• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.1-17
• /
• 2017

The deep learning framework is software designed to help develop deep learning models. Some of its important functions include "automatic differentiation" and "utilization of GPU". The list of popular deep learning framework includes Caffe (BVLC) and Theano (University of Montreal). And recently, Microsoft's deep learning framework, Microsoft Cognitive Toolkit, was released as open-source license, following Google's Tensorflow a year earlier. The early deep learning frameworks have been developed mainly for research at universities. Beginning with the inception of Tensorflow, however, it seems that companies such as Microsoft and Facebook have started to join the competition of framework development. Given the trend, Google and other companies are expected to continue investing in the deep learning framework to bring forward the initiative in the artificial intelligence business. From this point of view, we think it is a good time to compare some of deep learning frameworks. So we compare three deep learning frameworks which can be used as a Python library. Those are Google's Tensorflow, Microsoft's CNTK, and Theano which is sort of a predecessor of the preceding two. The most common and important function of deep learning frameworks is the ability to perform automatic differentiation. Basically all the mathematical expressions of deep learning models can be represented as computational graphs, which consist of nodes and edges. Partial derivatives on each edge of a computational graph can then be obtained. With the partial derivatives, we can let software compute differentiation of any node with respect to any variable by utilizing chain rule of Calculus. First of all, the convenience of coding is in the order of CNTK, Tensorflow, and Theano. The criterion is simply based on the lengths of the codes and the learning curve and the ease of coding are not the main concern. According to the criteria, Theano was the most difficult to implement with, and CNTK and Tensorflow were somewhat easier. With Tensorflow, we need to define weight variables and biases explicitly. The reason that CNTK and Tensorflow are easier to implement with is that those frameworks provide us with more abstraction than Theano. We, however, need to mention that low-level coding is not always bad. It gives us flexibility of coding. With the low-level coding such as in Theano, we can implement and test any new deep learning models or any new search methods that we can think of. The assessment of the execution speed of each framework is that there is not meaningful difference. According to the experiment, execution speeds of Theano and Tensorflow are very similar, although the experiment was limited to a CNN model. In the case of CNTK, the experimental environment was not maintained as the same. The code written in CNTK has to be run in PC environment without GPU where codes execute as much as 50 times slower than with GPU. But we concluded that the difference of execution speed was within the range of variation caused by the different hardware setup. In this study, we compared three types of deep learning framework: Theano, Tensorflow, and CNTK. According to Wikipedia, there are 12 available deep learning frameworks. And 15 different attributes differentiate each framework. Some of the important attributes would include interface language (Python, C ++, Java, etc.) and the availability of libraries on various deep learning models such as CNN, RNN, DBN, and etc. And if a user implements a large scale deep learning model, it will also be important to support multiple GPU or multiple servers. Also, if you are learning the deep learning model, it would also be important if there are enough examples and references.

• Proceedings of the Korean Environmental Health Society Conference
• /
• 2005.06a
• /
• pp.45-72
• /
• 2005

C. neoformans-associated cryptococcosis is primarily a disease of immunocompromised persons, has a world-wide distribution, and is often spread by pigeons in the urban environment. In contrast, C. gattii causes infection in normal hosts, has only been described in tropical and semi-tropical areas of the world, and has a unique niche in river gum Eucalyptus trees. Cryptococcosis is acquired through inhalation of the yeast propagules from the environment. C. gattii has been identified as the cause of an emerging infectious disease centered on Vancouver Island, British Columbia, Canada. No cases of C. gattii-disease were diagnosed prior to 1999; the current incidence rate is 36 cases per million population. A search was initiated in 2001 to find the ecological niche of this basidiomycetous yeast. C. gaftii was found in the environment in treed areas of Vancouver Island. The highest percentage of colonized-tree clusters were found around central Vancouver Island, with decreasing rates of colonization to the north and south. Climate, soil and vegetation cover of this area, called the Coastal Douglas fir biogeoclimatic zone, is unique to British Columbia and Canada. The concentration of airborne C. gattii was highest in the dry summer months, and lowest during late fall, winter, and early spring, months which have heavy rainfall. The study of the emerging colonization of this organism and subsequent cases of environmentally acquired disease will be informative in planning public health management of new routes of exposure to exotic agents in areas impacted by changing climate and land use patterns. Cryptococcosis is an infection associated with an encapsulated, basidiomycetous yeast Cryptococcus neoformans. The route of entry for this organism is through the lungs, with possible systemic spread via the circulatory system to the brain and meninges. There are four cryptococcal serogroups associated with disease in humans and animals, distinguished by capsular polysaccharide antigens. Cryptococcus neoformans: variety grubii (serotype A), variety neoformans (serotype D), and variety gattii (serotypes B and C) (Franzot et at. 1999). C. neoformans variety gattii has recently been elevated to species status, C. gattii. C. neoformans val. grubii and var. neoformans have a world-wide distribution, and are particularly associated with soil and weathered bird droppings. In contrast, C. gattii (CG) is not associated with bird excrement, is primarily found in tropical and subtropical climates, and has a restricted environmental niche associated with specific tree species. (Ellis & Pfiffer 1990) Ellis and Pfeiffer theorize that, as a basidiomycete, CG requires an association with a tree in order to become pathogenic to mammals. In Australia, CG has been found to be associated with five species of Eucalypts, Eucalyptus camaldulensis, E. tereticornis, E. blakelyi, E. gomphocephala, and E. rudis. Eucalypts, although originally native to Australia, now have a world-wide distribution. CG has been found associated with imported eucalypts in India, California, Brazil, and Egypt. In addition, in Brazil and Columbia, where eucalypts have been naturalized, native trees have been shown to harbour CG (Callejas et al. 1998; Montenegro et al. 2000). In British Columbia, Canada, since the beginning of 1999, there have been 120 confirmed cases of cryptococcal mycoses associated with CG in humans, including 4 fatalities (data from British Columbia Centre for Disease Control), and over 200 cases in animal pets in BC (data from Central Laboratory for Veterinarians). What is remarkable about the BC outbreak of C. gattii-cryptococcosis is that all of the cases have been residents of, or visitors to, a narrow area along the eastern coast of Vancouver Island, BC, from the tip of the island in the south (Victoria) to Courtenay on the north-central island as illustrated in Figure 1. Of the first 38 human cases, 58% were male with a mean age of 59.7 years (range 20 - 82): 36 cases (95%) were Caucasian. Ten cases (26%) presented with meningitis, the remainder presented with respiratory symptoms. Cultures recovered from cases of cryptococcosis associated with the outbreak were typed as serogroup B, which is specific to CG (Bartlett et al. 2003). This was the first reported outbreak of CVG in Canada, or indeed, the world. Where infection with CG is endemic, for example, Australia, the incidence of cryptococcosis ranges from 1.8 - 4.7 per million between the southern and northern states (Sorrell 2001). However, the overall incidence of cryptococcosis in immunocompenent individuals has been estimated at 0.2 per million population per year (Kwon-Chung et al. 1984). The population of Vancouver Island is approximately 720,000,consequently, even if the organism were endemic, one would expect a maximum of 0.15 cases of cryptococcal disease annually.

• Journal of Intelligence and Information Systems
• /
• v.17 no.1
• /
• pp.111-125
• /
• 2011

Clustering is a process of grouping similar or relevant documents into a cluster and assigning a meaningful concept to the cluster. By this process, clustering facilitates fast and correct search for the relevant documents by narrowing down the range of searching only to the collection of documents belonging to related clusters. For effective clustering, techniques are required for identifying similar documents and grouping them into a cluster, and discovering a concept that is most relevant to the cluster. One of the problems often appearing in this context is the detection of a complex concept that overlaps with several simple concepts at the same hierarchical level. Previous clustering methods were unable to identify and represent a complex concept that belongs to several different clusters at the same level in the concept hierarchy, and also could not validate the semantic hierarchical relationship between a complex concept and each of simple concepts. In order to solve these problems, this paper proposes a new clustering method that identifies and represents complex concepts efficiently. We developed the Hierarchical Overlapping Clustering (HOC) algorithm that modified the traditional Agglomerative Hierarchical Clustering algorithm to allow overlapped clusters at the same level in the concept hierarchy. The HOC algorithm represents the clustering result not by a tree but by a lattice to detect complex concepts. We developed a system that employs the HOC algorithm to carry out the goal of complex concept detection. This system operates in three phases; 1) the preprocessing of documents, 2) the clustering using the HOC algorithm, and 3) the validation of semantic hierarchical relationships among the concepts in the lattice obtained as a result of clustering. The preprocessing phase represents the documents as x-y coordinate values in a 2-dimensional space by considering the weights of terms appearing in the documents. First, it goes through some refinement process by applying stopwords removal and stemming to extract index terms. Then, each index term is assigned a TF-IDF weight value and the x-y coordinate value for each document is determined by combining the TF-IDF values of the terms in it. The clustering phase uses the HOC algorithm in which the similarity between the documents is calculated by applying the Euclidean distance method. Initially, a cluster is generated for each document by grouping those documents that are closest to it. Then, the distance between any two clusters is measured, grouping the closest clusters as a new cluster. This process is repeated until the root cluster is generated. In the validation phase, the feature selection method is applied to validate the appropriateness of the cluster concepts built by the HOC algorithm to see if they have meaningful hierarchical relationships. Feature selection is a method of extracting key features from a document by identifying and assigning weight values to important and representative terms in the document. In order to correctly select key features, a method is needed to determine how each term contributes to the class of the document. Among several methods achieving this goal, this paper adopted the $x^2$�� statistics, which measures the dependency degree of a term t to a class c, and represents the relationship between t and c by a numerical value. To demonstrate the effectiveness of the HOC algorithm, a series of performance evaluation is carried out by using a well-known Reuter-21578 news collection. The result of performance evaluation showed that the HOC algorithm greatly contributes to detecting and producing complex concepts by generating the concept hierarchy in a lattice structure.

• The Korean Journal of Nuclear Medicine
• /
• v.18 no.2
• /
• pp.17-23
• /
• 1984

For nearly a century it has been known that chemical activity accompanies mental activity, but only recently has it been possible to begin to examine its exact nature. Positron-emitting radioactive tracers have made it possible to study the chemistry of the human mind in health and disease, using chiefly cyclotron-produced radionuclides, carbon-11, fluorine-18 and oxygen-15. It is now well established that measurable increases in regional cerebral blood flow, glucose and oxygen metabolism accompany the mental functions of perception, cognition, emotion and motion. On May 25, 1983 the first imaging of a neuroreceptor in the human brain was accomplished with carbon-11 methyl spiperone, a ligand that binds preferentially to dopamine-2 receptors, 80% of which are located in the caudate nucleus and putamen. Quantitative imaging of serotonin-2, opiate, benzodiazapine and muscarinic cholinergic receptors has subsequently been accomplished. In studies of normal men and women, it has been found that dopamine and serotonin receptor activity decreases dramatically with age, such a decrease being more pronounced in men than in women and greater in the case of dopamine receptors than serotonin-2 receptors. Preliminary studies in patients with neuropsychiatric disorders suggests that dopamine-2 receptor activity is diminished in the caudate nucleus of patients with Huntington's disease. Positron tomography permits quantitative assay of picomolar quantities of neuro-receptors within the living human brain. Studies of patients with Parkinson's disease, Alzheimer's disease, depression, anxiety, schizophrenia, acute and chronic pain states and drug addiction are now in progress. The growth of any scientific field is based on a paradigm or set of ideas that the community of scientists accepts. The unifying principle of nuclear medicine is the tracer principle applied to the study of human disease. Nineteen hundred and sixty-three was a landmark year in which technetium-99m and the Anger camera combined to move the field from its latent stage into a second stage characterized by exponential growth within the framework of the paradigm. The third stage, characterized by gradually declining growth, began in 1973. Faced with competing advances, such as computed tomography and ultrasonography, proponents and participants in the field of nuclear medicine began to search for greener pastures or to pursue narrow sub-specialties. Research became characterized by refinements of existing techniques. In 1983 nuclear medicine experienced what could be a profound change. A new paradigm was born when it was demonstrated that, despite their extremely low chemical concentrations, in the picomolar range, it was possible to image and quantify the distribution of receptors in the human body. Thus, nuclear medicine was able to move beyond physiology into biochemistry and pharmacology. Fundamental to the science of pharmacology is the concept that many drugs and endogenous substances, such as neurotransmitters, react with specific macromolecules that mediate their pharmacologic actions. Such receptors are usually identified in the study of excised tissues, cells or cell membranes, or in autoradiographic studies in animals. The first imaging and quantification of a neuroreceptor in a living human being was performed on May 25, 1983 and reported in the September 23, 1983 issue of SCIENCE. The study involved the development and use of carbon-11 N-methyl spiperone (NMSP), a drug with a high affinity for dopamine receptors. Since then, studies of dopamine and serotonin receptors have been carried out in over 100 normal persons or patients with various neuropsychiatric disorders. Exactly one year later, the first imaging of opitate receptors in a living human being was performed [1].

• Journal of radiological science and technology
• /
• v.32 no.4
• /
• pp.361-370
• /
• 2009

The image quality management of bone mineral density is the responsibility and duty of radiologists who carry out examinations. However, inaccurate conclusions due to lack of understanding and ignorance regarding the methodology of image quality management can be a fatal error to the patient. Therefore, objective of this paper is to understand proper image quality management and enumerate methods for examiners and patients, thereby ensuring the reliability of bone mineral density exams. The accuracy and precision of bone mineral density measurements must be at the highest level so that actual biological changes can be detected with even slight changes in bone mineral density. Accuracy and precision should be continuously preserved for image quality of machines. Those factors will contribute to ensure the reliability in bone mineral density exams. Proper equipment management or control methods are set with correcting equipment each morning and after image quality management, a phantom, recommended from the manufacturer, is used for ten to twenty-five measurements in search of a mean value with a permissible range of ${\pm}1.5%$ set as standard. There needs to be daily measurement inspections on the phantom or at least inspections three times a week in order to confirm the existence or nonexistence of changes in values in actual bone mineral density. in addition, bone mineral density measurements were evaluated and recorded following the rules of Shewhart control chart. This type of management has to be conducted for the installation and movement of equipment. For the management methods of inspectors, evaluation of the measurement precision was conducted by testing the reproducibility of the exact same figures without any real biological changes occurring during reinspection. Bone mineral density inspection was applied as the measurement method for patients either taking two measurements thirty times or three measurements fifteen times. An important point when taking measurements was after a measurement whether it was the second or third examination, it was required to descend from the table and then reascend. With a 95% confidence level, the precision error produced from the measurement bone mineral figures came to 2.77 times the minimum of the biological bone mineral density change. The value produced can be stated as the least significant change (LSC) and in the case the value is greater, it can be stated as a section of genuine biological change. From the initial inspection to equipment moving and shifter, management must be carried out and continued in order to achieve the effects. The enforcement of proper quality control of radiologists performing bone mineral density inspections which brings about the durability extensions of equipment and accurate results of calculations will help the assurance of reliable inspections.

• The Korean Society of Law and Medicine
• /
• v.18 no.1
• /
• pp.297-341
• /
• 2017

We searched out court rulings on medical affairs through court library search sites and specialized articles on medically relevant judgments sentenced in 2016. And we selected and analyzed the judgements of the court we considered important as follows. In relation to the medical civil judgements, (1) In the case of applying surgery for female infertility during cesarean section operation but it has not been done, we expressed the regret for the lack of judgment in the process of entering the medical contract, introducing the rights infringed and the scope of compensation, (2) We pointed out that the ruling on the medical malpractice estimation goes out of limit of negligence estimation doctrine, and that the court asked very high degree duty of the traditional Korean medicine doctors to cooperate with Western medicine doctors. (3) In the case of admitting hospital's 100% responsibility, we pointed out the court overlooked the uncertainty and good intention of the medical practice. (4) Additionally, We introduced the cases admitted the hospital's responsibility in the accident related to the psychiatric patients in closed ward. Relating to a medical criminal ruling, we analyzed the supreme court decision about whether the dentist's Botox injection on the patient's face is a medical practice within the scope of the license from the viewpoint whether it is within the possible range of the word. And, concerning decisions on healthcare administration, (1) we analyzed the case about when medical personnel operate multiple medical institutions, whether it is possible to get back medical care costs under the National Health Insurance Law, (2) We commented on the ruling regarding explanation obligation in terms of object, degree, subject of explanation as a prerequisite for permissible arbitrary uninsured benefits. Finally, we reviewed the decision of the Constitutional Court about the Article 24 of the Mental Health Law, which it had allowed for a mental patient to be hospitalized forcibly by the consent of two guardians and a diagnosis of a psychiatrist. Also we indicated the problems of the revised Mental Health Law.

• Korean journal of communication and information
• /
• v.57
• /
• pp.33-48
• /
• 2012

Over the last decades, the issue of work or labor has played a critical role in prevailing discourses to represent the changed economic reality. Aesthetic labor, cultural work, network labor, team-work and alike, have played a dazzling role to represent the emerging economic order, employing the word of labor. Certainly, it is not less than a part of a wide range of shifts in order to make capital work with more effect by making up a workable and governable subject. In this article, I try to examine shifts around the media work which has contributed to expand the new discourse of 'labor.' I will say that it is quite crucial for accounting for the reality of media work to shed light on moves to represent media work, and, among others, one to transform the subjectivity involved in it among others. Furthermore, it would be necessary to take a close look at the subjectivity of media work and its modification to deal with and eliminate the precariousness of media work. Saying about media work without paying any attention to heterogenous and various practices to compose a media work, one is forced to regard media work as the matter of economic and legal interests. In addition, it would bring about that the cultural political concerns of media work will be detached from critical sight of the media cultural studies. Referring to major studies around media work in critical media studies, cultural studies and political economy of communication, this article will briefly look into the arrangement of contentions around subjectivity of media work in South Korea. And it will try to suggest what cultural-political strategy we need to investigate, fighting against the hegemonic power to generate and regulate media work and its workers in precarious conditions. It does not intend to search the media work and its complicated realities in detail in South Korea. I wish that it would make a preliminary step to propose and elaborate the critical analysis of media work and its form of subjectivities.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.45 no.1
• /
• pp.25-36
• /
• 2008

Eye localization means localization of the center of the pupils, and is necessary for face recognition and related applications. Most of eye localization methods reported so far still need to be improved about robustness as well as precision for successful applications. In this paper, we propose a robust eye localization method using multi-scale Gabor feature vectors without big computational burden. The eye localization method using Gabor feature vectors is already employed in fuck as EBGM, but the method employed in EBGM is known not to be robust with respect to initial values, illumination, and pose, and may need extensive search range for achieving the required performance, which may cause big computational burden. The proposed method utilizes multi-scale approach. The proposed method first tries to localize eyes in the lower resolution face image by utilizing Gabor Jet similarity between Gabor feature vector at an estimated initial eye coordinates and the Gabor feature vectors in the eye model of the corresponding scale. Then the method localizes eyes in the next scale resolution face image in the same way but with initial eye points estimated from the eye coordinates localized in the lower resolution images. After repeating this process in the same way recursively, the proposed method funally localizes eyes in the original resolution face image. Also, the proposed method provides an effective illumination normalization to make the proposed multi-scale approach more robust to illumination, and additionally applies the illumination normalization technique in the preprocessing stage of the multi-scale approach so that the proposed method enhances the eye detection success rate. Experiment results verify that the proposed eye localization method improves the precision rate without causing big computational overhead compared to other eye localization methods reported in the previous researches and is robust to the variation of post: and illumination.

• Journal of Intelligence and Information Systems
• /
• v.16 no.4
• /
• pp.195-211
• /
• 2010

An exhibition is defined as market events for specific duration to present exhibitors' main product range to either business or private visitors, and it also plays a key role as effective marketing channels. Especially, as the effect of the opinions of the visitors after the exhibition impacts directly on sales or the image of companies, exhibition organizers must consider various needs of visitors. To meet needs of visitors, ubiquitous technologies have been applied in some exhibitions. However, despite of the development of the ubiquitous technologies, their services cannot always reflect visitors' preferences as they only generate information when visitors request. As a result, they have reached their limit to meet needs of visitors, which consequently might lead them to loss of marketing opportunity. Recommendation systems can be the right type to overcome these limitations. They can recommend the booths to coincide with visitors' preferences, so that they help visitors who are in difficulty for choices in exhibition environment. One of the most successful and widely used technologies for building recommender systems is called Collaborative Filtering. Traditional recommender systems, however, only use neighbors' evaluations or behaviors for a personalized prediction. Therefore, they can not reflect visitors' dynamic preference, and also lack of accuracy in exhibition environment. Although there is much useful information to infer visitors' preference in ubiquitous environment (e.g., visitors' current location, booth visit path, and so on), they use only limited information for recommendation. In this study, we propose a booth recommendation methodology using Sequential Association Rule which considers the sequence of visiting. Recent studies of Sequential Association Rule use the constraints to improve the performance. However, since traditional Sequential Association Rule considers the whole rules to recommendation, they have a scalability problem when they are adapted to a large exhibition scale. To solve this problem, our methodology composes the confidence database before recommendation process. To compose the confidence database, we first search preceding rules which have the frequency above threshold. Next, we compute the confidences of each preceding rules to each booth which is not contained in preceding rules. Therefore, the confidence database has two kinds of information which are preceding rules and their confidence to each booth. In recommendation process, we just generate preceding rules of the target visitors based on the records of the visits, and recommend booths according to the confidence database. Throughout these steps, we expect reduction of time spent on recommendation process. To evaluate proposed methodology, we use real booth visit records which are collected by RFID technology in IT exhibition. Booth visit records also contain the visit sequence of each visitor. We compare the performance of proposed methodology with traditional Collaborative Filtering system. As a result, our proposed methodology generally shows higher performance than traditional Collaborative Filtering. We can also see some features of it in experimental results. First, it shows the highest performance at one booth recommendation. It detects preceding rules with some portions of visitors. Therefore, if there is a visitor who moved with very a different pattern compared to the whole visitors, it cannot give a correct recommendation for him/her even though we increase the number of recommendation. Trained by the whole visitors, it cannot correctly give recommendation to visitors who have a unique path. Second, the performance of general recommendation systems increase as time expands. However, our methodology shows higher performance with limited information like one or two time periods. Therefore, not only can it recommend even if there is not much information of the target visitors' booth visit records, but also it uses only small amount of information in recommendation process. We expect that it can give real?time recommendations in exhibition environment. Overall, our methodology shows higher performance ability than traditional Collaborative Filtering systems, we expect it could be applied in booth recommendation system to satisfy visitors in exhibition environment.

• Food Science and Preservation
• /
• v.15 no.3
• /
• pp.483-490
• /
• 2008

Several tartaric acid-degrading bacteria were isolated from Korean grape wine pomace after enrichment culture at $30^{\circ}C$ for 10 days in liquid media containing tartaric acid Among them, strains KMBL 5777 and KMBL 5778 exhibited the highest level in the growth and tartaric acid degradability in a medium containing 0.2%(w/v) tartaric acid as a sole carbon source. They were identified as Acetobacter tropicalis based on their morphological and physiological characteristics as well as their 16S rDNA sequences. Blast search of the 16S rDNA sequences revealed that the isolated strains are closest to Acetobacter tropicalis. Homologies of the sequences of KMBL 5777 and KMBL 5778 were 96.0 and 98.9%, respectively with those of A. tropicalis LMG 1663. Both the two bacteria showed higher tartaric acid degradation at $25^{\circ}C$ that those at 20 and $30^{\circ}C$. They could degrade tartaric acid at a wide range of pH between 4.0 and 7.0 with the most rapid degradability at pH 7.0. However, when the bacteria were grown for 8 days, the same level of tartaric acid degradation was observed at pH 4.0, 5.0, 6.0 and 7.0, which was 90.0% of degradation of the acid.