• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2359-2376
• /
• 2022

With the exponential growth of medical image big data represented by high-resolution CT images(CTI), the high-resolution CTI data is of great importance for clinical research and diagnosis. The paper takes lung CTI as an example to study. Retrieving answer CTIs similar to the input one from the large-scale lung CTI database can effectively assist physicians to diagnose. Compared with the conventional content-based image retrieval(CBIR) methods, the CBIR for lung CTIs demands higher retrieval accuracy in both the contour shape and the internal details of the organ. In traditional supervised deep learning networks, the learning of the network relies on the labeling of CTIs which is a very time-consuming task. To address this issue, the paper proposes a Weakly Supervised Similarity Evaluation Network (WSSENet) for efficiently support similarity analysis of lung CTIs. We conducted extensive experiments to verify the effectiveness of the WSSENet based on which the CBIR is performed.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.128-130
• /
• 2021

Smart cities collect data from thousands of IoT-based sensor devices for intelligent application-based services. Centralized cloud servers support application tasks with higher computation resources but introduce network latency. Fog layer-based data centers bring data processing at the edge, but fewer available computation resources and poor task allocation strategy prevent real-time data analysis. In this paper, tasks generated from devices are distributed as high resource and low resource intensity tasks. The novelty of this research lies in deploying a virtual node assigned to each cluster of IoT sensor machines serving a joint application. The node allocates tasks based on the task intensity to either cloud-computing or fog computing resources. The proposed Task Allocation Strategy provides seamless allocation of jobs based on process requirements.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2022.11a
• /
• pp.383-384
• /
• 2022

Gas Combustion Unit (GCU) onboard liquefied natural gas carriers handles boil-off to stabilize tank pressure. There are many factors for LNG cargo operators to take into consideration to determine whether to use GCU or not. Gas consumption of main engine and re-liquefied gas through the Partial Re-Liquefaction System (PRS) are good examples of these factors. Human gas operators have decided the operation so far. In this paper, some deep learning neural network models were developed to provide human gas operators with a decision support system. The models consider various factors specially into GCU operation. A deep learning model with Sigmoid activation functions in input layer and hidden layers made the best performance among eight different deep learning models.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.3
• /
• pp.279-286
• /
• 2015

The rapid increase of the current marine accidents is mainly due to the human execution errors. In an effort to address this, various kinds of researches such as construction of the digital vessels and vessel information monitoring systems have been conducted. But for safe navigation of vessels, it lack on systems study which can efficiently store, utilize and manage the mass data accepted by the vessel. In this paper, we propose a VWS(Virtual World System) that is based on the architecture of intelligent systems RVC(Reactive Layer-Virtual World-Considerative Layer) model of intelligent autonomous navigation system. VWS is responsible to store all the necessary information for safe navigation of the vessel and the information services to the sub-system of intelligent autonomous navigation system. VWS uses topology database to express the specific problem area, and utilizes a scheduling to reflect the characteristics of the real-time processing environment. Also, Virtual World defines API for the system to reflect the characteristics of the distributed processing environment. As a case study, the VWS is applied to a intelligent ship autonomous navigation system, and simulation is done to prove the effectiveness of the proposed system.

• Journal of the Korean Society for information Management
• /
• v.21 no.4 s.54
• /
• pp.109-131
• /
• 2004

As there are diverse delivery media and forms of information resources, their management schemes are diverse as well. In library community, cataloguing. rules for describing information resources such as AACR and KCR have been developed. The efforts to automate management of information resources based on these rules resulted in the development of MARC. However, MARC records are restricted in describing the information resources and MARC has various and distinct characteristics of the structural rigidity, which does not support the representation of extended semantic structures that exist among bibliographic entities. Therefore, since the data model for MARC is single-layer data model, it is not appropriate for describing information resources represented by multi-layer data model which can be used to set up the relationships among various objects in digital libraries. In this paper, we propose an a metadata model for digital libraries based on the IFLA FRBR basic model which supports multi-layer data model and a representation scheme based on XML Schema to manage the metadata about old books, old documents, resource related to music, conferences and seminars.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.12B
• /
• pp.1749-1757
• /
• 2010

This paper considers the cost minimization problem to satisfy QoS (Quality of Service) requirements for a given network, in particular when communication resources to each link can be additionally assigned. For the purpose of quantifying QoS requirements such as data transfer delay and packet loss, we introduce the cost function considering both the link utilization factor and the additionally assigned resource. To minimize this cost function, we firstly formulate a Basic Capacity Planning (BCP) problem, a special case of Network Utility Maximization (NUM). We show that the solution of this BCP problem cannot be optimal via a counter example. In this paper, we suggest the cross-layer design of both additionally assigned resource and routing path, whose initial values are set to the result of BCP problem. This cross-layer design is based on a heuristic approach which presents an effective way to plan how much communication resources should be added to support the QoS requirements in future. By simulation study, we investigate the convergence of the cost function in a more general network topology as well as in a given simple topology.

• Journal of the Microelectronics and Packaging Society
• /
• v.28 no.3
• /
• pp.33-38
• /
• 2021

Recently, rollable and foldable displays are attracting great attention in the flexible display market due to their excellent form factor. To predict and prevent the mechanical failure of the display panels, it is essential to accurately understand the mechanical properties of brittle SiN_x thin films, which have been used as an insulating film in flexible displays. In this study, tensile properties of the ~130 nm- and ~320 nm-thick SiN_x thin films were successfully measured by coating a ~190 nm-thick organic nano-support-layer (PMMA, PS, P3HT) on the fragile SiN_x thin films and stretching the films as a bilayer state. Young's modulus values of the ~130 nm and ~320 nm SiN_x thin films fabricated through the controlled chamber pressure and deposition power (A: 1250 mTorr, 450 W/B: 1000 mTorr, 600 W/C: 750 mTorr, 700 W) were calculated as A: 76.6±3.5, B: 85.8±4.6, C: 117.4±6.5 GPa and A: 100.1±12.9, B: 117.9±9.7, C: 159.6 GPa, respectively. As a result, Young's modulus of ~320 nm SiN_x thin films fabricated through the same deposition condition increased compared to the ~130 nm SiN_x thin films. The tensile testing method using the organic nano-support-layer was effective in the precise measurement of the mechanical properties of the brittle thin films. The method developed in this study can contribute to the robust design of the rollable and foldable displays by enabling quantitative measurement of mechanical properties of fragile thin films for flexible displays.

• Journal of the Korean Crystal Growth and Crystal Technology
• /
• v.23 no.6
• /
• pp.265-271
• /
• 2013

Czochralski (CZ) growth process is one of the most important techniques for growing high quality sapphire single crystal for LED application. In this study, the inductively-heated CZ growth processes for the sapphire crystal of 300 mm length have been analyzed numerically using finite element method. The hot zone structures were modified with the crucible geometry change and the additional insulation layer installed above the crucible. The results show that the solid-liquid interface height decreased from about 80 mm at initial stage to 40 mm after mid-stage due to achieve the growth speed balance. Also the optimal input power of the modified system was similar with the original one due to the compensation effects of the crucible geometry and additional insulation. The crystal temperature grown by the modified CZ grower was increased about 10 K than the original one. Therefore the sapphire crystal of 300 mm height was grown successfully.

• Journal of KIISE:Information Networking
• /
• v.32 no.3
• /
• pp.391-403
• /
• 2005

Several nv6 mobility support protocols for mobile Internet services are proposed in IETP : Mobile Ipv6, Hierarchical Mobile Ipv6, and Fast Handovers over Mobile Ipv6. Recently, IEEE 802.11 network has also been widely deployed in public areas for mobile Internet services. In the near future, IPv6 mobility support over IEEE 802.11 network is expected to be a key function to actualize the All If-based mobile various services. For appropriate application of these protocols, the IPv6 mobility support protocols should be analyzed according to their characteristics in terms of signaling, handover latency, lost packets, and required buffer sire as well as the impact of lower layer such as IEEE 802.11 network. In this paper, we analyze the performance of the protocols over IEEE 802.11 network. We define a packet-level traffic model, a network system model, and a mobility model. From these models, we construct a framework for the performance analysis. We also make cost functions to formalize each protocol's performance. Lastly, we analyze the effect of varying parameters used to show diverse numerical results, and compare with each other. From the analysis results, it is concluded that each Protocol has contrary or contrastive advantages with other Protocols, so there is no protocol that holds a dominant position.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.5 s.305
• /
• pp.123-136
• /
• 2005

We propose a recognition system for superimposed patterns based on selective attention model and SVM which produces better performance than artificial neural network. The proposed selective attention model includes attention layer prior to SVM which affects SVM's input parameters. It also behaves as selective filter. The philosophy behind selective attention model is to find the stopping criteria to stop training and also defines the confidence measure of the selective attention's outcome. Support vector represents the other surrounding sample vectors. The support vector closest to the initial input vector in consideration is chosen. Minimal euclidean distance between the modified input vector based on selective attention and the chosen support vector defines the stopping criteria. It is difficult to define the confidence measure of selective attention if we apply common selective attention model, A new way of doffing the confidence measure can be set under the constraint that each modified input pixel does not cross over the boundary of original input pixel, thus the range of applicable information get increased. This method uses the following information; the Euclidean distance between an input pattern and modified pattern, the output of SVM, the support vector output of hidden neuron that is the closest to the initial input pattern. For the recognition experiment, 45 different combinations of USPS digit data are used. Better recognition performance is seen when selective attention is applied along with SVM than SVM only. Also, the proposed selective attention shows better performance than common selective attention.