• Journal of Intelligence and Information Systems
• /
• v.18 no.2
• /
• pp.29-45
• /
• 2012

Bond rating is regarded as an important event for measuring financial risk of companies and for determining the investment returns of investors. As a result, it has been a popular research topic for researchers to predict companies' credit ratings by applying statistical and machine learning techniques. The statistical techniques, including multiple regression, multiple discriminant analysis (MDA), logistic models (LOGIT), and probit analysis, have been traditionally used in bond rating. However, one major drawback is that it should be based on strict assumptions. Such strict assumptions include linearity, normality, independence among predictor variables and pre-existing functional forms relating the criterion variablesand the predictor variables. Those strict assumptions of traditional statistics have limited their application to the real world. Machine learning techniques also used in bond rating prediction models include decision trees (DT), neural networks (NN), and Support Vector Machine (SVM). Especially, SVM is recognized as a new and promising classification and regression analysis method. SVM learns a separating hyperplane that can maximize the margin between two categories. SVM is simple enough to be analyzed mathematical, and leads to high performance in practical applications. SVM implements the structuralrisk minimization principle and searches to minimize an upper bound of the generalization error. In addition, the solution of SVM may be a global optimum and thus, overfitting is unlikely to occur with SVM. In addition, SVM does not require too many data sample for training since it builds prediction models by only using some representative sample near the boundaries called support vectors. A number of experimental researches have indicated that SVM has been successfully applied in a variety of pattern recognition fields. However, there are three major drawbacks that can be potential causes for degrading SVM's performance. First, SVM is originally proposed for solving binary-class classification problems. Methods for combining SVMs for multi-class classification such as One-Against-One, One-Against-All have been proposed, but they do not improve the performance in multi-class classification problem as much as SVM for binary-class classification. Second, approximation algorithms (e.g. decomposition methods, sequential minimal optimization algorithm) could be used for effective multi-class computation to reduce computation time, but it could deteriorate classification performance. Third, the difficulty in multi-class prediction problems is in data imbalance problem that can occur when the number of instances in one class greatly outnumbers the number of instances in the other class. Such data sets often cause a default classifier to be built due to skewed boundary and thus the reduction in the classification accuracy of such a classifier. SVM ensemble learning is one of machine learning methods to cope with the above drawbacks. Ensemble learning is a method for improving the performance of classification and prediction algorithms. AdaBoost is one of the widely used ensemble learning techniques. It constructs a composite classifier by sequentially training classifiers while increasing weight on the misclassified observations through iterations. The observations that are incorrectly predicted by previous classifiers are chosen more often than examples that are correctly predicted. Thus Boosting attempts to produce new classifiers that are better able to predict examples for which the current ensemble's performance is poor. In this way, it can reinforce the training of the misclassified observations of the minority class. This paper proposes a multiclass Geometric Mean-based Boosting (MGM-Boost) to resolve multiclass prediction problem. Since MGM-Boost introduces the notion of geometric mean into AdaBoost, it can perform learning process considering the geometric mean-based accuracy and errors of multiclass. This study applies MGM-Boost to the real-world bond rating case for Korean companies to examine the feasibility of MGM-Boost. 10-fold cross validations for threetimes with different random seeds are performed in order to ensure that the comparison among three different classifiers does not happen by chance. For each of 10-fold cross validation, the entire data set is first partitioned into tenequal-sized sets, and then each set is in turn used as the test set while the classifier trains on the other nine sets. That is, cross-validated folds have been tested independently of each algorithm. Through these steps, we have obtained the results for classifiers on each of the 30 experiments. In the comparison of arithmetic mean-based prediction accuracy between individual classifiers, MGM-Boost (52.95%) shows higher prediction accuracy than both AdaBoost (51.69%) and SVM (49.47%). MGM-Boost (28.12%) also shows the higher prediction accuracy than AdaBoost (24.65%) and SVM (15.42%)in terms of geometric mean-based prediction accuracy. T-test is used to examine whether the performance of each classifiers for 30 folds is significantly different. The results indicate that performance of MGM-Boost is significantly different from AdaBoost and SVM classifiers at 1% level. These results mean that MGM-Boost can provide robust and stable solutions to multi-classproblems such as bond rating.

• Journal of the Microelectronics and Packaging Society
• /
• v.24 no.4
• /
• pp.23-29
• /
• 2017

The packaged optical fiber Bragg grating sensors which were networked by multiplexing the Bragg grating sensors with WDM technology were investigated in application for the structural health monitoring of the marine trestle structure transporting the ship. The optical fiber Bragg grating sensor was packaged in a cylindrical shape made of aluminum tubes. Furthermore, after the packaged optical fiber sensor was inserted in polymeric tube, the epoxy was filled inside the tube so that the sensor has resistance and durability against sea water. The packaged optical fiber sensor component was investigated under 0.2 MPa of hydraulic pressure and was found to be robust. The number and location of Bragg gratings attached at the trestle were determined where the trestle was subject to high displacement obtained by the finite element simulation. Strain of the part in the trestle being subjected to the maximum load was analyzed to be ${\sim}1000{\mu}{\varepsilon}$ and thus shift in Bragg wavelength of the sensor caused by the maximum load of the trestle was found to be ~1,200 pm. According to results of the finite element analysis, the Bragg wavelength spacings of the sensors were determined to have 3~5 nm without overlapping of grating wavelengths between sensors when the trestle was under loads and thus 50 of the grating sensors with each module consisting of 5 sensors could be networked within 150 nm optical window at 1550 nm wavelength of the Bragg wavelength interrogator. Shifts in Bragg wavelength of the 5 packaged optical fiber sensors attached at the mock trestle unit were well interrogated by the grating interrogator which used the optical fiber loop mirror, and the maximum strain rate was measured to be about $235.650{\mu}{\varepsilon}$. The modelling result of the sensor packaging and networking was in good agreements with experimental result each other.

• Nuclear Medicine and Molecular Imaging
• /
• v.42 no.5
• /
• pp.354-361
• /
• 2008

Purpose: N-13 ammonia uptake and retention in the myocardium is related to perfusion and metabolism. There are several potential advantages of N-13 ammonia positron emission tomography (PET) to detect myocardial ischemia, such as higher spatial resolution, greater counting efficiencies, and robust attenuation correction. But there are few reports comparing Tc-99m myocardial perfusion single photon emission tomography (MPS) and N-13 ammonia PET. We thus compared adenosine stress N-13 ammonia PET/CT and Tc-99m sestamibi MPS in patients with suspected coronary artery stenosis. Materials and Methods: Seventeen patients (male 13 : $63{\pm}11$ years old) underwent adenosine stress N-13 ammonia PET/CT (Discovery ST, GE), Tc-99m sestamibi MPS (dual head gamma camera, Hawkeye, GE) and coronary angiography within 1 week. N-13 ammonia PET/CT and Tc-99m sestamibi MPS images were assessed with a 20-segment model by visual interpretation and quantitative analysis using automatic quantitative software (Myovation, GE). Results: Both sensitivities and specificities of detecting an individual coronary artery stenosis were higher for N-13 ammonia PET/CT than Tc-99m sestamibi MPS (PET/CT: 91%/89% vs MPS: 65%/82%). N-13 ammonia PET/CT showed reversibility in 52% of segments that were considered non-reversibile by Tc-99m sestamibi MPS. In the 110 myocardial segments supplied by the stenotic coronary artery, N-13 ammonia PET/CT showed higher count densities than Tc-99m MPS on rest study (p < 0.01), and the difference of count density between the stress and the rest studies was also larger on N-13 ammonia PET/CT. Conclusion: Adenosine stress N-13 ammonia PET/CT had higher diagnostic sensitivity and specificity, more reversibility of perfusion defects and greater stress/rest uptake differences than Tc-99m sestamibi MPS. Accordingly, N-13 ammonia PET/CT might offer better assessment of myocardial ischemia and viability.

• Korean Journal of Ecology and Environment
• /
• v.33 no.3 s.91
• /
• pp.187-196
• /
• 2000

This paper describes a simple method that uses differences among Carlson's (1977) trophic state index (TSI) values based on total phosphorus (TP), chlorophyll a (CHL) and Secchi depth (SD) to draw inferences regarding the factors that are limiting to phytoplankton growth and the composition of lake seston. Examples are provided regarding seasonal and spatial patterns in a large subtropical lake (Lake Okeechobee, Florida, USA) and inter- and intra-lake variations from a multilake data set developed from published studies. Once an investigator has collected routine water quality data and established TSI values based on TP, CHL, and SD, a number of inferences can be made. Additional information can be provided where it also is possible to calculate a TSI based on total nitrogen (TN). Where TSI (CHL)<>TSI (SD), light attenuating particles are large (large filaments or colonies of algae), and the phytoplankton may be limited by zooplankton grazing. Other limiting conditions are inferred by different relationships between the TSI values. Results of this study indicate that the analysis is quite robust, and that it generally gives good agreement with conclusions based on more direct methods (e.g., nutrientaddition bioassays, zooplankton size data, zooplankton removal experiments). The TSI approach, when validated periodically with these more costly and time-intensive methods, provides an effective, low cost method for tracking long-term changes in pelagic structure and function with potential value in monitoring lake ecology and responses to management.

• Progress in Medical Physics
• /
• v.13 no.4
• /
• pp.209-217
• /
• 2002

There have been many studies on the application of the reciprocal advantages of multimodality image to define accurate target volume in the Process of radiation treatment planning. For the proper use of the multimodality images, the registration works between different modality images should be performed in advance. In this study, we selected chamfer matching method and mutual information method as most popular methods in recent image registration studies considering the registration accuracy and clinical practicality. And the two registration methods were analyzed to deduce the optimal registration method according to the characteristics of images. Lung phantom of which multimodality images could be acquired was fabricated and CT, MRI and SPECT images of the phantom were used in this study. We developed the registration program which can perform the two registration methods properly and analyzed the registration results which were produced by the developed program in many different images' conditions. Although the overall accuracy of the registration in both chamfer matching method and mutual information method was acceptable, the registration errors in SPECT images which had lower resolution and in degraded images of which data were removed in some part were increased when chamfer matching method was applied. Especially in the case of degraded reference image, chamfer matching methods produce relatively large errors compared with mutual information method. Mutual information method can be estimated as more robust registration method than chamfer matching method in this study because it did not need the prerequisite works, the extraction of accurate contour points, and it produced more accurate registration results consistently regardless of the images' characteristics. The analysis of the registration methods in this study can be expected to provide useful information to the utilization of multimodality images in delineating target volume for radiation treatment planning and in many other clinical applications.

• Journal of Intelligence and Information Systems
• /
• v.19 no.4
• /
• pp.97-121
• /
• 2013

With the growing importance of knowledge for sustainable competitive advantages and innovation in a volatile environment, many researches on knowledge sharing have been conducted. However, previous researches have mostly relied on the questionnaire survey which has inherent perceptive errors of respondents. The current research has drawn the relationship among primary participant behaviors towards the participation level in knowledge sharing, basically from online user behaviors on Wikipedia, a representative community for online knowledge collaboration. Without users' participation in knowledge sharing, knowledge collaboration for creating knowledge cannot be successful. By the way, the editing patterns of Wikipedia users are diverse, resulting in different revisiting periods for the same number of edits, and thus varying results of shared knowledge. Therefore, we illuminated the participation level of knowledge sharing from two different angles of number of edits and revisiting period. The behavioral dimensions affecting the level of participation in knowledge sharing includes the article talk for public discussion and user talk for private messaging, and community registration, which are observable on Wiki platform. Public discussion is being progressed on article talk pages arranged for exchanging ideas about each article topic. An article talk page is often divided into several sections which mainly address specific type of issues raised during the article development procedure. From the diverse opinions about the relatively trivial things such as what text, link, or images should be added or removed and how they should be restructured to the profound professional insights are shared, negotiated, and improved over the course of discussion. Wikipedia also provides personal user talk pages as a private messaging tool. On these pages, diverse personal messages such as casual greetings, stories about activities on Wikipedia, and ordinary affairs of life are exchanged. If anyone wants to communicate with another person, he or she visits the person's user talk page and leaves a message. Wikipedia articles are assessed according to seven quality grades, of which the featured article level is the highest. The dataset includes participants' behavioral data related with 2,978 articles, which have reached the featured article level, with editing histories of articles, their article talk histories, and user talk histories extracted from user talk pages for each article. The time period for analysis is from the initiation of articles until their promotion to the featured article level. The number of edits represents the total number of participation in the editing of an article, and the revisiting period is the time difference between the first and last edits. At first, the participation levels of each user category classified according to behavioral dimensions have been analyzed and compared. And then, robust regressions have been conducted on the relationships among independent variables reflecting the degree of behavioral characteristics and the dependent variable representing the participation level. Especially, through adopting a motivational theory adequate for online environment in setting up research hypotheses, this work suggests a theoretical framework for the participation level of online knowledge sharing. Consequently, this work reached the following practical behavioral results besides some theoretical implications. First, both public discussion and private messaging positively affect the participation level in knowledge sharing. Second, public discussion exerts greater influence than private messaging on the participation level. Third, a synergy effect of public discussion and private messaging on the number of edits was found, whereas a pretty weak negative interaction effect of them on the revisiting period was observed. Fourth, community registration has a significant impact on the revisiting period, whereas being insignificant on the number of edits. Fifth, when it comes to the relation generated from private messaging, the frequency or depth of relation is shown to be more critical than the scope of relation for the participation level.

• Journal of Korean Orthopaedic Sports Medicine
• /
• v.1 no.1
• /
• pp.79-88
• /
• 2002

Purpose: In this prospective, randomized study, we compared the results of early motion versus conventional immobilization after the arthroscopic Bankart repair. Materilal and Methods : We performed an arthroscopic Bankart repair using suture anchors in 62 patients with traumatic anterior shoulder instability and randomized them into two groups; Group 1 (n=28; mean age, 28 years) underwent three-week of immobilization and conventional rehabilitation program, while Group 2 (n=34; mean age, 29 years) underwent an accelerated rehabilitation program with staged range of motion and strengthening exercises starting from the immediate postoperative day. Selected patients were non-athletes with a classic Bankart lesion and a robust labrum. Analysis of outcome included pain scores (6-week and follow-up: 31(9 months), range of motion, return to activity, recurrence, patients’ satisfaction with each program, and shoulder scores (ASES, UCLA, and Rowe). Results : The recurrent rate was not different between the two groups (2 anterior apprehension from each group) (p=0.842). Patients with accelerated rehabilitation resumed functional range-of-motion faster and returned earlier to the functional level of activity (p<0.05). Accelerated rehabilitation decreased postoperative pain and more patients were satisfied with this program (p<0.05). No differences were found between the two groups at the follow-up with regards to the shoulder scores, return to activity, pain score, and the range-of-motion. Conclusions : Early mobilization after arthroscopic Bankart repair does not increase the recurrence rate in selected patients. Although the final outcomes are similar in both groups, the accelerated rehabilitation program promotes functional recovery and reduces postoperative pain, which enables patients an early institution of desired activities.

• Journal of Technology Innovation
• /
• v.17 no.2
• /
• pp.55-80
• /
• 2009

Technology innovation activity plays a pivotal role in constructing the entrance barrier for other firms and making process improvement and new product. and these activities give a profit increase and growth to firms. Thus, technology innovation activity can reduce the default risk of firms. However, technology innovation activity can also increase the firm's default risk because technology innovation activity requires too much investment of the firm's resources and has the uncertainty on success. The purpose of this study is to examine the effect of technology innovation activity on the default risk of firms. This study's sample consists of manufacturing firms listed on the Korea Securities Market and The Kosdaq Market from January 1,2000 to December 31, 2008. This study makes use of R&D intensity as an proxy variable of technology innovation activity. The default probability which proxies the default risk of firms is measured by the Merton's(l974) debt pricing model. The main empirical results are as follows. First, from the empirical results, it is found that technology innovation activity has a negative and significant effect on the default risk of firms independent of the Korea Securities Market and Kosdaq Market. In other words, technology innovation activity reduces the default risk of firms. Second, technology innovation activity reduces the default risk of firms independent of firm size, firm age, and credit score. Third, the results of robust analysis also show that technology innovation activity is the important factor which decreases the default risk of firms. These results imply that a manager must show continuous interest and investment in technology innovation activity of one's firm. And a policymaker also need design an economic policy to promote the technology innovation activity of firms.

• Journal of Korea Water Resources Association
• /
• v.54 no.10
• /
• pp.835-848
• /
• 2021

In Jeju Island which has peculiarity for its geological features and hydrology system, hydrological factor analysis for the effective water management is necessary. Because in-situ hydro-meteorological data is affected by surrounding environment, the in-situ dataset could not be the spatially representative for the study area. For this reason, remote sensing data may be used to overcome the limit of the in-situ data. In this study, applicability assessment of MOD16 evapotranspiration data, Globas Land Data Assimilation System (GLDAS) based evapotranspiration/soil moisture data, and Advanced SCATterometer (ASCAT) soil moisture product which were evaluated their applicability on other study areas was conducted. In the case of evapotranspiration, comparison with total precipitation and flux-tower based evapotranspiration were conducted. And for soil moisture, 6 in-situ data and ASCAT soil moisture product were compared on each site. As a result, 57% of annual precipitation was calculated as evapotranspiration, and the correlation coefficient between MOD16 evapotranspiration and GLDAS evapotranspiration was 0.759, which was a robust value. The correlation coefficient was 0.434, indicating a relatively low fit. In the case of soil moisture, in the case of the GLDAS data, the RMSE value was less than 0.05 at all sites compared to the in-situ data, and a statistically significant result was obtained as a result of the significance test of the correlation coefficient. However, for satellite data, RMSE over than 0.05 were found at Wolgak and there was no correlation at Sehwa and Handong points. It is judged that the above results are due to insufficient quality control and spatial representation of the evapotranspiration and soil moisture sensors installed in Jeju Island. It is estimated as the error that appears when adjacent to the coast. Through this study, the necessity of improving the existing ground observation data of hydrometeorological factors is emphasized.

• (The) Research of the performance art and culture
• /
• no.34
• /
• pp.9-42
• /
• 2017

Of all Geommu(劍舞, Korean sword dance), Gisaeng-Geommu danced by gisaeng(妓生, Korean female entertainers) for private guests and at the royal court. The Sword dance in the late Joseon Dynasty used to be dynamic exuding menacing "sword spirit(劍氣)." Sword dance being transmitted today is more ritualistic and elegant. This study considers Korean sword dance has a core structure and motifs transcending generational differences, and based on this critical thinking, aims to analyze the structure of Korean sword dance. This study analyzed the prose "Geommugi(劍舞記)" by Park Je-ga(朴齊家) and the poem "Mugeompyeonjeongmiin(舞劍篇贈美人)" by Jeong Yak-yong(丁若鏞) out of literary works from the late Joseon Dynasty, and from official records of rituals(笏記), "Geomgimu(Sword Dance, 劍器舞)" and "Geommu" in "Gyobanggayo(敎坊歌謠)." In the introduction part of Sword dance, a dancer appears, bows and performs a hand dance or hansam(汗衫) dance to and fro. In the development part, a dancer meets with a sword but first hesitates to hold it and dances holding swords in both hands. The climax shows expert sword skills and combat scenes. In the conclusion part, the court dance involves a dancer bidding a formal farewell, while the dance for entertainment, a dance throws away the sword to finish. From literature materials, the structure of Korean sword dance could be divided into an introduction, a development, a climax and a conclusion. Based on this, this study analyzed sword dance movements by linking the beats accompanying the current sword dance, in the order of a Yeombul, the traditional Korean ballad Taryeong or Neujeun Taryeong, Jajin Taryeong, Taryeong and Jajin Taryeong. The introduction part includes a Buddhist prayer and the beginning of Taryeong. Dancers appear, and in two rows they dance facing each other. On the slow beat, their dances are relaxed and elegant. The development part is matched with Jajin Taryeong. Dancers sit in front of swords and grab them, and they dance holding a pair of swords. The beat gradually becomes faster, progressing the development of the dance. But then, the slower Taryeong is placed again. The reason behind it is to create a tension for a little while, before effectively reaching a climax by speeding up the tempo again. Moving on to Jajin Taryeong, dancers' movements are bigger and more dynamic. The highly elated Jajin Taryeong shows dance movements at the climax on fast, robust beats. In the conclusion part, the beat is quick-tempo and on the upbeat again on Jajin Taryeong. Driving on without a stop on the exciting Yeonpungdae(燕風臺) melody, dancers standing in a line dance wielding the swords and bow before finishing.