• ETRI Journal
• /
• v.41 no.2
• /
• pp.235-241
• /
• 2019

In this paper, we propose a rank-weighted reconstruction feature to improve the robustness of a feed-forward deep neural network (FFDNN)-based acoustic model. In the FFDNN-based acoustic model, an input feature is constructed by vectorizing a submatrix that is created by slicing the feature vectors of frames within a context window. In this type of feature construction, the appropriate context window size is important because it determines the amount of trivial or discriminative information, such as redundancy, or temporal context of the input features. However, we ascertained whether a single parameter is sufficiently able to control the quantity of information. Therefore, we investigated the input feature construction from the perspectives of rank and nullity, and proposed a rank-weighted reconstruction feature herein, that allows for the retention of speech information components and the reduction in trivial components. The proposed method was evaluated in the TIMIT phone recognition and Wall Street Journal (WSJ) domains. The proposed method reduced the phone error rate of the TIMIT domain from 18.4% to 18.0%, and the word error rate of the WSJ domain from 4.70% to 4.43%.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.6
• /
• pp.1035-1043
• /
• 2022

With the spread of the Internet of Things (IoT), cloud computing, and big data, the need for high-speed encryption for applications is emerging. GPU optimization can be used to validate cryptographic analysis results or reduced versions theoretically obtained by the GPU in a reasonable time. In this paper, PIPO lightweight encryption implemented in various environments was implemented on GPU. Optimally implemented considering the brute force attack on PIPO. In particular, the optimization implementation applying the bit slicing technique and the GPU elements were used as much as possible. As a result, the implementation of the proposed method showed a throughput of about 19.5 billion per second in the RTX 3060 environment, achieving a throughput of about 122 times higher than that of the previous study.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.611-614
• /
• 2023

갈비뼈 병변 진단 과정은 방사선 전문의가 CT 스캐너를 통해 생성된 2 차원 CT 이미지들을 해석하며 진행된다. 병변의 위치를 파악하고 정확한 진단을 내리기 위해 수백장의 2차원 CT 이미지들이 세밀하게 검토되며 갈비뼈를 분류한다. 본 연구는 이런 노동 집약적 작업의 문제점을 개선시키기 위해 Biaxial Rib Segmentation(BARS)을 제안한다. BARS 는 흉부 CT 볼륨의 관상면과 수평면으로 구성된 2 차원 이미지들을 U-Net 모델에 학습한다. 모델이 산출한 세그멘테이션 마스크들의 조합은 서로 다른 평면의 공간 정보를 보완하며 3 차원 갈비뼈 볼륨을 재건한다. BARS 의 성능은 DSC, Recall, Precision 지표를 사용해 평가하며, DSC 90.29%, Recall 89.74%, Precision 90.72%를 보인다. 향후에는 이를 기반으로 순차적 갈비뼈 레이블링 연구를 진행할 계획이다.

• Journal of Information Processing Systems
• /
• v.20 no.1
• /
• pp.38-52
• /
• 2024

The 5G is the 5th generation mobile network that provides enhanced mobile broadband, ultra-reliable & low latency communications, and massive machine-type communications. New services can be provided through multi-access edge computing, network function virtualization, and network slicing, which are key technologies in 5G mobile communication. However, these new technologies provide new attack paths and threats. In this paper, we analyzed the overall threats of 5G mobile communication through a literature review. First, defines 5G mobile communication, analyzes its features and technology architecture, and summarizes possible security issues. Addition, it presents security threats from the perspective of user devices, radio access network, multi-access edge computing, and core networks that constitute 5G mobile communication. After that, security requirements for threat factors were derived through literature analysis. The purpose of this study is to conduct a fundamental analysis to examine and assess the overall threat factors associated with 5G mobile communication. Through this, it will be possible to protect the information and assets of individuals and organizations that use 5G mobile communication technology, respond to various threat situations, and increase the overall level of 5G security.

• Journal of KIISE:Computing Practices and Letters
• /
• v.6 no.5
• /
• pp.528-535
• /
• 2000

This paper aims at providing guidelines enabling productive software construction by offering reusable modules which is used not only for effective maintenance for each step, but also for a re-engineering process after analyzing developed source code. There are four processing steps. The first is to analyze source code. The second is module slicing and clustering using McCabe and BP/Win Tools, The third is to transform the outputs extracted from the business model to reusable modules. The final step is to design repository and to construct a system. In this paper, we applied the fourth step to our case study, which was specified from the first step to the fourth. The specified fourth step contains various things for constructing repository. And the fourth step reanalyzes informal and unstructured information by using reverse engineering tools, in order to provide effective guidelines for productive software maintenance and re-engineering.

• Journal of Yeungnam Medical Science
• /
• v.37 no.4
• /
• pp.345-348
• /
• 2020

Papillary fibroelastomas are the second most common primary cardiac tumor in adults. Over 80% of fibroelastomas occur on the cardiac valves, usually on the left side of the heart, while the remaining lesions are typically scattered throughout the atria and ventricles. Although the optimal timing for surgery is controversial and depends on tumor size and location, prompt surgical resection is warranted in patients at high risk of embolism. A tumor on the cardiac valve can be removed using the slicing excision technique without leaflet injury. Here we present two cases of papillary fibroelastomas occurring on the ventricular surface of the aortic valve and in the right ventricle.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2005.06a
• /
• pp.1214-1218
• /
• 2005

As demands for precision parts are increased, existing methods to fabricate them such as MEMS, LIGA technology have the technical limitations like high precision, high functionality and ultra miniaturization. A micro-stereolithography technology based on $DMD^{TM}$(Digital Micromirror Device) can meet these demands. In this technology, STL file is the standard format as the same of conventional rapid prototyping system, and 3D part is fabricated by stacking layers that are sliced as 2D section from STL file. Whereas in conventional method, the resin surface is cured as scanning laser beam spot according to the section shape, but in this research, we use integral process which enables to cure the resin surface at one time. In this paper, we deal with the dynamic pattern generation and $DMD^{TM}$ operation to fabricate micro structures. Firstly, we address effective slicing method of STL file, conversion to bitmap, and dynamic pattern generation. Secondly, we suggest $DMD^{TM}$ operation and optimal support manufacturing for $DMD^{TM}$ mounting. Thirdly, we examine the problems on continuous stacking layers, and their improvements in software aspects.

• Korean Journal of Materials Research
• /
• v.22 no.9
• /
• pp.450-453
• /
• 2012

In crystalline solar cells, the substrate itself constitutes a large portion of the fabrication cost as it is derived from semiconductor ingots grown in costly high temperature processes. Thinner wafer substrates allow some cost saving as more wafers can be sliced from a given ingot, although technological limitations in slicing or sawing of wafers off an ingot, as well as the physical strength of the sliced wafers, put a lower limit on the substrate thickness. Complementary to these economical and techno-physical points of view, a device operation point of view of the substrate thickness would be useful. With this in mind, BC-BJ Si and GaAs solar cells are compared one to one by means of the Medici device simulation, with a particular emphasis on the substrate thickness. Under ideal conditions of 0.6 ${\mu}m$ photons entering the 10 ${\mu}m$-wide BC-BJ solar cells at the normal incident angle (${\theta}=90^{\circ}$), GaAs is about 2.3 times more efficient than Si in terms of peak cell power output: 42.3 $mW{\cdot}cm^{-2}$ vs. 18.2 $mW{\cdot}cm^{-2}$. This strong performance of GaAs, though only under ideal conditions, gives a strong indication that this material could stand competitively against Si, despite its known high material and process costs. Within the limitation of the minority carrier recombination lifetime value of $5{\times}10^{-5}$ sec used in the device simulation, the solar cell power is known to be only weakly dependent on the substrate thickness, particularly under about 100 ${\mu}m$, for both Si and GaAs. Though the optimum substrate thickness is about 100 ${\mu}m$ or less, the reduction in the power output is less than 10% from the peak values even when the substrate thickness is increased to 190 ${\mu}m$. Thus, for crystalline Si and GaAs with a relatively long recombination lifetime, extra efforts to be spent on thinning the substrate should be weighed against the expected actual gain in the solar cell output power.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.44 no.11
• /
• pp.941-948
• /
• 2016

Textile composite materials have been widely applied in aerospace structures due to their various advantages such as high specific stiffnesses and strengths, better out-of-plane performances, impact and delamination resistances, and net shape fabrications. In this paper, a modified geometric model of repeating unit cell (RUC) is suggested based on the Naik's model for 2D plain weave textile composites. The RUC geometry is defined by various parameters. The proposed model considers another parameter which is a gap length between adjacent yarns. The effective stiffnesses are predicted by using the yarn slicing technique and stress averaging technique based on iso-strain assumption. And the stiffnesses of RUC are evaluated by adjusting the gap ratio and verified by comparing with Naik's model and experimental data for 2D plain weave composite specimens.

• Journal of KIISE:Software and Applications
• /
• v.31 no.5
• /
• pp.643-650
• /
• 2004

Control flow graph is used for Performing many Program-analysis techniques, such as data-flow and control-dependence analysis, and software-engineering techniques, such as program slicing and testings. For these analyses to be safe and useful, the CFG should incorporate the exception flows that are induced by exceptions. In previous research to construct control flow graph, normal flows and exception flows are computed at the same time, since these two flows are known to be mutually dependent. By investigating realistic Java programs, we found that the cases when these two flows are mutually dependent rarely happen. So, we can decouple exception flow analysis from normal flow analysis. In this paper we propose an analysis that estimates exception flows. We also propose exception flow graph to represent exception flows. And we show that the control flow graph that accounts for exception flows can be constructed by merging exception flow graph onto normal control flow graph.