作者：
Yutaka MaedaKansai University
Faculty of Engineering Department of Electrical Engineering 3-3-35 Yamate-cho Suita Osaka 564 JAPAN
In this paper, we propose recursive algorithms using time difference simultaneous perturbation to find a minimum or maximum point of a function. Instead of a gradient vector used in the well-known gradient method, the...
详细信息
In this paper, we propose recursive algorithms using time difference simultaneous perturbation to find a minimum or maximum point of a function. Instead of a gradient vector used in the well-known gradient method, the algorithms need only one value of the function for every modification of an estimator. The algorithms are very simple. Therefore, applications of these algorithms to many problems such as control of dynamical system or learning rule of neural networks are very easy. A numerical simulation is shown. The paper contains a comparison with the random search, the finite difference and the simultaneous perturbation method.
A unified presentation of recursive algorithms for plant model identification in closed loop is given. From the basic formulation of the problem of finding the plant model which gives the best predictor for the closed...
详细信息
ISBN:
(纸本)0780335910
A unified presentation of recursive algorithms for plant model identification in closed loop is given. From the basic formulation of the problem of finding the plant model which gives the best predictor for the closed loop system the two families of algorithms using either a re-parameterized predictor for the closed loop or a plant predictor operating on filtered data are presented and their asymptotic properties are examined. Validation tests for the models identified in closed loop are proposed.
A recursive algorithm for optimizing the architecture of feedforward neural networks by the stepwise addition of a reasonable number of hidden nodes is proposed. The recursive algorithm retains the calculation results...
详细信息
A recursive algorithm for optimizing the architecture of feedforward neural networks by the stepwise addition of a reasonable number of hidden nodes is proposed. The recursive algorithm retains the calculation results and approximation precision already obtained in the previous iteration step and uses them in the next step to efficiently lighten the computational burden of network optimization and training. The commonly used genetic algorithm has been modified for network training to circumvent the local optimum problem. Some new genetic operators, competition and self-reproduction, have been introduced and used together with some substantially modified genetic operators, crossover and mutation, to form a modified genetic algorithm (MGA) which ensures asymptotic convergence to the global optima with relatively high efficiency. The proposed methods have been successfully applied to concentration estimation in chemical analysis and quantitative structure-activity relationship studies of chemical compounds.
Numerical computation of Fourier series is an important problem of the numerical application theory. In this paper, we present two recursive algorithms for evaluation of the Fourier series, f(x) = a0/2 + Σni=1(ai cos...
详细信息
Numerical computation of Fourier series is an important problem of the numerical application theory. In this paper, we present two recursive algorithms for evaluation of the Fourier series, f(x) = a0/2 + Σni=1(ai cos ix + bi sin ix), where ai and bi are known complex constants and x is an evaluation point. Compared to some traditional algorithms, the new algorithms take only half of the arithmetic operations.
In this paper several recursive algorithms for computing M-estimates in multivariate linear regression models are discussed. It is shown that the recursive M-estimators of regression coefficient and scatter parameters...
详细信息
In this paper several recursive algorithms for computing M-estimates in multivariate linear regression models are discussed. It is shown that the recursive M-estimators of regression coefficient and scatter parameters are strongly consistent. In particular, the asymptotic normality of the recursive M-estimators regression coefficients is established. (C) 1996 Academic Press, Inc.
A general, recursive algorithm is presented for computing the expected Fisher information matrix for state-space model parameters. Simulation results are featured where known Fisher information matrices corresponding ...
详细信息
A general, recursive algorithm is presented for computing the expected Fisher information matrix for state-space model parameters. Simulation results are featured where known Fisher information matrices corresponding to simple state-space models are estimated using both observed and expected information matrices. The accuracy of the two approaches is compared.
作者：
Lee, PZLiu, GS[a] Institute of Information Science
Academia Sinica 128 Sec. 2 Yen-Chiu-Yun Road Nankang Taipei 11529 Taiwan ROC
[b] Vanguard International Semiconductor Corporation Hsin Chu Taiwan ROC
In this paper, we present an efficient algorithm for computing the 2-D discrete cosine transform (DCT). First, we arrange the input data matrix in a certain order. Second, we formulate the output data matrix to a bloc...
详细信息
In this paper, we present an efficient algorithm for computing the 2-D discrete cosine transform (DCT). First, we arrange the input data matrix in a certain order. Second, we formulate the output data matrix to a block-structured matrix-vector multiplication, and show that the block-structured matrix has the same properties as the 1-D DCT kernel matrix. Therefore, the 2-D DCT problem is reduced to the 1-D DCT problem. We thus can provide a procedure for computing the 2-D DCT which is similar to that of the 1-D DCT. The primary appeal of this algorithm lies not only in its low computational cost but also in its regular structure as the transform size increases.
This paper introduces wavelets and shows that they may be efficient and useful for the power distribution relaying. The wavelet transform of a signal consists in measuring the ''similarity'' between th...
详细信息
This paper introduces wavelets and shows that they may be efficient and useful for the power distribution relaying. The wavelet transform of a signal consists in measuring the ''similarity'' between the signal and a set of translated and scaled versions of ''mother wavelet''. The ''mother wavelet'' is a chosen fast decaying oscillation function. Wavelets are used to analyse transient earth fault signals in a 20 kV resonant grounded network as generated by EMTP. It is shown that wavelets may be employed for analysing records to study efficiently the faulted network. Moreover, this new technique can actually be implemented in real time for protection devices. Thus, it is suitable for application to protective relays.
Efficient algorithms for shape preserving approximation to curves and surfaces are very important in shape design and modelling in CAD/CAM systems. In this paper, a local algorithm using piecewise generalized conic se...
详细信息
Efficient algorithms for shape preserving approximation to curves and surfaces are very important in shape design and modelling in CAD/CAM systems. In this paper, a local algorithm using piecewise generalized conic segments is proposed for shape preserving curve interpolation. It is proved that there exists a smooth piecewise generalized conic curve which not only interpolates the data points, but also preserves the convexity of the data. Furthermore, if the data is strictly convex, then the interpolant could be a locally adjustable GC(2) curve provided the curvatures at the data points are well determined. It is also shown that the best approximation order is O(h(6)). An efficient algorithm for the simultaneous computation of points on the curve is derived so that the curve can be easily computed and displayed. The numerical complexity of the algorithm for computing N points on the curve is about 2N multiplications and N additions. Finally, some numerical examples with graphs are provided and comparisons with both quadratic and cubic spline interpolants are also given.
The Wilcoxon signed rank test is a well known statistical test based on Wilcoxon's T-n(+) statistic. When n is large enough, the lower tail probability P-0 (T-n(+) less than or equal to x) of T-n(+) under H-0 : th...
详细信息
The Wilcoxon signed rank test is a well known statistical test based on Wilcoxon's T-n(+) statistic. When n is large enough, the lower tail probability P-0 (T-n(+) less than or equal to x) of T-n(+) under H-0 : theta = 0 may be computed easily by a normal approximation. We propose in this paper a recursive, real-time algorithm which computes this probability when n is small.
暂无评论