• Proceedings of the IEEK Conference
• /
• 1998.10a
• /
• pp.1191-1194
• /
• 1998

This paper presents a new computationally efficient adaptive algorithm for blind signal separation, which is able to recover the narrowband source signals in the presence of cochannel interference without a prior knowledge of array manifold. We derive a new blind signal separation algorithm using the Natural gradient 〔1〕from an information-theoretic approach. The resulting algorithm has the Bussgang property which has been widely used in blind equalization 〔12〕. Extensive computer simulation results comfirm the validity and high performance of the proposed algorithm.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1987.11a
• /
• pp.79-81
• /
• 1987

In this paper we propose a fast adaptive least squares algorithm for linear phase FIR filters. The algorithm requires 10m multiplications per data point where m is the filter order. Both linear phase cases with constant phase delay and constant group delay are examined. Simulation results demonstrate that the proeposed algorithm is superior to the LMS gradient algorithm and the averaging scheme used for the modified fast Kalman algorithm.

• Proceedings of the KIEE Conference
• /
• 1989.07a
• /
• pp.457-460
• /
• 1989

In this paper, an improvement of the EBP(error back propagation) learning speed using an acceleration algorithm is described. Using an acceleration algorithm known as the Partan method in the gradient search algorithm, learning speed is 25% faster than the original EBP algorithm in the simulaion results.

• Proceedings of the KIPE Conference
• /
• 1998.10a
• /
• pp.293-297
• /
• 1998

This paper presents a newly developed speed sensorless drive using RLS based on Neural Network Training Algorithm. The proposed algorithm has just the time-varying learning rate, while the wellknown back-propagation algorithm based on gradient descent has a constant learning rate. The number of iterations required by the new algorithm to converge is less than that of the back-propagation algorithm. The theoretical analysis and experimental results to verify the effectiveness of the proposed control strategy are described.

• Proceedings of the Korean Information Science Society Conference
• /
• 1998.10c
• /
• pp.360-362
• /
• 1998

In the original mixtures of experts framework, the parameters of the network are determined by gradient descent, which is naturally slow. In [2], the Expectation-Maximization(EM) algorithm is used instead, to obtain the network parameters, resulting in substantially reduced training times. This paper presents the new EM algorithm for prediction. We show that an Efficient training algorithm may be derived for the HME network. To verify the utility of the algorithm we look at specific examples in time series prediction. The application of the new EM algorithm to time series prediction has been quiet successful.

• ETRI Journal
• /
• v.33 no.3
• /
• pp.401-406
• /
• 2011

This paper presents a fast block-matching algorithm based on the normalized cross-correlation, where the elimination order is determined based on the gradient magnitudes of subblocks in the current macroblock. Multilevel Cauchy-Schwartz inequality is derived to skip unnecessary block-matching calculations in the proposed algorithm. Also, additional complexity reduction is achieved re-using the normalized cross correlation values for the spatially neighboring macroblock because the search areas of adjacent macroblocks are overlapped. Simulation results show that the proposed algorithm can improve the speed-up ratio up to about 3 times in comparison with the existing algorithm.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.23 no.8
• /
• pp.1984-1989
• /
• 1998

Convergence properties of the stochastic gradient adaptive algorithm based on the least mean absolute third (LMAT) error criterion is presented.In particular, the performnce of the algorithmis examined and compared with least mena square (LMS) algorithm for several different probability densities of the measurement noisein a system identification mode. It is observedthat the LMAT algorithm outperforms the LMS algorithm for most of the noise probability densities, except for the case of the exponentially distributed noise.

• 전기의세계
• /
• v.21 no.3
• /
• pp.48-52
• /
• 1972

The application of pontryagin's Maximum Principle to the optimal control eventually leads to the problem of solving the two point boundary value problem. Most of problems have been related to their own special factors, therfore it is very hard to recommend the best method of deriving their optimal solution among various methods, such as iterative Runge Kutta, analog computer, gradient method, finite difference and successive approximation by piece-wise linearization. The gradient method has been applied to the optimal control of two point boundary value problem in the power systems. The most important thing is to set up some objective function of which the initial value is the function of terminal point. The next procedure is to find out any global minimum value from the objective function which is approaching the zero by means of gradient projection. The algorithm required for this approach in the relevant differential equations by use of the Runge Kutta Method for the computation has been established. The usefulness of this approach is also verified by solving some examples in the paper.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2013.06a
• /
• pp.260-262
• /
• 2013

본 논문에서는 gradient inverse weighted filtering(GIWF) 보간법 기반의 화면 내 디인터레이싱 알고리듬을 제안한다. 소실된 화소 줄을 보간하기 위해서 먼저 선 처리 과정을 통해 정해진 마스크의 중간 지점의 예측 값을 생성한다. 이 때는 4-tap 필터를 이용을 한다. 다음으로 이웃 픽셀들 사이에서 마스크 내의 기울기 정보를 이용하여 gradient-weighted 필터의 가중치 계수를 계산한다. 그리고 마지막으로 새롭게 제시된 GIWF 보간법을 이용하여 소실된 화소 줄을 보간하게 된다. GIWF를 이용하여 영상의 디테일을 보존하고 잡음을 제거하는 효과를 얻게 되었다. 제안된 방법의 영상 시퀀스에 대한 실험 결과는 기존의 방법들에 비하여 성능의 우수함을 보여준다.

• Journal of Ocean Engineering and Technology
• /
• v.20 no.6 s.73
• /
• pp.61-66
• /
• 2006

Two iterative solvers are applied to solve the modified mild slope equation. The elliptic formulation of the governing equation is selected for numerical treatment because it is partly suited for complex wave fields, like those encountered inside harbors. The requirement that the computational model should be capable of dealing with a large problem domain is addressed by implementing and testing two iterative solvers, which are based on the Stabilized Bi-Conjugate Gradient Method (BiCGSTAB) and Generalized Conjugate Gradient Method (GCGM). The characteristics of the solvers are compared, using the results for Berkhoff's shoal test, used widely as a benchmark in coastal modeling. It is shown that the GCGM algorithm has a better convergence rate than BiCGSTAB, and preconditioning of these algorithms gives more than half a reduction of computational cost.