训练样本

采用了和学习向量量化神经网络相同的训练样本，构建了三层的BP神经网络。

Use the same training samples and build a three-layer BP neural network .

本文使用200幅交通标志作为训练样本对BP神经网络进行训练。

200 traffic signs are used as training samples to train the BP neural network .

将计算结果作为训练样本，对BP神经网络进行训练，从而实现BP网络对Moses时域计算的仿真。

The results as the training sample are used to train the BP neural network .

它通过对程序的运行轨迹进行编码，得到编码后的数据作为BP神经网络的训练样本。

Through processing the program running traces , raw data are converted as the training samples of BP neural network .

基于Gabor变换的每类单个训练样本人脸识别研究

Face recognition from single sample per class based on Gabor filtering

针对纹理图像分割中的训练样本自动获取问题，提出了一种基于模糊C均值算法的支持向量机半监督图像分割方法。

A support vector machine half-supervised machine learning method based on fuzzy C-mean algorithm objective to feature vector automatic acquirement was proposed and used in texture image segmentation .

通过判别分析方法和BP神经网络对训练样本和检验样本的分析，回判准确率较好，纳税评估指标具有可操作性。

Through the approaches of discriminant analysis and BP neural network , the veracious ratio of the training and testing samples are preferable .

深入研究了在高维多光谱数据分类中，SVM的性能与核函数类型、核函数参数、支持向量（SupportVector&SV）、训练样本数目、数据维数等之间的关系。

The relation between the performance of SVM and kernel function , support vector , training set , data dimension and so on is studied .

简要分析RBF网络的结构特点及最近邻聚类学习算法，以大量粉土地基实测数据为学习训练样本及预测样本，建立了预测模型。

The structure features of radial basis function and the nearest neighbor-clustering algorithm have been discussed in detail .

最后，用人脸训练样本在这些人脸局部特征子空间上的投影得到的权值向量训练支撑向量机SVM（supportVectorMachine）获得最终的人脸检测分类器。

At last , we trained the SVM ( Support Vector Machine ) with the power value vectors which the training examples project on the face features subspace .

提出了一种K-最近邻改进算法，该算法用模糊自适应共振理论（FuzzyART）对K-最近邻的训练样本集进行浓缩，以改善K-最近邻的计算速度。

This paper implemented an improved K-Nearest Neighbor ( Fuzzy KNN ) algorithm in which Fuzzy Adaptive Resonance Theory ( ART ) is applied in K-NN classification to make a new algorithm .

采用AAM对初始训练样本建模，在此基础上构造SVM分类器。

The original training examples were modeled by using AAM before the SVM classifier learning process .

该方法遵循网络的隐节点数与训练样本数相匹配的网络结构设计的最简原则构建BP网络；

The method follows the minimum principle of the structural design of the neural network , in which the training sample numbers match the hidden node numbers .

在大训练样本情况下，用传统的方法求解SVM问题计算复杂，针对该问题探讨了一系列的SVM训练算法，并对其进行了比较。

Under large samples , it is considerable complex to solve SVM questions by traditional methods . A series of training algorithms are discussed and compared .

在训练样本很大时，选择利用RLS算法来训练网络。

When the training sample is very large , RLS algorithm is used to train the networks .

对于如何用BP神经网络解界面反问题，给出了其基本步骤，并根据上述训练样本集容量的概念及界面反问题的特殊性，给出了组织界面反问题训练样本集的方法。

The principle and method used in forming a training sample set were presented . Secondly , the basic procedure solving the inversion of interfaces using BP neural networks was given .

为了提高样本的多样性，采用动态预测的方法，即将样本反馈、增加到训练样本中，不断重新训练BP神经网络。

In order to improve the diversity of the sample , the dynamic prediction method , taking sample feedback , increasing training samples , and re-training of BP neural network constantly .

此部分使用从发动机采集到的信号做为训练样本，然后运用BP网络算法，对网络权值进行正向和反向的不断训练。

This section uses the collected signal from the engine as training samples , and then use the BP Network algorithm constantly to train network weights from the forward and reverse .

以训练样本识别正确率为适应度函数，采用GA对初始特征进行组合优化。

Then , a GA is used to combine and optimize initial features . The fitness function of the GA is the recognition rate of training samples .

在此基础上，结合模糊C均值（fuzzyC-Means，FCM）非监督聚类算法，提出了一种自动标注训练样本的方法。

On this basis , a kind of automatic selection methods of training samples was presented by using the Fuzzy c-Means ( FCM ) which is unsupervised clustering algorithm .

由于算法对单个类别的多训练样本进行特征提取，因此更为有效的描述了样本的特征空间，同时结合SVM处理高维数据分割的强大能力，取得了很好的识别效果。

Because SVM has the strong ability of segmenting high dimension data and PCA can distill the eigenspace effectively , the result of the experiment is satisfied .

以保定市各县供电企业专家评价数据作为BP神经网络的训练样本，进行仿真试验，得到了满意的结果。

Simulation tests were carried out with evaluation data given by experts from power supply enterprises in Baoding taken as training samples for BP neural network , and the results are good .

RBF神经网络训练样本的选取则由均匀试验设计确定，以提高样本的代表性并大幅减少样本数量，从而加快网络的训练过程，加强网络的逼近能力。

The training samples of RBF Neural Network are determined by Uniform Experiment Design to accelerate training process and enhance approximate capacity of the neural network .

通过引入两个调节变量，改进后的SVM可根据训练样本与预测样本之间的时间跨度，对不同样本的训练误差采取不同的惩罚力度。

By inducting two variables , empirical error of the training sample is treated differently based on the time span between the forecasting sample and the training sample .

利用实验室测量数据组成训练样本，分别采用了BP（误差反向传播）网络和RBF（径向基函数）网络两种神经网络建立软测量模型。

Taking laboratory measurement data for the training sample , a BP ( back-propagation ) network and RBF ( radial basis function ) network is separately used for soft-measurement modeling .

用NB算法在新训练样本中进行学习，获得分类器，从而将NB与判别分析方法有机地结合起来。

This classifier is trained on new samples with NB algorithm . In this way , Discriminant Analysis and NB algorithm is combined organically together .

研究了高维多光谱图像分类中SVM的分类性能与训练样本数目和数据维数之间的关系。

So the classification accuracy of SVM is better than traditional ones . The relationship between the classification performance of SVM and training set , data dimension is studied .

提出了基于BP网络的无刷直流电机相角控制技术，将实验数据作为训练样本进行离线训练，网络收敛后用作在线控制。

Thus the BP neural network based phase control method is presented , in which experimental data is applied to train BP neural network in off-line way and afterwards converged network to control on-line .

对采集到的焊缝偏差信息数据集进行主成分分析，映射到低维的PCA空间，作为关联向量机的训练样本集；

The data set is projected into low dimensional PCA space using PCA , and the low dimensional data set used as training data set of RVM .

通过实验获得进化神经网络的训练样本，在RBF神经网络的训练中，提出了基于Elitist竞争机制的遗传进化训练方法。

The training samples were obtained by experiments and the genetic evolutionary method based on elitist rule was presented for neural network training .