROBE METHOD IN INFECTED ROOT CANALS)

  • 김민겸;윤수한;정종평
    • Restorative Dentistry and Endodontics
    • /
    • 제21권1호
    • /
    • pp.1-18
    • /
    • 1996
  • There are many advantages when using IIF and DNA probe methods over anaerobic culture method in that they are time-and effort-saving, more precise and more sensitive. Furthermore, in IIF and DNA probe methods, the detection is possible only with small amount of bacteria, the quantitative analysis is possible, and the cell viability is not necessary. The purpose of this study is to observe the incidence of P.endodontalis by carrying out anaerobic culture, IIF and colony lift using DNA probe method respectively, and to compare these 3 methods in terms of effectiveness and sensitivity in order to identify the most effective detection method. 30 teeth with at least one clinical symptoms, with single canal, and with pulp necrosis were sampled. For sampling bacteria, access cavity was prepared after disinfecting tooth and its surroundings. Then the paper point was inserted up to the periapical area, leave there for a while, and finally it was placed into PRAS Ringer's sol. and PBS sol. In anaerobic culture method, P.endodontalis was identified by biochemical tests after subculturing black and brown colonies which were produced after 7 days of incubation on BAP and Brucella BAP in anaerobic chamber. To identify P.endodontalis in IIF method, species-specific polyclonal rabbit-antisera of P.endodontalis(ATCC 35406) was reacted with sampled PBS sol. dispensed onto glass slide, and then P.endodontalis was examined by phase contrast microscopy after incubating with Goat anti-rabbit lgG conjugated to Fluorescein isothiocyanate. For colony lift using DNA probe method, membranes were laid over colonies on the surface of BAP and were hybridized with cloned DNA probe of P.endodontalis. The existence of P.endodontalis was then identified by the methods of chemiluminescent detection and color metric detection. Black colony was found in 11 teeth out of 30 teeth and P.endodontalis was detected in 6 teeth (20 %) by anaerobic culture method, 16 teeth (53 %) by IIF method, and 7 teeth (23 %) by DNA probe method. IIF method is significantly better in detecting P.endodontalis than DNA probe method and anaerobic culture method. There was no significant differences between DNA probe method and anaerobic culture method. There was significant correlation between the formation of black colony and the existence of P.endodontalis. The probability of detecting P.endodontalis when black colony being present is 2.89 times higher than when not being present. There was significant relationship between the foul odor of clinical symptoms and P.endodontalis. The sensitivity of existing P.endodontalis when foul odor being present was 93.75 %, while the specificity of not existing P.endodontalis when foul odor not being present was 28.57 %. These results suggested that the probes of P.endodontalis will be used to decide the method and prognosis in endodontic treatments.

  • PDF

레이저유도 플라즈마 분광법을 이용한 폐금속 분류를 위한 추정 연성정보 기반의 최빈 분류 기술 (Estimated Soft Information based Most Probable Classification Scheme for Sorting Metal Scraps with Laser-induced Breakdown Spectroscopy)

  • 김에덴;장혜민;신성호;정성호;황의석
    • 자원리싸이클링
    • /
    • 제27권1호
    • /
    • pp.84-91
    • /
    • 2018
  • 본 연구에서는 레이저유도 플라즈마 분광법(Laser induced breakdown spectroscopy, LIBS) 기반의 금속 종류별 스펙트럼 데이터를 이용하여 연성정보(soft information)를 추정하고 최빈 클래스로 분류하는(most probable classification) 금속 분류 방법을 제안한다. 폐금속 자원과 같이 사전 정보가 없는 금속을 분류하는 경우 몇 가지 핵심 구성성분에 대한 정량 분석을 통해서 클래스를 추정하는 방법이 효율적이다. 이에 따라 부분 집합 기반의 부분최소제곱회귀법(Partial Least Square Regression, PLSR)을 이용하여 LIBS 검출 스펙트럼으로부터 각 성분의 농도를 독립적으로 신뢰성 있게 추정하고, 인증 표준물질(CRM) 등 알려진 모집합의 농도정보에 기반하여 최고 확률을 갖도록 분류하는 기술을 제안한다. 샘플 스펙트럼들의 다변량 분석을 통해서 여러 성분의 추정 농도를 다변량 정규 분포를 갖는 것으로 가정하고 통합(Joint) 추정 연성정보를 구할 수 있으며, 이를 활용한 최빈 확률 검출이나 추가적인 사전 정보의 결합 등을 통해서 분류 성능을 향상시킬 수 있다. 제안된 기술의 평가를 위해서 9가지 종류의 CRM 금속시료의 LIBS 스펙트럼 데이터를 사용하며, 부분 집합 기반의 PLSR 농도 추정 기술을 기반으로 단변량 혹은 다변량 정규 분포 연성 정보추정을 통해 미지 금속의 검출과 연성 정보의 검출 등을 테스트 하였다. 또한 방사형 차트(Radar chart)를 이용하여 추정된 농도와 획득한 연성정보를 효과적으로 시각화함으로써 기존 라이브러리에 포함된 부분 집합의 금속과 비교하여 해당 금속과의 유사성을 그래프를 통해 추정할 수 있다.

한 대학병원 자의퇴원 환자의 특성 연구 - 퇴원환자 지료정보 DB를 이용하여 - (A Study on the Characteristics of the Patients Discharged Against Medical Advice)

  • 홍준현;최귀숙;이정화;이은미
    • 한국의료질향상학회지
    • /
    • 제8권2호
    • /
    • pp.208-217
    • /
    • 2001
  • Background : The objective of this study is proving the basic data for developing a management system for the discharges against medical advice(AMA) by identifying the characteristics of the AMA patients of an university hospital for 10 years. Methods : By using discharge abstract data base, we divided the total discharges(435,254) into two groups, discharge against medical advice and discharge with discharge order. We confirmed the characteristics of AMA group by analyzing discharge abstract data of the both groups by SAS software V6.12 and $x^2$ test. Medical records of AMA patients in the year 2000 were reviewed to identify the reasons for AMA which we couldn't extract from discharge abstract DB. Result : The total number of AMA for 10 years were 9,358(2.15%) and the AMA rate has been continuously decreased for 10 years. Male, admission through emergency room, discharges admission via other hospital, patients without operation during hospitalization, discharges in hopeless or not improved condition showed higher AMA rate. The AMA rate was higher as the age of the patients was higher, and the average length of stay was longer in AMA patients than in those with discharge order. The AMA rate in psychiatry was highest(14.3%) and it was higher in surgery departments than those of medical or other sections. The AMA rate varied by attending physicians even in the same department and it was statistically significant. Patients with the principal diagnosis of "medical observation and evaluation for suspected diseases" showed the highest AMA rate(15.5%), and that of schizophrenia or psychosis was the nest. One hundred twenty-one patients(19.5%) out of 622 AMA in 2000 discharged against medical advice for transfer to order health care facilities. Among them 71 patients(58.7%) discharged with their medical care information, such as copies of medical record, medical certificates, summaries, etc. Written oath of the patients discharged AMA was filed in their medical records in 466 cases(74.9%) although some of them were incomplete. Conclusion : Characteristics of AMA discharge could be used as the basic data in developing a system to manage the patients who have risk factors to leave the hospital against medical advice. By reducing number of patients leaving the hospital against medical advice we can increase satisfaction of medical providers and consumers.

  • PDF

토마토 역병균 항균 활성 데이터의 이분번 근사모델링 (Two Class Approximation of TLB (Tomato Late Blight) Activity Data)

  • 한호규;;조승주
    • 농약과학회지
    • /
    • 제9권2호
    • /
    • pp.140-145
    • /
    • 2005
  • 정량적 구조 활성관계 모델링은 물리적인 성질과 생물학적 활성이 관계 있다는 것을 전제로 한다. 그러나, 퍼센트 활성과 같은 데이터들은 모델링에 많이 활용되지 않았다. 이것의 중요한 이유중의 하나는 이러한 값들이 정량적이 아니고 정성적인 데에 있다. 본 연구에서는 분자모델링에 퍼센트활성 데이터를 활용하기 위하여 데이터 값들을 2개의 계층으로 분류하고 CoMFA(비교분자장)를 판별함수로 활용하였다. 즉, 베타-케토아세트아닐라이드 유도체들의 토마토 역병균에 대한 항균력 시험의 퍼센트 활성 데이터를, 한 계층은 활성이 있는 것, 다른 계층은 활성이 없는 것으로 나누었다. 특히, CoMFA를 활용함으로써 화학적인 이해에 중요한 3차원적인 정보를 얻을 수 있었다. 이 모델은 주어진 데이타를 98%의 정확도로 설명하였으며, LOO 검증을 해본 결과 예측력은 약 69% 정도였다 이 결과는 활성 데이터를 근사적으로 2개의 계급으로 나누고 CoMFA를 활용하는 방식이 구조활성관계를 이해하고 화합물 유도체를 합성하는데 활용될 수 있음을 보여준다.