Treatment of skewed multi-dimensional training data to facilitate the task of engineering neural models

Successful application of neural network models relies heavily on problem-dependent internal parameters. As the theory does not facilitate the choice of the optimal parameters of neural models, these can solely be obtained through a tedious trial-and-error process. The process requires performing multiple training simulations with various network parameters, until satisfactory performance criteria of a neural model are met. In literature, it has been shown that neural models are not consistently good in prediction under highly skewed data. Consequently, the cost of engineering neural models rises in such circumstance to seek for appropriate internal parameters. In this paper the aim is to show that a recently proposed treatment of highly skewed data eases the task of practitioners in engineering neural network models to meet satisfactory performance criteria. As the applications of neural models grows dramatically in diverse engineering domains, the understanding of the treatment show indispensable practical values.     

Assessment of the performances of multilayer perceptron neural networks in comparison with recurrent neural networks and two statistical methods for diagnosing coronary artery disease

Abstract: We aimed to examine the diagnostic performances of multilayer perceptron neural networks (MLPNNs) for predicting coronary artery disease and to compare them with different types of artificial neural network methods, namely recurrent neural networks (RNNs) and two statistical methods (quadratic discriminant analysis (QDA) and logistic regression (LR)). MLPNNs were trained with backpropagation, quick propagation, delta-bar-delta and extended delta-bar-delta algorithms as classifiers; the RNN was trained with the Levenberg–Marquardt algorithm; LR and QDA were used for predicting coronary artery disease. Coronary artery disease was classified with accuracy rates varying from 79.9% to 83.9% by MLPNNs. Even though MLPNNs achieved higher accuracy rates than the statistical methods, LR (73.2%) and QDA (58.4%), their performances were lower compared to the RNN (84.7%). Among the four different types of training algorithms that trained MLPNNs, quick propagation achieved the highest accuracy rate; however, it was lower than the RNN trained with the Levenberg–Marquardt algorithm. RNNs, which demonstrated 84.7% accuracy and 86.5% positive predictive rates, may be a helpful tool in medical decision making for diagnosis of coronary artery disease.     

基于小波描述子和神经网络的形状识别

提出一种把小波描述子和神经网络相结合的形状识别方法。通过小波描述子提取待识别物体的形状特征,所提取的特征值不受物体位移、缩放和旋转的影响,接着用多层分类器网络对物体的形状进行识别分类,并采用BP算法对神经网络进行学习和训练。最后得出令人满意的实验结果。     

A versatile strategy for achieving the second-order advantage when applying different artificial neural networks to non-linear second-order data: Unfolded principal component analysis/residual bilinearization

Second-order instrumental signals showing a non-linear behaviour with respect to analyte concentration can still be adequately processed in order to achieve the important second-order advantage. The combination of unfolded principal component analysis with residual bilinearization, followed by application of a variety of neural network models, allows one to obtain the second-order advantage. While principal component analysis models the training data, residual bilinearization models the contribution of the potential interferents which may be present in the test samples. Neural networks such as multilayer perceptron, radial basis functions and support vector machines, are all able to model the non-linear relationship between analyte concentrations and sample principal component scores. Three different experimental systems have been analyzed, all requiring the second-order advantage: 1) pH–UV absorbance matrices for the determination of two active principles in pharmaceutical preparations, 2) fluorescence excitation–emission matrices for the determination of polycyclic aromatic hydrocarbons, and 3) UV-induced fluorescence excitation–emission matrices for the determination of amoxicillin in the presence of salicylate. In all cases, reasonably accurate predictions can be made with the proposed techniques, which cannot be reached using traditional methods for processing second-order data.     

基于GWO-MLP的光伏系统输出功率短期预测模型

准确预测光伏系统的输出功率对微网系统的优化调度具有重要意义。为了减小光伏系统输出功率短期预测误差,文中采用多层感知器(Multi Layer Perceptron, MLP)神经网络作为主要的预测载体,将光照强度、温度、风速数据作为MLP的输入,光伏系统的输出功率作为MLP的输出,采用光伏电站的历史数据对MLP进行训练,并针对MLP在初始化权重和偏置量中存在的随机性问题,提出运用改进灰狼算法(Grey Wolf Optimizer, GWO)对MLP的初始权重和偏置量进行优化,减小MLP随机初始化的误差。仿真结果显示,文中提出的GWO-MLP在均方误差(Mean Square Error, MSE)、均方根误差(Root Mean Square Error, RMSE)、平均绝对误差(Mean Absolute Error, MAE)方面较MLP、Elman神经网络、支持向量机(Support Vector Machine, SVM)、极限学习机(Extreme Learning Machine, ELM)都有明显提高,表明所提方法可以准确预测光伏系统的输出功率。     

多裂缝有水气藏动态预测方法探讨

人工神经网络计算是一门崭新的建立在神经生理学和认知科学研究成果基础之上的信息处理学科,它采用数学的方法来研究并行分布的、非程序的、大脑风格的信息处理的本质和能力。利用人工神经网络中三层感知器及BP学习算法,根据日常生产资料,对多裂缝有水气藏的动态预测进行了探讨和偿试。有验预测结果表明,历史拟合和预测精度都非常高,有很好的应用前景。     

An MLP-based texture segmentation method without selecting a feature set

A texture segmentation technique which employs a multilayer perceptron (MLP) and does not consider the selection of features is presented in this paper. Thus, users can avoid selection and computation of the feature set and hence real-time segmentation may be possible. The technique apparently works in a fashion similar to our visual system whereby we do not consciously compute any feature for texture discrimination. A detailed study has been made for the selection of the network size. A newly proposed variant of the back-propagation algorithm has been used for more efficient training of the network. An edge-preserving noise-smoothing approach has been proposed to remove noise from the segmented image.     

关于单体模糊神经网络感知机收敛定理的讨论

考虑单体模糊神经网络模型,讨论了单体模糊神经网络的感知机收敛定理与传统神经网络感知机收敛定理的关系,指出了在证明单体模糊神经网络的感知机收敛定理时存在的问题,提出了两种修改方案,即给出了两个改进的感知机收敛定理,并给出相应的证明和说明.     

多层感知器的局部化设计

结合前馈网络的全局化设计和局部化设计的各自特点,提出一种由多个既独立又关 联的同构子网络构成的局部化多层感知器.仿真结果表明,该网络对复杂非线性系统具有良 好的学习性能,并适用于学习控制的直接逆模型法和远程学习法.     

微带不连续性的神经网络模型

本文采用多层感知器建立了微带不连续性的神经网络模型。