• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.540-543
• /
• 2020

기존의 초해상도 딥러닝 기법은 모델의 깊이가 깊어지면서, 좋은 성능을 내지만 점점 더 복잡해지고 있고, 실제로 사용하는데 있어 많은 시간을 요구한다. 이를 해결하기 위해, 우리는 딥러닝 모델의 가중치를 양자화 하여 추론시간을 줄이고자 한다. 초해상도 모델은 feature extraction, non-linear mapping, reconstruction 세 부분으로 나누어져 있으며, 레이어 사이에 많은 skip-connection 이 존재하는 특징이 있다. 따라서 양자화 시 최종 성능 하락에 미치는 영향력이 레이어 별로 다르며, 이를 감안하여 강화학습으로 레이어 별 최적 bit 를 찾아 성능 하락을 최소화한다. 본 논문에서는 Skip-connection 이 많이 존재하는 MSRN 을 사용하였으며, 결과에서 feature extraction, reconstruction 부분과 블록 내 특정 위치의 레이어가 항상 높은 bit 를 가짐을 알 수 있다. 기존에 영상 분류에 한정되어 사용되었던 혼합 bit 양자화를 사용하여 초해상도 딥러닝 기법의 모델 사이즈를 줄인 최초의 논문이며, 제안 방법은 모바일 등 제한된 환경에 적용 가능할 것으로 생각된다.

• Journal of the Society of Naval Architects of Korea
• /
• v.56 no.2
• /
• pp.152-160
• /
• 2019

In this study, we propose a piping material supply management method using CAD system. The piping materials installed in super large offshore plants have very complicated connection conditions. Therefore, it is very difficult to determine the order of receipt of a large number of installation materials. Therefore, we have developed a system that can automatically check the preparation rate of installation materials prior to the installation process. We have developed an algorithm to obtain connection information among installation items from PDMS system. We have developed an algorithm that can determine the order of installation materials to be installed using the connection information. The order of the installation material is determined by taking into account the constraint conditions for the complete installation of the piping material. We confirm the effectiveness of the developed algorithms in the operating system. This system is also used to manage installation schedules and plan the installation manpower.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.18 no.1
• /
• pp.71-77
• /
• 2018

The optical time domain reflectometer (OTDR) is the most widely used method to monitor problems with currently installed optical fibers. The OTDR is an instrument designed to test the FTTx network and evaluates the physical properties of the fiber, such as transmission loss and connection loss. It is important to improve the spatial resolution in order to accurately grasp the optical path problems by using the OTDR. When the pulse width is less than twice the distance between the two reflectors, the signals reflected from the two reflectors are reflected without overlap, so that the reflected signal can be distinguished. However, when the pulse width is larger than twice the distance between the two reflectors, so that the reflected signal can not be distinguished. In order to overcome these limitations, this paper proposed a method of improving spatial resolution by applying a super resolution algorithm. As a result of the simulation, the resolution is improved when the super resolution algorithm is applied, and the event interval can be analyzed more precisely.

• Steel and Composite Structures
• /
• v.36 no.2
• /
• pp.213-228
• /
• 2020

Shape Memory Alloys (SMAs) are new materials used in various fields of science and engineering, one of which is civil engineering. Owing to their distinguished capabilities such as super elasticity, energy dissipation, and tolerating cyclic deformations, these materials have been of interest to engineers. On the other hand, the connections of a steel structure are of paramount importance because of their vulnerabilities during an earthquake. Therefore, it is indispensable to find approaches to augment the efficiency and safety of the connection. This research investigates the behavior of steel connections with extended end plates equipped hybridly with 8 rows of high strength bolts as well as Nitinol superelastic SMA bolts. The connections are studied using component method in dual form. In this method, the components affecting the connections behavior, such as beam flange, beam web, column web, extended end plate, and bolts are considered as parallel and series springs according to the Euro-Code3. Then, the nonlinear force- displacement response of the connection is presented in the form of moment-rotation curve. The results obtained from this survey demonstrate that the connection has ductility, in addition to its high strength, due to high ductility of SMA bolts.

• Structural Engineering and Mechanics
• /
• v.72 no.1
• /
• pp.99-111
• /
• 2019

To investigate the seismic performance of long-span partially earth-anchored cable-stayed bridge, a super long-span partially earth-anchored cable-stayed bridge scheme with main span of 1400m is taken as example, structural response of the bridge under E1 seismic action is investigated numerically by the multimode seismic response spectrum and time-history analysis, seismic behavior and also the effect of structural geometric nonlinearity on the seismic responses of super long-span partially earth-anchored cable-stayed bridges are revealed. The seismic responses are also compared to those of a fully self-anchored cable-stayed bridge with the same main span. The effects of structural parameters including the earth-anchored girder length, the girder width, the girder depth, the tower height to span ratio, the inclination of earth-anchored cables, the installation of auxiliary piers in the side spans and the connection between tower and girder on the seismic responses of partially ground-anchored cable-stayed bridges are investigated, and their reasonable values are also discussed in combination with static performance and structural stability. The results show that the horizontal seismic excitation produces significant seismic responses of the girder and tower, the seismic responses of the towers are greater than those of the girder, and thus the tower becomes the key structural member of seismic design, and more attentions should be paid to seismic design of these sections including the tower bottom, the tower and girder at the junction of tower and girder, the girder at the auxiliary piers in side spans; structural geometric nonlinearity has significant influence on the seismic responses of the bridge, and thus the nonlinear time history analysis is proposed to predict the seismic responses of super long-span partially earth-anchored cable-stayed bridges; as compared to the fully self-anchored cable-stayed bridge with the same main span, several stay cables in the side spans are changed to be earth-anchored, structural stiffness and natural frequency are both increased, the seismic responses of the towers and the longitudinal displacement of the girder are significantly reduced, structural seismic performance is improved, and therefore the partially earth-anchored cable-stayed bridge provides an ideal structural solution for super long-span cable-stayed bridges with kilometer-scale main span; under the case that the ratio of earth-anchored girder length to span is about 0.3, the wider and higher girder is employed, the tower height-to-span ratio is about 0.2, the larger inclination is set for the earth-anchored cables, 1 to 2 auxiliary piers are installed in each of the side spans and the fully floating system is employed, better overall structural performance is achieved for long-span partially earth-anchored cable-stayed bridges.

• The Journal of the Convergence on Culture Technology
• /
• v.6 no.4
• /
• pp.663-667
• /
• 2020

By 2050, the technologies of the fourth industrial revolution currently being discussed will become highly mature and face a turning point in human civilization. Among them, the most innovative technologies will be artificial intelligence, robots, virtual reality and hyper-connected networks. These technologies will transform not only human life but also the appearance of the battlefield. Therefore, in 2050, the Army should concentrate all its capabilities on developing new concept weapons systems based on superintelligence and hyperconnectivity that converge and integrate these technologies.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.5
• /
• pp.185-191
• /
• 2022

In this study, we propose a novel super-resolution neural network based on U-Net, residual neural network, and sub-pixel convolution. To prevent the loss of detailed information due to the max pooling of U-Net, we propose down-sampling and connection using sub-pixel convolution. This uses all pixels in the filter, unlike the max pooling that creates a new feature map with only the max value in the filter. As a 2×2 size filter passes, it creates a feature map consisting only of pixels in the upper left, upper right, lower left, and lower right. This makes it half the size and quadruple the number of feature maps. And we propose two methods to reduce the computation. The first uses sub-pixel convolution, which has no computation, and has better performance, instead of up-convolution. The second uses a layer that adds two feature maps instead of the connection layer of the U-Net. Experiments with a banchmark dataset show better PSNR values on all scale and benchmark datasets except for set5 data on scale 2, and well represent local area patterns.

• Journal of Broadcast Engineering
• /
• v.24 no.1
• /
• pp.132-141
• /
• 2019

The necessity of transmitting video data over a narrow-bandwidth exists steadily despite that video service over broadband is common. In this paper, we propose a scalable video coding framework for low-resolution video transmission over a very narrow-bandwidth network by super-resolution of decoded frames of a base layer using a convolutional neural network based super resolution technique to improve the coding efficiency by using it as a prediction for the enhancement layer. In contrast to the conventional scalable high efficiency video coding (SHVC) standard, in which upscaling is performed with a fixed filter, we propose a scalable video coding framework that replaces the existing fixed up-scaling filter by using the trained convolutional neural network for super-resolution. For this, we proposed a neural network structure with skip connection and residual learning technique and trained it according to the application scenario of the video coding framework. For the application scenario where a video whose resolution is $352{\times}288$ and frame rate is 8fps is encoded at 110kbps, the quality of the proposed scalable video coding framework is higher than that of the SHVC framework.

• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.12
• /
• pp.403-410
• /
• 2020

In addition to the research on noise removal and super-resolution using the data restoration (Output result) function of Autoencoder, research on the performance improvement of clustering using the dimension reduction function of autoencoder are actively being conducted. The clustering function and data restoration function using Autoencoder have common points that both improve performance through the same learning. Based on these characteristics, this study conducted an experiment to see if the autoencoder model designed to have excellent data recovery performance is superior in clustering performance. Skip connection technique was used to design autoencoder with excellent data recovery performance. The output result performance and clustering performance of both autoencoder model with Skip connection and model without Skip connection were shown as graph and visual extract. The output result performance was increased, but the clustering performance was decreased. This result indicates that the neural network models such as autoencoders are not sure that each layer has learned the characteristics of the data well if the output result is good. Lastly, the performance degradation of clustering was compensated by using both latent code and skip connection. This study is a prior study to solve the Hanja Unicode problem by clustering.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.5 no.11
• /
• pp.2143-2159
• /
• 2011

The hybrid architecture of content delivery networks (CDN) and peer-to-peer overlay networks (P2P) is a promising technology enables effective real-time streaming services. It complements the advantages of quality control and reliability in a CDN, and the scalability of a P2P system. With real-time streaming services, however, high connection setup and media delivery latency are becoming the critical issues in deploying the CDN-P2P system. These issues result from biased peer selection without location awareness or content awareness, and can lead to significant service disruption. To reduce service disruption latency, we propose a group-based CDN-P2P hybrid architecture (iCDN-P2P) with a location/content-aware selection of peers. Specifically, a SuperPeer network makes a location-aware peer selection by employing a content addressable network (CAN) to distribute channel information. It also manages peers with content awareness, forming a group of peers with the same channel as the sub-overlay. Through a performance evaluation, we show that the proposed architecture outperforms the original CDN-P2P hybrid architecture in terms of connection setup delay and media delivery time.