Deep Learning-based Professional Image Interpretation Using Expertise Transplant (전문성 이식을 통한 딥러닝 기반 전문 이미지 해석 방법론)
-
- Journal of Intelligence and Information Systems
- /
- v.26 no.2
- /
- pp.79-104
- /
- 2020
Recently, as deep learning has attracted attention, the use of deep learning is being considered as a method for solving problems in various fields. In particular, deep learning is known to have excellent performance when applied to applying unstructured data such as text, sound and images, and many studies have proven its effectiveness. Owing to the remarkable development of text and image deep learning technology, interests in image captioning technology and its application is rapidly increasing. Image captioning is a technique that automatically generates relevant captions for a given image by handling both image comprehension and text generation simultaneously. In spite of the high entry barrier of image captioning that analysts should be able to process both image and text data, image captioning has established itself as one of the key fields in the A.I. research owing to its various applicability. In addition, many researches have been conducted to improve the performance of image captioning in various aspects. Recent researches attempt to create advanced captions that can not only describe an image accurately, but also convey the information contained in the image more sophisticatedly. Despite many recent efforts to improve the performance of image captioning, it is difficult to find any researches to interpret images from the perspective of domain experts in each field not from the perspective of the general public. Even for the same image, the part of interests may differ according to the professional field of the person who has encountered the image. Moreover, the way of interpreting and expressing the image also differs according to the level of expertise. The public tends to recognize the image from a holistic and general perspective, that is, from the perspective of identifying the image's constituent objects and their relationships. On the contrary, the domain experts tend to recognize the image by focusing on some specific elements necessary to interpret the given image based on their expertise. It implies that meaningful parts of an image are mutually different depending on viewers' perspective even for the same image. So, image captioning needs to implement this phenomenon. Therefore, in this study, we propose a method to generate captions specialized in each domain for the image by utilizing the expertise of experts in the corresponding domain. Specifically, after performing pre-training on a large amount of general data, the expertise in the field is transplanted through transfer-learning with a small amount of expertise data. However, simple adaption of transfer learning using expertise data may invoke another type of problems. Simultaneous learning with captions of various characteristics may invoke so-called 'inter-observation interference' problem, which make it difficult to perform pure learning of each characteristic point of view. For learning with vast amount of data, most of this interference is self-purified and has little impact on learning results. On the contrary, in the case of fine-tuning where learning is performed on a small amount of data, the impact of such interference on learning can be relatively large. To solve this problem, therefore, we propose a novel 'Character-Independent Transfer-learning' that performs transfer learning independently for each character. In order to confirm the feasibility of the proposed methodology, we performed experiments utilizing the results of pre-training on MSCOCO dataset which is comprised of 120,000 images and about 600,000 general captions. Additionally, according to the advice of an art therapist, about 300 pairs of 'image / expertise captions' were created, and the data was used for the experiments of expertise transplantation. As a result of the experiment, it was confirmed that the caption generated according to the proposed methodology generates captions from the perspective of implanted expertise whereas the caption generated through learning on general data contains a number of contents irrelevant to expertise interpretation. In this paper, we propose a novel approach of specialized image interpretation. To achieve this goal, we present a method to use transfer learning and generate captions specialized in the specific domain. In the future, by applying the proposed methodology to expertise transplant in various fields, we expected that many researches will be actively conducted to solve the problem of lack of expertise data and to improve performance of image captioning.
This study aims to explore the direction for Korea's effective response to Target 3 (30by30), which can be said to be the core of the Kunming-Montreal Global Biodiversity Framework (K-M GBF) of the Convention on Biological Diversity (CBD), to find the direction of systematic OECM (Other Effective area-based Conservation Measures) discovery at the national level through a survey of global conceptual review and expert perception of OECM. This study examined ① the use of Korean terms related to OECM, ② derivation of determining criteria reflecting global standards, ③ deriving types of potential OECM candidates in Korea, and ④ considerations for OECM identification and reporting to explore the direction for identifying systematic, national-level OECM that complies with global standards and reflects the Korean context. First, there was consensus for using Korean terminology that reflects the concept of OECM rather than simple translations, and it was determined that "nature coexistence area" was the most preferred term (12 people) and had the same context as CBD 2050 Vision of "a world of living in harmony with nature." This study suggests utilizing four criteria (1. No protected areas, 2. Geographic boundaries, 3. Governance/management, and 4. Biodiversity value) that reflect OECM's core characteristics in the first-stage selection process, carrying out the consensus-building process (stage 2) with the relevant agencies, and adding two criteria (3-1 Effectiveness and sustainability of governance and management and 4-1 Long-term conservation) and performing the in-depth diagnosis in stage 3 (full assessment for reporting). The 28 types examined in this study were generally compatible with OECMs (4.45-6.21/7 points, mean 5.24). In particular, the "Conservation Properties (6.21 points)" and "Conservation Agreements (6.07 points)", which are controlled by National Nature Trust, are shown to be the most in line with the OECM concept. They were followed by "Buffer zone of World Natural Heritage (5.77 points)", "Temple Forest (5.73 points)", "Green-belt (Restricted development zones, 5.63 points)", "DMZ (5.60 points)", and "Buffer zone of biosphere reserve (5.50 point)" to have high potential. In the case of "Uninhabited Islands under Absolute Conservation", the response that they conformed to the protected areas (5.83/7 points) was higher than the OECM compatibility (5.52/7 points), it is determined that in the future, it would be preferable to promote the listing of absolute unprotected islands in the Korea Database on Protected Areas (KDPA) along with their surrounding waters (1 km). Based on the results of a global OECM standard review and expert perception survey, 10 items were suggested as considerations when identifying OECM in the Korean context. In the future, continuous research is needed to identify the potential OECMs through site-level assessment regarding these considerations and establish an effective in-situ conservation system at the national level by linking existing protected area systems and identified OECMs.
Introduction Recently, changes in the patterns and concepts of maternity care, in both developing and developed countries have been accelerating. An outstanding development in this field is the number of deliveries taking place in hospitals or maternity centers. In Korea, however, more than 90% of deliveries are carried out at home with the help of untrained relatives or even without helpers. It is estimated that less than 10% of deliveries are assisted by professional persons such as a physician or a midwife. Taking into account the shortage of professional person i11 rural Korea, it is difficult to expect widespread prenatal, postnatal, and delivery care by professional persons in the near future, It is unrealistic, therefore, to expect rapid development of MCH care by professional persons in rural Korea due to economic and sociological reasons. Given these conditions. it is reasonable that an educated village women could used as a "maternity aid", serving simple and technically easy roles in the MCH field, if we could give such a women incentive to do so. The midwife and physician are assigned difficult problems in the MCH field which could not be solved by the village worker. However, with the application of the village worker system, we could expect to improve maternal and child hoalth through the replacement of untrained relatives as birth attendants with educated and trained maternity aides. We hope that this system will be a way of improving MCH care, which is only one part of the general health services offered at the local health centre level. Problems of MCH in rural Korea The field of MCH is not only the weakest point in the medical field in our country hut it has also dropped behind other developing countries. Regarding the knowledge about pregnancy and delivery, a large proportion of our respondents reported having only a little knowledge, while 29% reported that they had "sufficient" knowledge. The average number of pregnancies among women residing in rural areas was 4.3 while the rate of women with 5 or more pregnancies among general women and women who terminated childbearing were 43 and 80% respectively. The rate of unwanted pregnancy among general women was 19.7%. The total rate for complications during pregnancy was 15.4%, toxemia being the major complication. The rate of pregnant women with chronic disease was 7%. Regarding the interval of pregnancy, the rates of pregnancy within 12 months and within 36 months after last delivery were 9 and 49% respectively. Induced abortion has been increasing in rural areas, being as high as 30-50% in some locations. The maternal death rate was shown 10 times higher than in developed countries (35/10,000 live births). Prenatal care Most women had no consultation with a physician during the prenatal period. Of those women who did have prenatal care, the majority (63%) received such care only 1 or 2 times throughout the entire period of pregnancy. Also, in 80% of these women the first visit Game after 4 months of gestation. Delivery conditions This field is lagging behind other public health problems in our country. Namely, more than 95% of the women deliveried their baby at home, and delivery attendance by a professional person occurred only 11% of the time. Attendance rate by laymen was 78% while those receiving no care at all was 16%. For instruments used to cut the umbilical corn, sterilized scissors were used by 19%, non-sterilized scissors by 63% and 16% used sickles. Regarding delivery sheets, the rate of use of clean sheets was only 10%, unclean sheets, vinyl and papers 72%, and without sheets, 18%. The main reason for not using a hospital as a place of delivery was that the women felt they did not need it as they had previously experience easy deliveries outside hospitals. Difficult delivery composed about 5% of the total. Child health The main food for infants (95%) was breast milk. Regarding weaning time, the rates within one year, up to one and half, two, three and more than three years were 28,43,60,81 and 91% respectively, and even after the next pregnancy still continued lactation. The vaccination of children is the only service for child health in rural Korea. As shown in the Table, the rates of all kinds of vaccination were very low and insufficient. Infant death rate was 42 per 1,000 live births. Most of the deaths were caused by preventable diseases. Death of infants within the neonatal period was 83% meaning that deaths from communicable diseases decreased remarkably after that time. Infant deaths which occurred without medical care was 52%. Methods of improvement in the MCH field 1. Through the activities of village health workers (VHW) to detect pregnant women by home visiting and. after registration. visiting once a month to observe any abnormalities in pregnant women. If they find warning signs of abnormalities. they refer them to the public health nurse or midwife. Sterilized delivery kits were distributed to the expected mother 2 weeks prior to expected date of delivery by the VHW. If a delivery was expected to be difficult, then the VHW took the mother to a physician or call a physician to help after birth, the VHW visits the mother and baby to confirm health and to recommend the baby be given proper vaccination. 2. Through the midwife or public health nurse (aid nurse) Examination of pregnant women who are referred by the VHW to confirm abnormalities and to treat them. If the midwife or aid nurse could not solve the problems, they refer the pregnant women to the OB-GY specialist. The midwife and PHN will attend in the cases of normal deliveries and they help in the birth. The PHN will conduct vaccination for all infants and children under 5, years old. 3. The Physician will help only in those cases referred to him by the PHN or VHW. However, the physician should examine all pregnant women at least three times during their pregnancy. First, the physician will identify the pregnancy and conduct general physical examination to confirm any chronic disease that might disturb the continuity of the pregnancy. Second, if the pregnant woman shows any abnormalities the physician must examine and treat. Third, at 9 or 10 months of gestation (after sitting of the baby) the physician should examine the position of the fetus and measure the pelvis to recommend institutional delivery of those who are expected to have a difficult delivery. And of course. the medical care of both the mother and the infants are responsible of the physician. Overall, large areas of the field of MCH would be served by the VHW, PHN, or midwife so the physician is needed only as a parttime worker.
Sasang-Medicine can classify all sympton with more simple classifying system than past Oriental-Medicine, because Sasang Byeon-Zeung(=classifying system of the sympton) separate by four clearly. The merit of this Sasang Byeon-Zeung can be seen more clearly on the part of the pathology of the expiratory-scattering and inspiratory-gathering of the Tae-Eum and Tae-Yang. On this view point, this thesis discussed the following subjects. 1. Investigate the theory of raising-falling and scattering-gathering developed in the Dongeui-Suse-Bowon. 2. Investigate the changes of the recognition of the Yang-Dog sympton and Jo-Yeol sympton argued as Interior-overheating-sympton of the Tae-Eum-In caused by liver's receiving heat. 3. Investigate the Yi-Je-Ma's view on the Eum-Yang in the argument of interior-overheating-sympton of the Tae-Eum-In caused by liver's receiving heat. As a result, the following conclusions were led to. 1. Dongeui-Suse-Bowon considers Spleen-Kidney has the couple motion of the raising Yang and falling Eum, and Liver-Lung has the couple motion of the expiratory-scattering and inspiratory-gathering. This theory of raising-falling and scattering-gathering is same as in the concept with the gathering. This theory of raising-falling and scattering-gathering is same as in the concept with the theory of raising-falling and floating-sinking of past Oriental-Medicine, but more consistently systematized in the pathology and prescription. 2. Dongeui-Suse-Bowon considers the Yang-Dog sympton and Jo-Yeol sympton as the interior-overheating-sympton of the Tae-Eum-In. As following the book, the fire of desire weeken the expiratory-scattering power of the lung, and deepen the shortage of the expiratory-scattering power comparison to the inspiratory-gathering power. Therfore the sympton can be treated by releasing ourselves from the desire and taking medicine strengthening the expiratory-scattering power. 3. In the early stage of the orintal medicine, they used prescriptions composed of So-Yang medicine and Tae-Eum medicine which can cool heat. Galgeun, Mawhang and Seungma were used in the age of Sanghanron, thereafter Jugoing's Jojung-Tang and Gongsin's Galgeunhaegi-Tang were developed as prescriptions of the interior-overheating-sympton of the Tae-Eum-In, and finally Tea-Uem-In Galgeunhaegi-Tang was settled by Yi-Je-Ma.
The purpose of this thesis is to analyze characteristics of everyday movement showed in performances of Steve Paxton. A work of art has been realized as a special object enjoyed by high class people as high culture for a long time. Therefore, a gap between everyday life and art has been greatly existed, and the emergence of everyday elements in a work of art means that public awareness involving social change is changed. The postmodernism as the period when a boundary between art and everyday life is uncertain was a postwar society after the Second World War and a social situation that rapidly changes into a capitalistic society. Changes in this time made scholars gain access academically concepts related to everyday life, and affected artists as the spirit of the times of pluralistic postmodernism refusing totality. At the same period of the time, modern dance also faced a turning point as post-modern dance. After the Second World War, modern dance started to be evaluated as it reaches the limit, and at this juncture, headed by dancers including the Judson Dance Theatre. Acting as a dancer in a dance company of Merce Cunningham, Steve Paxton, one of founders of the Judson Dance Theatre, had a critical mind of the conditions of dance company with the social structure and the process that movement is made. This thinking is showed in early performances as an at tempt to realize everyday motion it self in performances. His early activity represented by a walking motion attracted attention as a simple motion that excludes all artful elements of existing dance performances and is possible to conduct by a person who is not a dancer. Although starting the use of everyday movement is regarded as an open characteristic of post-modern dance, advanced researches on this were rare, so this study started. In addition, studies related to Steve Paxton are skewed towards Contact Improvisation that he rose as an active practician. As the use of ordinary movement before he focused on Contact Improvisation, this study examines other attempts including Contact Improvisation as attempts after the beginning of his performances. Therefore, the study analyzes Satisfyin' Lover, Contact at 10th & 2nd and Bound that are performances of Steve Paxton, and based on this, draws everyday characteristics. In addition, related books, academic essays, dance articles and reviews are consulted to consider a concept related to everyday life and understand dance historical movement of post-modern dance. Paxton attracted attention because of his activity starting at critical approach of movement of existing modern dance. As walking of performers who are not dancers, a walking motion showed in Satisfyin' Lover gave esthetic meaning to everyday movement. After that, he was affected by Eastern ideas, so developed Contact Improvisation making a motion through energy of the natural laws. In addition, he had everyday things on his performances, and used a method to deliver various images by using mundane movement and impromptu gestures originating from relaxed body. Everyday movement of his performances represents change in awareness of performances of the art of dancing that are traditionally maintained including change of dance genre of an area. His activity with unprecedented attempt and experimentation should be highly evaluated as efforts to overcome the limit of modern dance.
Recently banks and large financial institutions have introduced lots of Robo-Advisor products. Robo-Advisor is a Robot to produce the optimal asset allocation portfolio for investors by using the financial engineering algorithms without any human intervention. Since the first introduction in Wall Street in 2008, the market size has grown to 60 billion dollars and is expected to expand to 2,000 billion dollars by 2020. Since Robo-Advisor algorithms suggest asset allocation output to investors, mathematical or statistical asset allocation strategies are applied. Mean variance optimization model developed by Markowitz is the typical asset allocation model. The model is a simple but quite intuitive portfolio strategy. For example, assets are allocated in order to minimize the risk on the portfolio while maximizing the expected return on the portfolio using optimization techniques. Despite its theoretical background, both academics and practitioners find that the standard mean variance optimization portfolio is very sensitive to the expected returns calculated by past price data. Corner solutions are often found to be allocated only to a few assets. The Black-Litterman Optimization model overcomes these problems by choosing a neutral Capital Asset Pricing Model equilibrium point. Implied equilibrium returns of each asset are derived from equilibrium market portfolio through reverse optimization. The Black-Litterman model uses a Bayesian approach to combine the subjective views on the price forecast of one or more assets with implied equilibrium returns, resulting a new estimates of risk and expected returns. These new estimates can produce optimal portfolio by the well-known Markowitz mean-variance optimization algorithm. If the investor does not have any views on his asset classes, the Black-Litterman optimization model produce the same portfolio as the market portfolio. What if the subjective views are incorrect? A survey on reports of stocks performance recommended by securities analysts show very poor results. Therefore the incorrect views combined with implied equilibrium returns may produce very poor portfolio output to the Black-Litterman model users. This paper suggests an objective investor views model based on Support Vector Machines(SVM), which have showed good performance results in stock price forecasting. SVM is a discriminative classifier defined by a separating hyper plane. The linear, radial basis and polynomial kernel functions are used to learn the hyper planes. Input variables for the SVM are returns, standard deviations, Stochastics %K and price parity degree for each asset class. SVM output returns expected stock price movements and their probabilities, which are used as input variables in the intelligent views model. The stock price movements are categorized by three phases; down, neutral and up. The expected stock returns make P matrix and their probability results are used in Q matrix. Implied equilibrium returns vector is combined with the intelligent views matrix, resulting the Black-Litterman optimal portfolio. For comparisons, Markowitz mean-variance optimization model and risk parity model are used. The value weighted market portfolio and equal weighted market portfolio are used as benchmark indexes. We collect the 8 KOSPI 200 sector indexes from January 2008 to December 2018 including 132 monthly index values. Training period is from 2008 to 2015 and testing period is from 2016 to 2018. Our suggested intelligent view model combined with implied equilibrium returns produced the optimal Black-Litterman portfolio. The out of sample period portfolio showed better performance compared with the well-known Markowitz mean-variance optimization portfolio, risk parity portfolio and market portfolio. The total return from 3 year-period Black-Litterman portfolio records 6.4%, which is the highest value. The maximum draw down is -20.8%, which is also the lowest value. Sharpe Ratio shows the highest value, 0.17. It measures the return to risk ratio. Overall, our suggested view model shows the possibility of replacing subjective analysts's views with objective view model for practitioners to apply the Robo-Advisor asset allocation algorithms in the real trading fields.
The wall shear stress in the vicinity of end-to end anastomoses under steady flow conditions was measured using a flush-mounted hot-film anemometer(FMHFA) probe. The experimental measurements were in good agreement with numerical results except in flow with low Reynolds numbers. The wall shear stress increased proximal to the anastomosis in flow from the Penrose tubing (simulating an artery) to the PTFE: graft. In flow from the PTFE graft to the Penrose tubing, low wall shear stress was observed distal to the anastomosis. Abnormal distributions of wall shear stress in the vicinity of the anastomosis, resulting from the compliance mismatch between the graft and the host artery, might be an important factor of ANFH formation and the graft failure. The present study suggests a correlation between regions of the low wall shear stress and the development of anastomotic neointimal fibrous hyperplasia(ANPH) in end-to-end anastomoses. 30523 T00401030523 ^x Air pressure decay(APD) rate and ultrafiltration rate(UFR) tests were performed on new and saline rinsed dialyzers as well as those roused in patients several times. C-DAK 4000 (Cordis Dow) and CF IS-11 (Baxter Travenol) reused dialyzers obtained from the dialysis clinic were used in the present study. The new dialyzers exhibited a relatively flat APD, whereas saline rinsed and reused dialyzers showed considerable amount of decay. C-DAH dialyzers had a larger APD(11.70