• Journal of Information Processing Systems
• /
• v.19 no.6
• /
• pp.803-816
• /
• 2023

Explainable artificial intelligence is a method that explains how a complex model (e.g., a deep neural network) yields its output from a given input. Recently, graph-type data have been widely used in various fields, and diverse graph neural networks (GNNs) have been developed for graph-type data. However, methods to explain the behavior of GNNs have not been studied much, and only a limited understanding of GNNs is currently available. Therefore, in this paper, we propose an explanation method for node classification using graph convolutional networks (GCNs), which is a representative type of GNN. The proposed method finds out which features of each node have the greatest influence on the classification of that node using GCN. The proposed method identifies influential features by backtracking the layers of the GCN from the output layer to the input layer using the gradients. The experimental results on both synthetic and real datasets demonstrate that the proposed explanation method accurately identifies the features of each node that have the greatest influence on its classification.

• Structural Engineering and Mechanics
• /
• v.68 no.1
• /
• pp.103-119
• /
• 2018

In the present research, an attempt is made to obtain a semi analytical solution for both nonlinear natural frequency and forced vibration of embedded functionally graded double layered nanoplates with all edges simply supported based on nonlocal strain gradient elasticity theory. The interaction of van der Waals forces between adjacent layers is included. For modeling surrounding elastic medium, the nonlinear Winkler-Pasternak foundation model is employed. The governing partial differential equations have been derived based on the Mindlin plate theory utilizing the von Karman strain-displacement relations. Subsequently, using the Galerkin method, the governing equations sets are reduced to nonlinear ordinary differential equations. The semi analytical solution of the nonlinear natural frequencies using the homotopy analysis method and the exact solution of the nonlinear forced vibration through the Harmonic Balance method are then established. The results show that the length scale parameters give nonlinearity of the hardening type in frequency response curve and the increase in material length scale parameter causes to increase in maximum response amplitude, whereas the increase in nonlocal parameter causes to decrease in maximum response amplitude. Increasing the material length scale parameter increases the width of unstable region in the frequency response curve.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.2
• /
• pp.130-136
• /
• 2012

In this paper, we propose a near-field range estimation method for a uniform linear array that can calibrate bearing estimation error which give a bad influence on a range estimation process. When a range is fixed, the bearing error is calibrated to maximize the beamformer output by the proposed algorithm based on the gradient method. Simulation results show that the proposed algorithm can compensate the bearing error which is less than the mainlobe beamwidth so that reduce the range estimation error as similar as the case of no bearing error.

• Journal of the Korean earth science society
• /
• v.35 no.7
• /
• pp.529-536
• /
• 2014

In this paper, the automatic determination algorithm of strike and dip of a line source using gravity gradient on a single profile is proposed. In general, the gravity gradient tensor due to a line source has only two independent components because of its 2-Dimensional (2-D) characteristics. However, if the line source has the strike and dip regarding the observation profile, it comes to have five independent components. The proposed algorithm of the determination both strike and dip is based on the rotational transform that converts full gravity gradient tensor to reduced 2-D gravity gradient tensor. The least-square method is applied in order to find optimum rotational angles that make one of the row components minimalized simultaneously. The two synthetic cases of a line source are represented; one has strike only and the other has both strike and dip. This study finds that the automatic determination method using gravity gradient tensor can find directions of a line source in each case.

• Advances in nano research
• /
• v.8 no.1
• /
• pp.13-24
• /
• 2020

In this paper, for the first time based on the nonlocal strain gradient theory the effect of size dependency in torsional vibration of bi-direction functionally graded (FG) nonlinear nano-cone is study. The material properties were assumed to vary according to the arbitrary function in radial and axial directions. The Navier equation and boundary conditions of the size-dependent bidirectional FG nonlinear nano-cone were derived by Hamilton's principle. These equations were solved by employing the generalized differential quadrature method (GDQM). The presented model can turn into the classical model if the material length scale parameters are taken to be zero. The effects of some parameters, such as inhomogeneity constant, cross-sectional area parameter and small-scale parameters, were studied. As an essential result of this study can be stated that an FG nano-cone model based on the nonlocal elasticity theory behaves softer and based on the strain gradient theory behaves harder.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.2
• /
• pp.59-68
• /
• 1993

The motion vector estimation and motion boundary detection have been briskly studied since they are an important clue for analysis of object structure and 3-d motion. The purpose of this researches is more exact estimation, but there are two main causes to make inaccurate. The one is the erroneous measurement of gradients in brightness values and the other is the blurring of motion boundries which is caused by the smoothness constraint. In this paper, we analyze the gradient measurement error of conventional methods and propose new technique based on it. When the proposed method is applied to the motion boundary detection in Schunck and motion vector estimation in Horn & Schunck, it is shown to have much better performance than conventional method is some artificial and real image sequences.

• Smart Structures and Systems
• /
• v.18 no.5
• /
• pp.839-864
• /
• 2016

This paper presents an efficient method for updating the structural finite element model. Model updating is performed through minimizing the difference between the recorded acceleration of a real damaged structure and a hypothetical damaged one. This is performed by updating physical parameters (module of elasticity in this study) in each step using iterative process of modified nonlinear conjugate gradient (M-NCG) and modified Broyden-Fletcher-Goldfarb-Shanno algorithm (M-BFGS) separately. These algorithms are based on sensitivity analysis and provide a solution for nonlinear damage detection problem. Three illustrative test examples are considered to assess the performance of the proposed method. Finally, it is demonstrated that the proposed method is satisfactory for detecting the location and ratio of structural damage in presence of noise.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.10B
• /
• pp.1832-1840
• /
• 2000

The error diffusion algorithm is good for reproducing continuous images to binary images. However the reproduction of edge characteristics is weak in power spectrum an analysis of display error. In this paper an edge enhanced error diffusion method is proposed to improve the edge characteristic enhancement. Spatial gradient information in original image is adapted for edge enhance in threshold modulation of error diffusion. First the horizontal and vertical second order differential values are obtained from the gradient of peripheral pixels(3x3) in original image. second weighting function is composed by function including absolute value and sign of second order differential values. The proposed method presents a good visual results which edge characteristics is enhanced. The performance of the proposed method is compared with that of the conventional edge enhanced error diffusion by measuring the edge correlation and the local average accordance over a range of viewing distances and the RAPSD of display error.

• Proceedings of the Korean Society of Surveying, Geodesy, Photogrammetry, and Cartography Conference
• /
• 2010.04a
• /
• pp.227-229
• /
• 2010

This study presents an application of the 3D patch extraction method which is based on gradient-driven properties to obtain 3D planar patches over the terrain and man-made objects from lidar data. The method which was exploited in this study is composed of a sequence of processes: segmentation by slope, initiation of triggering patches by mode selection, and expansion of the triggering patches. Since urban areas contain many planar regions over the terrain surface, application of the method has been experimented to extract 3D planar patches not only from non-terrain objects but also from the terrain. The experimental result shows that the method is efficient to acquire 3D planar patches.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.1
• /
• pp.118-124
• /
• 2004

In this paper, we present a indirect adaptive control method using a wavelet neural network (WNN) for the control of chaotic nonlinear systems without precise mathematical models. The proposed indirect adaptive control method includes the off-line identification and on-line control procedure for chaotic nonlinear systems. In the off-line identification procedure, the WNN based identification model identifies the chaotic nonlinear system by using the serial-parallel identification structure and is trained by the gradient-descent method. And, in the on-line control procedure, a WNN controller is designed by using the off-line identification model and is trained by the error back-propagation algorithm. Finally, the effectiveness and feasibility of the proposed control method is demonstrated with applications to the chaotic nonlinear systems.