• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.17 no.4
• /
• pp.279-294
• /
• 2013

We are interested in an adaptive grid method for hyperbolic equations. A multiresolution analysis, based on a biorthogonal family of interpolating scaling functions and lifted interpolating wavelets, is used to dynamically adapt grid points according to the physical field profile in each time step. Traditional finite-difference schemes with fixed stencils produce high oscillations around sharp discontinuities. In this paper, we hybridize high-resolution schemes, which are suitable for capturing singularities, and apply a finite-difference approach to the scaling functions at non-singular points. We use a total variation diminishing Runge-Kutta method for the time integration. The computational cost is proportional to the number of points present after compression. We provide several numerical examples to verify our approach.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.4
• /
• pp.775-782
• /
• 2006

We generated blocks from the direction-extracted fingerprint during the pre-process of the fingerprint recognition algorithm and performed training by using the direction minutiae of each block as the input pattern of the neural network, so that we extracted the core points to use in the matching. Based on this, we designed the fingerprint recognition embedded system and tested it using the control board and the serial communication to utilize it for a variety of application systems. As a result, we can verify the reliance satisfactorily.

• Journal of Advanced Marine Engineering and Technology
• /
• v.33 no.5
• /
• pp.738-745
• /
• 2009

Underwater acoustic communication has multipath error because of reflection by sea-level and sea-bottom. The multipath of underwater channel causes receive signal to make error floor. In this paper, we propose the underwater communication system using various channel coding schemes such as RS coding, convolutional code, turbo code and concatenated code for overcoming the multipath effect in underwater channel. As shown in simulation results, characteristic of multipath error is similar to that of random error. So interleaver has not effect on error correcting. For correcting of error floor by multipath, it is necessary to use strong channel codes like turbo code. Turbo code is one of the iterative codes. And the performance of concatenated codes including RS code has better performance than using singular channel codes.

• Journal of Mechanical Science and Technology
• /
• v.19 no.8
• /
• pp.1611-1620
• /
• 2005

This paper shows the use of wavelet transformation combined with inverse acoustics to reconstruct the surface velocity of a noise source. This approach uses the boundary element analysis based on the measured sound pressure at a set of field points, the Helmholtz integral equations and wavelet transformation for reconstructing the normal surface velocity field. The reconstructed field can be diverged due to the small measurement errors in the case of nearfield acoustic holography (NAH) using an inverse boundary element method. In order to avoid this instability in the inverse problem, the reconstruction process should include some form of regularization for enhancing the resolution of source images. The usual method of regularization has been the truncation of wave vectors associated with small singular values, although the order of an optimal truncation is difficult to determine. In this paper, a wavelet transformation is applied to reduce the computation time for inverse acoustics and to enhance the reconstructed vibration field. The computational speed-up is achieved, with solution time being reduced to $14.5\%$.

• Proceedings of the IEEK Conference
• /
• summer
• /
• pp.263-268
• /
• 2004

The inverse kinematics problem is to find a set of joint variable values that will place the end effector of a robot manipulator into a given pose. Pieper has shown that a sufficient condition for a manipulator to have a closed form solution is that three adjacent joint axes intersects, hence the six axes robot with spherical wrist allows closed form solution. But many industrial robots have a non-spherical wrist to provide a stronger wrist configuration so that they can handle heavy payloads. Also, the use of a non-spherical wrist can result in a cheap and simple wrist arrangement than when all three axes intersect at a common point. In these cases, closed form solutions cannot be found. Therefore numerical technique must be used to solve the inverse kinematics equations. This paper proposes a new algorithm that can be used for finding inverse kinematics solution of the six axes robot with non-spherical wrist. Computer simulations are provided to prove the usefulness of our method.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.4 no.3
• /
• pp.347-352
• /
• 2004

We generated blocks from the direction-extracted fingerprint during the pre-process of the fingerprint recognition algorithm and performed training by using the direction minutiae of each block as the input pattern of the neural network, so that we extracted the core points to use in the matching. Based on this, we designed the fingerprint recognition embedded system and tested it by using the control board and the serial communication to utilize it for a variety of application systems. As a result, we can verify the reliance satisfactorily.

• Proceedings of the IEEK Conference
• /
• 2007.07a
• /
• pp.481-482
• /
• 2007

This paper demonstrates OFDM with adaptive modulation applied to Multiple-Input Multiple-Output (MIMO) systems. We apply an optimization algorithm to obtain a bit and power allocation for each subcarrier assuming instantaneous channel knowledge. The analysis and simulation is considered in two stages. The first stage involves the application of a variable-rate variable-power MQAM technique for a Single-Input Single-Output(SISO) OFDM system. This is compared with the performance of fixed OFDM transmission where a constant rate is applied to each subcarrier. The second stage applies adaptive modulation to a general MIMO system by making use of the Singular Value Decomposition to separate the MIMO channel into parallel subchannels. For a two-input antenna, two-output antenna system, the performance is compared with the performance of a system using selection diversity at the transmitter and maximal ratio combining at the receiver.

• East Asian mathematical journal
• /
• v.32 no.1
• /
• pp.129-138
• /
• 2016

In this paper, we propose an efficient regularization technique (The Method of Regularized Ratios) for the reconstruction of the shape of a rigid elastic scatterer from far field measurements. The approach used is based on the factorization method and creates via Picard's condition ratios, baptized Regularized Ratios, that serve to effectively remove unwanted singular values that may lead to poor reconstructions. This is achieved through the use of a sophisticated algorithm that progressively adjusts an initially set moderate tolerance. In comparison with the well established Tikhonov-Morozov regularization techniques our new algorithm appears to be more computationally efficient as it doesn't require computation of the regularization parameter for each point in the grid.

• Kyungpook Mathematical Journal
• /
• v.48 no.2
• /
• pp.195-206
• /
• 2008

The purpose of this note is to compare the rings of continuous functions, integer-valued or real-valued, in pointfree topology with those in classical topology. To this end, it first characterizes the Boolean frames (= complete Boolean algebras) whose function rings are isomorphic to a classical one and then employs this to exhibit a large class of frames for which the functions rings are not of this kind. An interesting feature of the considerations involved here is the use made of nonmeasurable cardinals. In addition, the integer-valued function rings for Boolean frames are described in terms of internal lattice-ordered ring properties.

• ETRI Journal
• /
• v.41 no.1
• /
• pp.109-116
• /
• 2019

This paper explores new techniques that are based on a hidden-layer linear transformation for fast speaker adaptation used in deep neural networks (DNNs). Conventional methods using affine transformations are ineffective because they require a relatively large number of parameters to perform. Meanwhile, methods that employ singular-value decomposition (SVD) are utilized because they are effective at reducing adaptive parameters. However, a matrix decomposition is computationally expensive when using online services. We propose the use of an extended diagonal linear transformation method to minimize adaptation parameters without SVD to increase the performance level for tasks that require smaller degrees of adaptation. In Korean large vocabulary continuous speech recognition (LVCSR) tasks, the proposed method shows significant improvements with error-reduction rates of 8.4% and 17.1% in five and 50 conversational sentence adaptations, respectively. Compared with the adaptation methods using SVD, there is an increased recognition performance with fewer parameters.