• Asian Pacific Journal of Cancer Prevention
• /
• v.16 no.17
• /
• pp.7595-7597
• /
• 2015

Background: The aim of this study was to evaluate the effect of whole brain radiotherapy (WBRT) combined with streotactic radiosurgery versus stereotactic radiosurgery (SRS) alone for patients with brain metastases. Materials and Methods: This was a retrospective study that evaluated the results of 46 patients treated for brain metastases at Dr. Abdurrahman Yurtaslan Ankara Oncology Training and Research Hospital, Radiation Oncology Department, between January 2012 and January 2015. Twenty-four patients were treated with WBRT+SRS while 22 patients were treated with only SRS. Results: Time to local recurrence was 9.7 months in the WBRT+SRS arm and 8.3 months in SRS arm, the difference not being statistically significant (p=0.7). Local recurrence rate was higher in the SRS alone arm but again without significance (p=0,06). Conclusions: In selected patient group with limited number (one to four) of brain metastases SRS alone can be considered as a treatment option and WBRT may be omitted in the initial treatment.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.9
• /
• pp.445-456
• /
• 2017

In this paper, we propose a method to detect spam tweets containing unhealthy information by using an n-gram dictionary under limited labeling. Spam tweets that contain unhealthy information have a tendency to use similar words and sentences. Based on this characteristic, we show that spam tweets can be effectively detected by applying a Naive Bayesian classifier using n-gram dictionaries which are constructed from spam tweets and normal tweets. On the other hand, constructing an initial training set requires very high cost because a large amount of data flows in real time in a twitter. Therefore, there is a need for a spam detection method that can be applied in an environment where the initial training set is very small or non exist. To solve the problem, we propose a method to generate pseudo-labels by utilizing twitter's retweet function and use them for the configuration of the initial training set and the n-gram dictionary update. The results from various experiments using 1.3 million korean tweets collected from December 1, 2016 to December 7, 2016 prove that the proposed method has superior performance than the compared spam detection methods.

• Journal of Power Electronics
• /
• v.17 no.4
• /
• pp.1014-1026
• /
• 2017

The traditional fault diagnosis method for photovoltaic (PV) inverters has a difficult time meeting the requirements of the current complex systems. Its main weakness lies in the study of nonlinear systems. In addition, its diagnosis time is long and its accuracy is low. To solve these problems, a hidden Markov model (HMM) is used that has unique advantages in terms of its training model and its recognition for diagnosing faults. However, the initial value of the HMM has a great influence on the model, and it is possible to achieve a local minimum in the training process. Therefore, a genetic algorithm is used to optimize the initial value and to achieve global optimization. In this paper, the HMM is combined with a genetic algorithm (GHMM) for PV inverter fault diagnosis. First Matlab is used to implement the genetic algorithm and to determine the optimal HMM initial value. Then a Baum-Welch algorithm is used for iterative training. Finally, a Viterbi algorithm is used for fault identification. Experimental results show that the correct PV inverter fault recognition rate by the HMM is about 10% higher than that of traditional methods. Using the GHMM, the correct recognition rate is further increased by approximately 13%, and the diagnosis time is greatly reduced. Therefore, the GHMM is faster and more accurate in diagnosing PV inverter faults.

• Proceedings of KOSOMES biannual meeting
• /
• 2017.04a
• /
• pp.38-38
• /
• 2017

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.27 no.2
• /
• pp.289-297
• /
• 2009

In the paper, we propose the methodology to extract training dataset automatically for supervised classification of road networks. For the preprocessing, we co-register the airborne photos, LIDAR data and large-scale digital maps and then, create orthophotos and intensity images. By overlaying the large-scale digital maps onto generated images, we can extract the initial training dataset for the supervised classification of road networks. However, the initial training information is distorted because there are errors propagated from registration process and, also, there are generally various objects in the road networks such as asphalt, road marks, vegetation, cars and so on. As such, to generate the training information only for the road surface, we apply the Expectation Maximization technique and finally, extract the training dataset of the road surface. For the accuracy test, we compare the training dataset with manually extracted ones. Through the statistical tests, we can identify that the developed method is valid.

• Korean Journal of Applied Biomechanics
• /
• v.28 no.3
• /
• pp.151-158
• /
• 2018

Objective: In this study, we analyzed kinematic changes in the start phase of speed skating before and after physical training. Method: We introduced a new strength training program (2017) that was improved in terms of exercise type and intensity [%, one repetition maximum (1RM)] compared with the previous strength training program (2016). The new program was applied to elite speed skating athletes (four males and four females). To determine the improvement in starting technique, we recorded race images during the start phase of the 500-m race held in 2016 and 2017. The race images were collected using five high-speed cameras and kinematic characteristics of the start phase were analyzed by three-dimensional image analysis. Results: The 1RMs were improved by 11% on an average after the strength training. In 2017, records of four out of the eight athletes were shortened in terms of the initial lap time (100 m), and 500-m records were shortened in six athletes. The time to nine strokes was shortened in five athletes, and the ratio of correct kinetic chain was increased or maintained at a high level in six athletes. Conclusion: In this study, the new strength training program (2017), applied to elite speed skating athletes, showed a positive effect on starting technique and reduced the record times.

• Journal of the Korea Safety Management & Science
• /
• v.18 no.3
• /
• pp.91-97
• /
• 2016

Currently many geriatric hospitals have been built in Korea because younger people don't want to care their parents and have been decreasing for labor. However, the increasing geriatric hospitals make the increasing fire safety accidents. Therefore, this study is conducted by survey and face-to-face talk for analyzing fire safety problem of twelve among 15 geriatric hospitals in the north of Chung-Buk area. The result of this study is that infection and fall accident are higher than others and fire safety implement rate of safety rule followed by CEO is 71%. Monthly safety training rate is 49% and initial fire safety training not conducted is 33%. Yearly outside fire safety training rate is 97% but workers who know how to use fire evacuation facility are 61%. Furthermore, safety instruction rate of fire safety manager is much higher than supervisor's safety instruction. The cause of accident is facility (33%). In conclusion, the institution and rule improvement need for decreasing infection and falling, increasing implement level of fire safety rule and fire safety training, participation rate of supervisor for fire safety, quality of fire safety training, and investment of fire safety facility.

• Journal of Institute of Control, Robotics and Systems
• /
• v.21 no.8
• /
• pp.709-712
• /
• 2015

This paper will introduce the development case study about virtual-reality-based operation training system for steel making process. Steel making process consist iron making process to create liquid steel, pig iron, by reduction process, steel making process to make molten steel by refining, continuous casting process to make slab, and rolling process to make final product like coil, plate. This steel making process deals with liquid and solid products, so facilities of steel making process are very various and complicated. In addition, according to various customer requirements, the recycle of facilities and recipes changing have been fast. So the training for skilled operators is very important point. In this paper, we develop steel making training system based virtual reality for training skilled operator. This system consists of virtual machine, virtual HMI, and virtual control panel. And for fitting the characteristics of each process and increasing the education effectiveness, we develop dynamic methods like the method of dynamic education system configuration, initial facilities setup operation education system, and etc.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.1C
• /
• pp.113-119
• /
• 2009

The training sequence with the appropriate length for equalization and initial synchronization is necessary before sending the pure data in the burst transmission type DS-UWB system. The length of the training sequence is one of the factors which make throughput decreased. The noble structure with the variable length of the training sequence whose length can be adaptively tailored according to the channel conditions (CM1,CM2,CM3,CM4) in the DS-USB systems is proposed. This structure can increase the throughput without sacrificing the performance than the method with fixed length of training sequence considering the worst case channel conditions. Simulation results under IEEE 802.15.3a channel model show that the proposed scheme can achieve higher throughput than a conventional one with the slight loss of BER performance. And this structure can reduce the computation complexity and power consumption with selecting the short length of the training sequence.

• Proceedings of the KSPS conference
• /
• 2007.05a
• /
• pp.183-186
• /
• 2007

This study analyzes Korean high school students' pronunciation errors in word-initial onglides in English. For this study, 24 Korean high school students read 34 English words including glide-vowel sequences in word-initial positions and vowel-initial words in a frame sentence. The results showed 2 different error types: glide deletion and vowel distortion. After the analysis of the first recording, the subjects were taught how to pronounce glide-vowel sequences properly in a 60-minute class. Comparison of the analyses of the first and second recordings showed that the subjects improved on the pronunciation of glide-vowel sequences. After the training, the pronunciation errors of diphthongs unique to English, [$j_I$], decreased substantially. However, most subjects still had difficulties in pronouncing [$w{\mho}$], [wu], and [wo]. There was no significant correlation between English course grade and error reduction.