一种基于权重选择虚拟样本的新分类算法

虚拟样本是一种在学习过程中引入先验知识的有效手段,一定程度上提高了分类器的性能。然而由于生成的虚拟样本集的数据分布与原始训练集的分布不一致,因此利用虚拟样本扩充后的训练样本集进行学习的分类器分类性能上存在波动。针对这种不足,提出一种基于权重选择虚拟样本的新分类算法。该方法首先利用TrAdaBoost算法对扩充后的样本集进行预处理,然后选取权重大于某一给定阈值的样本构造新训练样本集,最后根据新样本集进行训练得到分类器。由于排出了不重要的样本,因此在新样本集上得到的分类器具有更高的精度。在部分UCI标准数据集与KDD cup 99网络入侵检测数据集上的对比实验说明了该算法较不产生虚拟样本的直接分类算法和利用虚拟样本全集进行训练的分类算法具有更高的精度。     

基于改进卷积神经网络的多源数字识别算法

现有的数字识别算法多是对单一类型数字进行识别,无法应对识别多源数字。针对包含手写体数字与数码管数字的字符识别场景,提出一种基于改进卷积神经网络（CNN）的多源数字识别算法。首先,使用从数显仪表生产企业现场采集的样本,结合MINIST数据集,建立起包含手写体和数码管的混合数据集;然后,考虑更好的鲁棒性,提出一种改进的CNN,并用上述混合数据集对其训练,实现了一个网络识别多类型数字;最后,训练好的神经网络模型被成功应用于RoboMaster机甲大赛的多源数字识别场景中。测试结果表明,所提算法整体识别准确率稳定且较高,具有较好的鲁棒性和泛化能力。     

数据增强策略在人脸识别技术实现的研究

以小数据集为样本进行卷积神经网络模型的训练过程,容易出现所得到的神经网络模型泛化能力不足的问题。传统的处理方法大都通过数据增强的方式来提高训练数据的样本总数。本文选取多个网络模型进行对比实验,验证不同神经网络在训练过程中是否使用数据随机增强方式的模型识别准确率提高的效果,为如何选取小数据集样本训练神经网络提供参考。     

基于结构误差的图卷积网络

针对图卷积网络选取交叉熵作为损失函数在小样本数据集上可能导致模型训练过拟合、模型泛化能力不强等问题,提出了基于结构误差的图卷积网络。将改进的基于结构误差的支持向量机(support vector machine, SVM)作为图卷积网络的分类器,能够降低模型过拟合的风险。在SVM的泛化误差理论基础上,对SVM的损失函数进行改进,所提出的方法在最大化异类样本间隔的同时限制同类样本间的间隔,提升了模型的泛化能力。首先计算特征空间中特征向量到中心点的平均距离,由它近似替换最小包含球的半径,然后由新的损失函数指导模型的学习。在基于骨骼的行为识别领域的NTU RGB+D60和NTU RGB+D120数据集上的实验证明,相比于传统的图卷积网络模型,所提出的方法能够明显提升识别准确率并且具有更好的泛化性能。     

多神经网络在高维数据分类中的应用研究

针对BP神经网络在高维数据分类中存在训练时间长的缺点,提出一种新的多神经网络分类模型,该模型采用自组织特征映射(SOFM)网络对训练样本集进行无监督聚类,通过优化竞争层神经元权值,并以此训练BP神经网络实现数据分类.最后对自由手写数字样本进行识别,仿真实验表明,这一模型具有较强的分类能力和泛化能力.     

边缘智能背景下的手写数字识别

随着边缘智能的快速发展,现有手写数字识别卷积网络模型的发展已越来越不适应边缘部署、算力下降的要求,且存在小样本泛化能力较差和网络训练成本较高等问题。借鉴卷积神经网络（CNN）经典结构、Leaky_ReLU算法、dropout算法和遗传算法及自适应和混合池化思想构建了基于LeNet-DL改进网络的手写数字识别模型,分别在大样本数据集MNIST和小样本真实数据集REAL上与LeNet、LeNet+sigmoid、AlexNet等算法进行对比实验。改进网络的大样本识别精度可达99.34%,性能提升约0.83%;小样本识别精度可达78.89%,性能提升约8.34%。实验结果表明,LeNet-DL网络相较于传统CNN在大样本和小样本数据集上的训练成本更低、性能更优且模型泛化能力更强。     

基于Boosting算法的入侵检测

提出了一种新颖的基于boosting BP 神经网络的入侵检测方法。为了提高BP神经网络的泛化能力,采用改进的Boosting方法,进行网络集成。Boosting方法采用更有效的参数求解方法,即弱分类器的加权参数不但与错误率有关,还与其对正样本的识别能力有关。对“KDD Cup 1999 Data”网络连接数据集进行特征选择和归一化处理之后用于训练神经网络并仿真实验,得到了较高的检测率和较低的误报率,仿真结果表明,提出的入侵检测方法是有效的。     

数据失衡下基于WGAN和GAPCNN的轴承故障诊断研究

针对轴承故障数据严重失衡导致所训练的模型诊断能力和泛化能力较差等问题,提出基于Wasserstein距离的生成对抗网络来平衡数据集的方法。该方法首先将少量故障样本进行对抗训练,待网络达到纳什均衡时,再将生成的故障样本添加到原始少量故障样本中起到平衡数据集的作用;提出基于全局平均池化卷积神经网络的诊断模型,将平衡后的数据集输入到诊断模型中进行训练,通过模型自适应地逐层提取特征,实现故障的精确分类诊断。实验结果表明,所提诊断方法优于其他算法和模型,同时拥有较强的泛化能力和鲁棒性。     

一种复杂模式识别问题的求解方法

针对复杂的模式识别问题,提出了一种用串－并行混合结构的多种神经网络模型进行求解的方法。首先用ART网络对训练集中的样本进行粗分类,以减小训练集的样本规模,然后用多个BP网络并行地对小训练集进行训练。用于汉字识别问题,取得了很好的效果。     

基于Bayes法则和BP神经网络的高速动态情形下车型识别

针对在高速动态情形下的车型识别,介绍了一种对汽车提取特征、基于红外线检测的汽车分类仪;阐述了采用汽车特征参数作为样本向量训练BP网络的方法和识别车型原理;采用共轭梯法修正BP网络,提高了训练速度和全局收敛性;对于样本向量存在的数据“噪声”,则以Bayes法则对大量样本去除“噪声”,使特征样本向量更有代表性,理论与实际证明,这样得到BP网有强容错能力,能识别没有看过的汽车样本,从而提高了车型识别精度。     

An optimizing BP neural network algorithm based on genetic algorithm

A back-propagation (BP) neural network has good self-learning, self-adapting and generalization ability, but it may easily get stuck in a local minimum, and has a poor rate of convergence. Therefore, a method to optimize a BP algorithm based on a genetic algorithm (GA) is proposed to speed the training of BP, and to overcome BP’s disadvantage of being easily stuck in a local minimum. The UCI data set is used here for experimental analysis and the experimental result shows that, compared with the BP algorithm and a method that only uses GA to learn the connection weights, our method that combines GA and BP to train the neural network works better; is less easily stuck in a local minimum; the trained network has a better generalization ability; and it has a good stabilization performance.     

An exploration of the uncertainty relation satisfied by BP network learning ability and generalization ability

An overfit phenomenon exists in the BP network. The so-called overfit means that as long as the network is allowed to be sufficiently complicated, the BP network can minimize the error of the training sample set; however, in the case of a limited number of samples, the generalization ability of the network will decrease. This indicates that there is a relation between the learning ability and the generalization ability. Therefore, studying the relationship between the learning ability is the…     

基于改进BP网络8/6极SRM非线性建模的研究

针对开关磁阻电机(SRM)磁化曲线高度饱和、非线性的特点,提出一种基于改进的BP神经网络建立开关磁阻电机模型的方法。该方法构造了一个将连接权值变为参数可调函数的BP神经网络。通过分析开关磁阻电机磁链与转矩特性获得神经网络的训练样本,经过训练,实现开关磁阻电机非线性建模,并在Matlab/Simulink中建立开关磁阻电机控制系统(SRD)仿真模型。仿真与实验结果的对比,证明了此建模方法可行。与传统BP神经网络建模相比,该方法节约了计算时间,具有很强的泛化能力和较高精度,有效地提高了收敛速度。     

A novel virtual sample generation method based on Gaussian distribution

Traditional machine learning algorithms are not with satisfying generalization ability on noisy, imbalanced, and small sample training set. In this work, a novel virtual sample generation (VSG) method based on Gaussian distribution is proposed. Firstly, the method determines the mean and the standard error of Gaussian distribution. Then, virtual samples can be generated by such Gaussian distribution. Finally, a new training set is constructed by adding the virtual samples to the original training set. This work has shown that training on the new training set is equivalent to a form of regularization regarding small sample problems, or cost-sensitive learning regarding imbalanced sample problems. Experiments show that given a suitable number of virtual sample replicates, the generalization ability of the classifiers on the new training sets can be better than that on the original training sets.     

BP网络进化及其在雷达目标识别中的应用

针对常规ＢＰ神经网络的ＢＰ算法只能训练固定结构的神经网络,存在诸如易落入局部极值、没有引入提高泛化能力的训练机制等固有不足之处,以及一些神经网络进化算法的进化机制中存在的缺陷,本文提出一种ＢＰ神经网络进化算法,并用于高分辨雷达目标一维距离像的识别问题。实验结果表明,经所述方法优化后的神经网络结构简单、泛化能力优于ＢＰ算法和一些进化算法训练的网络。     

电动车用MH-Ni电池剩余电量的预测研究

蓄电池剩余容量为电动汽车可持续行驶提供有力的判据，因此，对它的准确估计具有重要意义。该文利用BP神经网络对剩余电量进行预测，采用试错法确定隐层节点数，尺度化共轭梯度反向传播算法对网络进行训练。通过对仿真结果的分析，发现网络出现了过拟合现象；通过引入BP网络学习能力和泛化能力的不确定关系，尝试建立新的网络；并对误差进行比较，取得最佳的泛化能力；进而对网络进行验证，证明新的网络是成功的。最后对进一步的研究提供思路。     

NeC4.5: neural ensemble based C4.5

Decision tree is with good comprehensibility while neural network ensemble is with strong generalization ability. These merits are integrated into a novel decision tree algorithm NeC4.5. This algorithm trains a neural network ensemble at first. Then, the trained ensemble is employed to generate a new training set through replacing the desired class labels of the original training examples with those output from the trained ensemble. Some extra training examples are also generated from the trained ensemble and added to the new training set. Finally, a C4.5 decision tree is grown from the new training set. Since its learning results are decision trees, the comprehensibility of NeC4.5 is better than that of neural network ensemble. Moreover, experiments show that the generalization ability of NeC4.5 decision trees can be better than that of C4.5 decision trees.     

Improving the Performance of Feedforward Neural Networks by Noise Injection into Hidden Neurons

The generalization ability of feedforward neural networks (NNs) depends on the size of training set and the feature of the training patterns. Theoretically the best classification property is obtained if all possible patterns are used to train the network, which is practically impossible. In this paper a new noise injection technique is proposed, that is noise injection into the hidden neurons at the summation level. Assuming that the test patterns are drawn from the same population used to generate the training set, we show that noise injection into hidden neurons is equivalent to training with noisy input patterns (i.e., larger training set). The simulation results indicate that the networks trained with the proposed technique and the networks trained with noisy input patterns have almost the same generalization and fault tolerance abilities. The learning time required by the proposed method is considerably less than that required by the training with noisy input patterns, and it is almost the same as that required by the standard backpropagation using normal input patterns.