Fourier三角基神经元网络的权值直接确定法

根据Fourier变换理论,本文构造出一类基于三角正交基的前向神经网络模型。该模型由输入层、隐层、输出层构成,其输入层和输出层采用线性激励函数,以一组三角正交基为其隐层神经元的激励函数。依据误差回传算法(即BP算法),推导了权值修正的迭代公式。针对BP迭代法收敛速度慢、逼近目标函数精度较低的缺点,进一步提出基于伪逆的权值直接确定法,该方法避免了权值反复迭代的冗长过程。仿真和预测结果表明,该方法比传统的BP迭代法具有更快的计算速度和更高的仿真与测试精度。     

切比雪夫正交基神经网络的权值直接确定法

经典的BP神经网络学习算法是基于误差回传的思想.而对于特定的网络模型,采用伪逆思想可以直接确定权值进而避免以往的反复迭代修正的过程.根据多项式插值和逼近理论构造一个切比雪夫正交基神经网络,其模型采用三层结构并以一组切比雪夫正交多项式函数作为隐层神经元的激励函数.依据误差回传(BP)思想可以推导出该网络模型的权值修正迭代公式,利用该公式迭代训练可得到网络的最优权值.区别于这种经典的做法,针对切比雪夫正交基神经网络模型,提出了一种基于伪逆的权值直接确定法,从而避免了传统方法通过反复迭代才能得到网络权值的冗长训练过程.仿真结果表明该方法具有更快的计算速度和至少相同的工作精度,从而验证了其优越性.     

改造前向神经网络结构以求网络权值直接确定——Jacobi正交基神经网络实例

根据多项式插值与逼近理论,以一组零阀值特例Jacobi正交多项式(第二类Chebyshev正交多项式)作为隐层神经元的激励函数,构造一种基于Jacobi正交基的前向神经网络模型.该神经网络模型采用三层前向结构,其中输入、输出层神经元采用零阀值线性激励函数.为改进传统神经网络收敛速度较慢及其局部极小点问题,针对该Jacobi正交基前向神经网络模型,提出了一种基于伪逆的直接计算神经网络权值的方法(即,一步确定法),并利用该神经网络进行预测.计算机仿真结果表明,相对比传统的BP迭代训练方法,权值直接确定法计算速度更快、预测精度更高.     

基于权值与结构确定法的单极Sigmoid神经网络分类器

构造了以单极Sigmoid函数作为隐层神经元激励函数的神经网络分类器,网络中输入层到隐层的权值和隐层神经元的阈值均为随机生成。同时,结合利用伪逆思想一步计算出隐层和输出层神经元之间连接权值的权值直接确定(WDD)法,进一步提出了具有边增边删和二次删除策略的网络结构自确定法,用来确定神经网络最优权值和结构。数值实验结果表明,该算法能够快速有效地确定单极Sigmoid激励函数神经网络分类器的最优网络结构; 分类器的分类性能良好。     

两输入幂激励前向神经网络权值与结构确定

基于多元函数逼近与二元幂级数展开理论,构建了一个以二元幂函数序列为隐神经元激励函数的两输入幂激励前向神经网络模型.以该网络模型为基础,基于权值直接确定法以及隐神经元数目与逼近误差的关系,提出了一种网络权值与结构确定算法.计算机仿真与数值实验结果验证了所构建的网络在逼近与去噪方面具有优越的性能,所提出的权值与结构确定算法能够快速、有效地确定网络的权值与最优结构,保证网络的最佳逼近能力.     

基于最小二乘法的SIMO傅里叶神经网络研究

利用傅里叶级数的原理，构造单输入、多输出(SIMO)傅里叶神经网络，将非线性映射转化成为线性映射，将求解神经网络权值的方法由非线性优化方法转化成为线性优化方法，并采用最小二乘法计算网络的权值，从而大大提高了神经网络的收敛速度并避免了局部极小问题.而且，在训练输出样本受白噪声影响时，最小二乘法具有良好的降低噪声影响的功能．     

基函数神经网络权值直接确定的图像复原

给出了基函数神经网络图像复原的模型,该神经网络模型是由三层构成的前向神经网络,以一组正交基为隐层神经元的激励函数。为了避免反复迭代权值修正的冗长BP训练过程,提出了一种权值直接确定的算法。实验结果表明,该种权值直接确定算法不仅能一步确定权值而获得更快的运算速度,而且能达到更高的精度。     

多输入La}uerre正交多项式前向神经网络权值与结构确定法

为克服邵神经网络模型及其学习算法中的固有缺陷,根据多项式插值和逼近理论,构造出一种以工;agucrre正交多项式作为隐层神经元激励函数的多输入前向神经网络模型。针对该网络模型,提出了权值与结构确定法,以便快速、自动地确定该网络的最优权值和最优结构。计算机仿真与实验结果显示:该算法是有效的,并且通过该算法所得到的网络具有较优的逼近性能和良好的去噪能力。     

量子小波神经网络及其在漏钢预报中的应用

针对传统神经网络收敛速度慢,收敛精度低,以及用于模式识别泛化能力差的问题。提出了将量子神经网络与小波理论相结合的量子小波神经网络模型。该模型隐层量子神经元采用小波基函数的线性叠加作为激励函数,称之为多层小波激励函数,这样隐层神经元既能表示更多的状态和量级,又能提高网络收敛精度和速度。给出了网络学习算法。并以之在漏钢预报波形识别中的应用验证了该模型和学习算法的有效性。     

一种权值直接确定及结构自适应的Chebyshev基函数神经网络

基于函数逼近理论,构造一种Chebyshev基函数神经网络模型.推导出该网络模型的权值直接确定方法,可一步计算出权值,克服了传统BP神经网络学习率选取困难、学习过程冗长和易陷入局部极小等缺点.在此基础上,设计了基于二分搜索的结构自适应算法,根据精度要求自动确定网络最优结构.理论分析及仿真验证均表明,该网络不仅能够快速地完成网络权值确定和结构自适应,且具有优异的学习与逼近能力,而且对随机加性噪声也具有较好的抑制作用.     

基于Fourier神经网络的图像复原算法*

由于退化图像的点扩散函数难以准确确定,提出一种基于Fourier正交基函数的前向神经网络图像复原模型,该模型以一组Fourier正交基为隐层神经元的激励函数,根据误差传递算法进行权值修正,达到收敛目标。给出Fourier神经网络及其相应的衍生算法的图像恢复实现步骤。实验表明,该方法能较好地实现图像的复原。     

Optimization of the weights and asymmetric activation function family of neural network for time series forecasting

The use of neural network models for time series forecasting has been motivated by experimental results that indicate high capacity for function approximation with good accuracy. Generally, these models use activation functions with fixed parameters. However, it is known that the choice of activation function strongly influences the complexity and neural network performance and that a limited number of activation functions has been used in general. We describe the use of an asymmetric activation functions family with free parameter for neural networks. We prove that the activation functions family defined, satisfies the requirements of the universal approximation theorem We present a methodology for global optimization of the activation functions family with free parameter and the connections between the processing units of the neural network. The main idea is to optimize, simultaneously, the weights and activation function used in a Multilayer Perceptron (MLP), through an approach that combines the advantages of simulated annealing, tabu search and a local learning algorithm. We have chosen two local learning algorithms: the backpropagation with momentum (BPM) and Levenberg–Marquardt (LM). The overall purpose is to improve performance in time series forecasting.     

Adaptive iterative learning control for nonlinear time-delay systems with periodic disturbances using FSE-neural network

An adaptive iterative learning control scheme is presented for a class of strict-feedback nonlinear time-delay systems, with unknown nonlinearly parameterised and time-varying disturbed functions of known periods. Radial basis function neural network and Fourier series expansion (FSE) are combined into a new function approximator to model each suitable disturbed function in systems. The requirement of the traditional iterative learning control algorithm on the nonlinear functions (such as global Lipschitz c...     

Multiplicative neuron model artificial neural network based on Gaussian activation function

Multiplicative neuron model-based artificial neural networks are one of the artificial neural network types which have been proposed recently and have produced successful forecasting results. Sigmoid activation function was used in multiplicative neuron model-based artificial neural networks in the previous studies. Although artificial neural networks which involve the use of radial basis activation function produce more successful forecasting results, Gaussian activation function has not been used for multiplicative neuron model yet. In this study, rather than using a sigmoid activation function, Gaussian activation function was used in multiplicative neuron model artificial neural network. The weights of artificial neural network and parameters of activation functions were optimized by guaranteed convergence particle swarm optimization. Two major contributions of this study are as follows: the use of Gaussian activation function in multiplicative neuron model for the first time and the optimizing of central and propagation parameters of activation function with the weights of artificial neural network in a single optimization process. The superior forecasting performance of the proposed Gaussian activation function-based multiplicative neuron model artificial neural network was proved by applying it to real-life time series.     

Neural network approximation for periodically disturbed functions and applications to control design

This paper addresses the approximation problem of functions affected by unknown periodically time-varying disturbances. By combining Fourier series expansion into multilayer neural network or radial basis function neural network, we successfully construct two kinds of novel approximators, and prove that over a compact set, the new approximators can approximate a continuously and periodically disturbed function to arbitrary accuracy. Then, we apply the proposed approximators to disturbance rejection in the first-order nonlinear control systems with periodically time-varying disturbances, but it is straightforward to extend the proposed design methods to higher-order systems by using adaptive backstepping technique. A simulation example is provided to illustrate the effectiveness of control schemes designed in this paper.     

Novel method and system for pattern recognition and processing using data encoded as Fourier series in Fourier space

A method and system for pattern recognition and processing is reported that has a data structure and theoretical basis that are unique. This novel approach anticipates the signal processing action of an ensemble of neurons as a unit and intends to simulate aspects of brain that give rise to capabilities such as intelligence, pattern recognition, and reasoning that have not been reproduced with past approaches such as neural networks that are based individual simulated “neuronal units.” Information representative of physical characteristics or representations of physical characteristics is transformed into a Fourier series in Fourier space within an input context of the physical characteristics that is encoded in time as delays corresponding to modulation of the Fourier series at corresponding frequencies. Associations are formed between Fourier series by filtering the Fourier series and by using a spectral similarity between the filtered Fourier series to determine the association based on Poissonian probability. The associated Fourier series are added to form strings of Fourier series. Each string is ordered by filtering it with multiple selected filters to form multiple time order formatted subset Fourier series, and by establishing the order through associations with one or more initially ordered strings to form an ordered string. Associations are formed between the ordered strings to form complex ordered strings that relate similar items of interest. The components of the system based on the algorithm are active based on probability using weighting factors based on activation rates.     

A neural-network-based model for 2D microwave imaging of cylinders

In this article, a neural network with radial-basis functions (RBF-NN) is applied to microwave imaging of cylinders. Initially, the shape function of the target cylinder is expanded by a Fourier series. The RBF-NN is trained by some direct-scattering data sets and thus can predict the images of the target cylinders. © 2004 Wiley Periodicals, Inc. Int J RF and Microwave CAE 14, 398–403, 2004.     

采用双折线步方法的傅里叶神经网络

目前神经网络已经成为解决非线性系统辨识问题的一类有效的方法,但是常用的多层感知器存在网络稳定性差、收敛速度慢的问题.在多层感知器和傅里叶级数基础上提出的傅里叶神经网络具有较好的泛化性、模式识别能力,但其学习算法主要采用最速下降法,易产生陷入局部极小,学习速度慢等问题.提出一种采用双折线步方法的傅里叶神经网络,避免了局部极小问题,且具有二阶收敛速度.通过相应的数值算例验证新算法的性能,并应用于非线性系统的识别问题中,其结果和几类经典的神经网络算法做了相应的对比和分析.