• Journal of the Korea Society of Computer and Information
• /
• v.25 no.10
• /
• pp.15-22
• /
• 2020

In this study, we propose an improvement method that can create U-Net model which detect fine concrete cracks by applying a weighted loss function. Because cracks in concrete are a factor that threatens safety, it is important to periodically check the condition and take prompt initial measures. However, currently, the visual inspection is mainly used in which the inspector directly inspects and evaluates with naked eyes. This has limitations not only in terms of accuracy, but also in terms of cost, time and safety. Accordingly, technologies using deep learning is being researched so that minute cracks generated in concrete structures can be detected quickly and accurately. As a result of attempting crack detection using U-Net in this study, it was confirmed that it could not detect minute cracks. Accordingly, as a result of verifying the performance of the model trained by applying the suggested weighted loss function, a highly reliable value (Accuracy) of 99% or higher and a harmonic average (F1_Score) of 89% to 92% was derived. The performance of the learning improvement plan was verified through the results of accurately and clearly detecting cracks.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.11 no.3
• /
• pp.181-188
• /
• 2011

With the rapid proliferation of wireless networks and mobile computing applications, the landscape of the network security has greatly changed recently. Especially, Vehicular Ad Hoc Networks maintaining network topology with vehicle nodes of high mobility are self-organizing Peer-to-Peer networks that typically have short-lasting and unstable communication links. VANETs are formed with neither fixed infrastructure, centralized administration, nor dedicated routing equipment, and vehicle nodes are moving, joining and leaving the network with very high speed over time. So, VANET-security is very vulnerable for the intrusion of malicious and misbehaving nodes in the network, since VANETs are mostly open networks, allowing everyone connection without centralized control. In this paper, we propose a weighted intrusion detection method using rough set that can identify malicious behavior of vehicle node's activity and detect intrusions efficiently in VANETs. The performance of the proposed scheme is evaluated by a simulation study in terms of intrusion detection rate and false alarm rate for the threshold of deviation number ${\epsilon}$.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.36 no.12
• /
• pp.1163-1170
• /
• 2008

A mission scheduling optimization algorithm according to the purpose of satellite operations is developed using genetic algorithm. Satellite mission scheduling is making a timetable of missions which are slated to be performed. It is essential to make an optimized timetable considering related conditions and parameters for effective mission performance. Thus, as important criterions and parameters related to scheduling vary with the purpose of satellite operation, those factors should be fully considered and reflected when the satellite mission scheduling algorithm is developed. The developed algorithm in this study is implemented and verified through a comprehensive simulation study. As a result, it is shown that the algorithm can be applied into various type of the satellite mission operations.

• The Journal of the Acoustical Society of Korea
• /
• v.13 no.4
• /
• pp.40-49
• /
• 1994

In this paper, we propose methods for improving the performance of word recognition system. The ray stratey of the first method is to apply the inertia to the feature vector sequences of speech signal to stabilize the transitions between VQ cdoes. The second method is generating the new observation probabilities using the transition matrix of VQ codes as weights at the observation probability of the output symbol, so as to take into account the time relation between neighboring frames in DHMM. By applying the inertia to the feature vector sequences, we can reduce the overlapping of probability distribution of the response paths for each word and stabilize state transitions in the HMM. By using the transition matrix of VQ codes as weights in conventional DHMM. we can divide the probability distribution of feature vectors more and more, and restrict the feature distribution to a suitable region so that the performance of recognition system can improve. To evaluate the performance of the proposed methods, we carried out experiments for 50 DDD area names. As a result, the proposed methods improved the recognition rate by $4.2\%$ in the speaker-dependent test and $12.45\%$ in the speaker-independent test, respectively, compared with the conventional DHMM.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.3
• /
• pp.260-266
• /
• 2018

The concept of customer utility, which affects customer's adoption, is important to understand the process of technology diffusion and substitution regarding multiple generation technology. This research defined the concept of attribute-based customer utility and developed a model for measuring attribute-based customer utility. Based on the literature review and modeling, we provided the definition and a model regarding customer utility and the accuracy of the model is verified through a case study of the semiconductor industry. Customer utility for a multiple generation technology needs to consider changes by generation, or time within the same generation, and is defined as the summation of both technological and economic utilities. In addition, we can model the measurement of customer utility after converting technological and economical attributes into utilities. This research is valuable in understanding not only customer utility as a driver of customer adoption, but also for establishing technological strategy after forecasting diffusion and substitution paths based on customer utility.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.10
• /
• pp.2187-2193
• /
• 2010

A specific radar system can be implemented more easily using the frequency modulated continuous wave comparing with the pulse Doppler radar. It also has the advantage of LPI (low probability of interception) because of the low power and wide bandwidth characteristics. These radars are usually used to cover the short range area and to obtain the high resolution measurements of the target range and velocity information. The transmitted waveform is used in the mixer to demodulate the received echo signal and the resulting beat signal can be obtained. This beat signal is analyzed using the FFT method for the purpose of clutter removal, detection of a target, extraction of velocity and range information, etc. However, for the case of short signal acquisition time, this FFT method can cause the serious leakage effect which disables the detection of weaker echo signals masked by strong side lobes of the clutter. Therefore, in this paper, the weighting window method is analyzed to suppress the strong side lobes while maintaining the proper main lobe width. Also, the results of FFT beat spectrum analysis are shown under various environments.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.8
• /
• pp.3785-3793
• /
• 2011

The purpose of this study is to devise the overall problems (such as, rights and obligations between users, owners, managers regarding commercial rental buildings, benefit/benefit in return, and the relevant regulations and systems) of the building maintenance system through field surveys, expert interviews, and literature reviews. In this rapidly changing society, there are a huge amount of needs for various reforms which can meet the diversity in each area, so it is greatly important to choose priority for the best improvement for efficient and economic expenses due to limited time, human resources and budgets. In this regard, the study extracted the evaluation elements from the devised problems with the viewpoint of BSC, calculated the weights through application of AHP to the extracted evaluation elements, evaluated the difference between before/after application of weights, and set the status, which makes it possible to realize the effective solutions and the reasonable expenses for all parties. By doing so, it is expected that the building maintenance system which maximizes the value and guarantees the user's convenience with smart functions of the building, will be realized in the near future.

• Journal of KIISE:Software and Applications
• /
• v.34 no.2
• /
• pp.160-168
• /
• 2007

For developing a text classifier using supervised learning, a manually labeled corpus of large size is required. However, it takes a lot of time and human effort. Recently a research paradigm was proposed to use a raw corpus and a small amount of seed information instead of manually labeled corpus. In this paper we introduce an unsupervised learning method that makes it possible to achieve better performance than other related works. The characteristics of our approach is that average mutual information is used to learn representative words and their weights and then update of the weights is done using a technique inspired by the works in information retrieval. By iterating this teaming process it was shown that a high performance system can be developed.

• The KIPS Transactions:PartB
• /
• v.9B no.4
• /
• pp.471-480
• /
• 2002

In this paper, we implemented a real-time system improving the performance of the paper currency discrimination by integrating a weighted region of interest matching algorithm with a weighted shape feature matching algorithm of the blocked image. The system classifies the paper currency by comparing a query image with compared images based on the database that contain images of paper currency. Especially, the system has good efficiency at the contaminated, rotated, and translated paper currency. The system hardware consists of three parts as follows : the paper currency image acquired by CIS(contact image sensor) is applied to the pre-processing part with A/D converter and PLD. Finally the pre-processed image data are classified by the main image processing part with a high-speed DSP based on the proposed algorithm.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2008.05a
• /
• pp.1007-1012
• /
• 2008

유역통합수자원관리의 시작은 기상예측정보의 제공으로부터 시작된다. 하지만, 기상예측정보는 단기, 중기, 장기로 구분되며, 제공되는 정보가 수자원 운영에 필요한 정보와 시간적으로나 공간적으로 차이가 나며, 가공에 많은 전문가들의 노력이 필요하여 실무에서의 적용에 많은 어려움이 따른다. 따라서 본 연구에서는 이러한 문제들을 해결하고 용이하게 수자원 운영자에게 필요한 기상정보를 적절한 형태의 가공을 통하여 자동적으로 제공해 주는데 그 목적이 있다. 이러한 시스템의 구축을 통해 향후 수자원 운영에 있어 필수적인 의사결정 정보를 제공해 주어 수자원의 이용효율을 높이고자 한다. 구축된 시스템은 금강 유역에 대해 소유역단위로 장기 유출의 입력자료인 일단위 예측 강수를 30일간 제공하도록 시스템을 구축하였다. 단기(1일$\sim$2일)에는 RDAPS의 모의 결과인 Grib파일을 자동 추출하여 예측 강수를 제공한다. 1일에 두 번 모의되는 RDAPS의 결과를 일단위로 제공하기 위해 여러 가지 case별 분석을 실시하여 가장 적합한 기법을 이용하여 일단위 시계열을 구축하는 시스템을 설계하였다. 중기(3일$\sim$10일)에는 GDAPS 결과인 Grib파일을 자동 추출하여 유역단위 시계열을 구축한 뒤 과거 자료를 이용한 연 평균 자료를 이용하여 가중치를 곱하여 시계열을 구축하였다. 장기(11일$\sim$30일) 시계열의 구축을 위해서는 단기 및 중기 예측 시계열을 이용하여 과거 시계열 자료와의 통계적 비교 분석을 이용하여 유사 시계열을 추출한 후 과거 자료에 대한 평균값과 기상 전망을 이용하여 가중치를 부여하는 방법 등을 이용하여 시스템을 구축하였다. 본 시스템은 한국수자원공사에서 운영 중인 RRFS모형의 입력 자료를 자동 생성할 수 있는 기능을 제공하도록 설계되었다. 이러한 시스템의 구축을 통해 기상정보를 다루는데 익숙하지 않은 수자원 운영자들에게 비교적 용이하게 유역단위 기상예측 정보를 추출하는데 큰 도움이 될 것으로 기대한다.