• ETRI Journal
• /
• v.41 no.3
• /
• pp.371-382
• /
• 2019

Machine reading comprehension is the task of understanding a given context and finding the correct response in that context. A simple recurrent unit (SRU) is a model that solves the vanishing gradient problem in a recurrent neural network (RNN) using a neural gate, such as a gated recurrent unit (GRU) and long short-term memory (LSTM); moreover, it removes the previous hidden state from the input gate to improve the speed compared to GRU and LSTM. A self-matching network, used in R-Net, can have a similar effect to coreference resolution because the self-matching network can obtain context information of a similar meaning by calculating the attention weight for its own RNN sequence. In this paper, we construct a dataset for Korean machine reading comprehension and propose an $S^2-Net$ model that adds a self-matching layer to an encoder RNN using multilayer SRU. The experimental results show that the proposed $S^2-Net$ model has performance of single 68.82% EM and 81.25% F1, and ensemble 70.81% EM, 82.48% F1 in the Korean machine reading comprehension test dataset, and has single 71.30% EM and 80.37% F1 and ensemble 73.29% EM and 81.54% F1 performance in the SQuAD dev dataset.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.18 no.2
• /
• pp.73-79
• /
• 2018

In this paper, TM(transverse magnetic) scattering problems by a conductive strip grating between a double dielectric layer are analyzed by applying the PMM(point matching method) known as a numerical method of electromagnetic fileld. The boundary conditions are applied to obtain the unknown field coefficients, the scattered electromagnetic fields are expanded in a series of Floquet mode functions, and the conductive boundary condition is applied to analysis of the conductive strip. The most normalized reflected powers of the sharp variations in minimum values are scattered in direction of the other angles except incident angle. Generally, in the case of numerical analysis except for reflection and transmission power in free space, as the dielectric constants of the double dielectric layer increases, the reflected power increases and the transmitted power decreases relatively, respectively. The numerical results for the presented structure of this paper having a grounded double dielectric layer are shown in good agreement compared to those of the existing papers.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.1
• /
• pp.49-54
• /
• 2021

In this paper, n this paper, E-polarized electromagnetic scattering problems by a resistive strip grating between a double dielectric layer are analyzed by applying the PMM(Point Matching Method) known as a numerical method of electromagnetic fileld. The boundary conditions are applied to obtain the unknown field coefficients, and the resistive boundary condition is applied to analysis of the resistive strip. The numerical results for the normalized reflected and transmitted power are analyzed by according as the relative permittivity and thickness of the double dielectric layers, and the resistivity of resistive strip. Overall, when the resistivity of the resistive strip decreased or the relative permittivity of the dielectric layer increased, the reflected power increased, and as the reflected power increased, the transmitted power decreased relatively. Especially, as the relative permittivity of double dielectric layer increases, the minimum value of the variation curve of the reflected power shifted in the direction that the grating period decreased. The numerical results for the presented structure of this paper are shown in good agreement compared to those of the existing papers.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.1
• /
• pp.29-34
• /
• 2022

In this paper, thr H-polarized scattering problems by a resistive strip grating in a grounded double dielectric layer are analyzed by applying the PMM(point matching method) known as a numerical method of electromagnetic fileld. The boundary conditions are applied to obtain the unknown field coefficients, the scattered electromagnetic fields are expanded in a series of Floquet mode functions, and the resistive boundary condition is applied to analysis of the resistive strip. The %error of the convergence of the reflected power according to the relative permittivity of the dielectric layer and the size of the number of rows in the square matrix was compared, as the size of the number of rows in the square matrix increased, the accuracy of the reflected power increased. As the resistivity of the resistive strip decreased, the thickness of the dielectric layers decreased, and the relative permittivity of the dielectric layers increased, the reflected power increased. The numerical results for the presented structure of this paper having a grounded double dielectric layer are shown in good agreement compared to those of the existing papers.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.31 no.8
• /
• pp.680-687
• /
• 2007

The paper presents a numerical simulation of flow of a turbulent boundary layer, representing a typical wind environment and matching a series of wind tunnel observations. The simulations are carried out at a Reynolds number of 20,000, based on the velocity U at a pseudo-height h, and large enough that the flow be effectively Reynolds number independent. Some wall models are proposed for the LES(Large Eddy Simulation) of the turbulent boundary layer over a rough surface. The Jenson number, $J=h/z_0$, based on the roughness length $z_0$, is 600 to match the wind tunnel data. The computational mesh is uniform with a spacing of h/32, as this aids rapid convergence of the multigrid solver, and the governing equations are discretised using second order finite differences within a parallel multiblock environment. The results presented include the comparison between wind tunnel measurements and LES computations of the turbulent boundary layer over rough surface.

• Composites Research
• /
• v.14 no.6
• /
• pp.38-46
• /
• 2001

In this study, An optimization code that can design microwave absorbing composite laminates is developed, and 3-layered microwave absorbing composite laminates are developed by optimizing the thickness of each layer. The layers are 3 different composite laminates. Many variables including lay-up angles of electromagnetically orthotropic composite layer can be considered in this code. The developed laminate is composed of an impedance matching layer of glass/epoxy fabric laminate, a glass/epoxy fabric laminate layer containing aluminum filler and carbon/epoxy fabric laminate layer. Permittivities of the materials are obtained using a network analyzer and a coaxial air line.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.7 no.3 s.18
• /
• pp.165-173
• /
• 2004

Matching inviscid and boundary layer methods are developed for analysis of hypersonic flow with thick boundary layer. The new equations match all the boundary layer properties with a variation in the inviscid solution near the edge, except for the normal velocity. Computational comparison are peformed for incompressible and compressible flows over a flat plate. Results from the present method are compared with Wavier-Stokes solutions. The present results are in good agreement with Wavier-Stokes solutions. They show that the new technique can provide improved predictions of heating rates and skin friction predictions for preliminary design of vehicles where shear layers and entropy layer swallowing are importantfor for preliminary design.

• 한국신재생에너지학회:학술대회논문집
• /
• 2011.11a
• /
• pp.64.1-64.1
• /
• 2011

다중접합 태양전지는 흡수대역이 다른 juntion으로 구성되어, 각각의 태양전지 간의 전류정합(current matching)이 효율 향상에 중요하다. 본 실험에서는 Top cell에 i-a-Si:H(Thinckness:100nm), Middle cell에는 i-a-SiGe:H(Thickness:800nm)을 적용하였고, bottom cell에는 i-${\mu}c$-Si:H(Thickness:1800nm), 수광부의 p-layer에 에 SiOx을 이용하여 triple juntion amorphous silicon solar cell(삼중접합태양전지)을 구현하였다. 이를 최적화 시키기 위해 ASA simulation을 이용하여 각 Cell의 intrinsic layer의 밴드갭을 가변하였다. 가변 결과 i-a-Si:H : 1.85 eV, i-a-SiGe:H: 1.6 eV, i-${\mu}c$-Si:H: 1.4 eV에서 태양전지 효율 14.5 %을 기록 하였다. 본 연구를 통해 Triple juntion cell에서의 intrinsic layer의 밴드갭 최적화를 구현해 볼 수 있었다.

• 제어로봇시스템학회:학술대회논문집
• /
• 1989.10a
• /
• pp.1135-1139
• /
• 1989

Three layered neural network was applied for the pattern recognition problem of human spermatozoa in clinical test. The goodness of recognition rate was studied in relation to the number of hidden layer cells and of output layer cells. The proposed method provided better results than conventional template matching technique. Parallel processing of the back propagation learning algorithm was also studied using transputers and its performance was evaluated.

• Journal of the Korean Graphic Arts Communication Society
• /
• v.9 no.1
• /
• pp.1-11
• /
• 1991

In this reseach, double-layered photoconductor consist of the carrier generation layer(CGL) of $\varepsilon$ type copper phthalocyanine thin film by an aqueous coating method and the carrier transport layer(CGL) of polyvinyl carbazol(PVK) by spin coating. We inverstigated effect of the surfactant solution and cathod electrolysis to the crystal type of $\varepsilon$-CuPc in CGL with TEM, SEM and X- ray diffraction spectroscopy and studied the mechanism of an aqueous coating for the preparation of CGL. The effect of the washing of CGL about the electrophotographic characteristics of the $\varepsilon$-CuPc/PVK doublelayered photoconductors is studied also.