基于CapsNet的汉字字形表征模型

提出基于胶囊神经网络(CapsNet)的汉字字形表征模型,通过表征汉字字形中的部件实现汉字字形的表征.首先,对任一汉字字形生成所有部件类别的表征向量.然后,根据部件存在概率,利用基于欧氏距离的离群点检测,选取相应的部件表征向量.最后,由选出的部件表征向量组成该汉字的字形表征.实验表明,文中模型在仅经过部件字形训练的情况下,即可有效识别汉字部件,同时自动生成汉字字形的有效表征.     

多尺度混合注意力胶囊网络的海洋鱼类识别

针对胶囊网络(capsule network,CapsNet)特 征提取结构单一和数据处理中参数量过大的问题,提出 多尺度混合注意力胶囊网络 模型。首先,在网络初始端添加不同尺度的卷积核来多角度提取 特征,并引 入混合注意力机制,通过聚焦更具分辨性的特征区域来降低复杂背景干扰。其次,采用局部 剪枝算法优 化动态路由,减少参数量,缩短模型训练时间。最后,在海洋鱼类数据集F4K(Fish4Knowled ge)上验证, 结果表明,与传统残差网络(residual network50,ResNet-50)、双线性网络(bilinear convo l utional neural network,B-CNN)、分层精简双线性注意力网络(spatial transformation netw ork and hierarchical compact bilinear pooling,STN-H-CBP)以及CapsNet模型相比,该算法 识别精度为98.65%,比ResNet-50模型提升 了5.92%;训练时间为2.2 h,相比于CapsNet 缩短了近40 min,验证了该算法的可行性。     

结合SoftPool的VGG19与CapsNet相级联的表情识别模型研究

表情识别作为计算机视觉领域的研究热点,在情感识别、人机交互、智能安防等领域有着重要的应用.文章针对VGG19在训练人脸表情数据集时由于全连接层参数量过大而易过拟合的问题,利用胶囊网络CapsNet对VGG19的全连接层进行替换,以实现VGG19与CapsNet相级联,从而改善训练时过拟合的问题,同时使得级联后的模型在RAF-DB数据集上的精度提高了5.28％.针对VGG19特征提取网络的MaxPool易丢失人脸特征图信息的问题,利用SoftPool对MaxPool进行替换,从而在最大程度上保留了人脸的细粒度特征.实验结果表明,改进后的模型在RAF-DB数据集上取得了 84.21％的精度,在FER2013数据集上取得了 73.16％的精度,表情识别效果更优.     

梯度直方图卷积特征的胶囊网络在交通监控下的车型分类

为了充分利用图像信息以提高现有交通监控下车型分类的效果,在胶囊网络的基础上增加梯度直方图卷积（HOG-C）特征提取方法,提出HOG-C特征的胶囊网络模型——HOG-C CapsNet。首先,使用梯度统计特征提取层对图像中的梯度信息进行统计,构建方向梯度直方图（HOG）特征图;其次,使用卷积层提取出图像的颜色信息,把提取出的颜色信息与HOG特征图融合构成HOG-C特征图;最后,输入卷积层提取HOG-C特征图的抽象特征,并通过胶囊网络对提取的抽象特征进行具有三维空间特征表达的胶囊封装,使用动态路由算法实现车型分类。在BIT-Vehicle数据集上对该模型和其他相关模型进行的对比实验中,该模型得到98.17%的准确率、97.98%的平均精确率均值（MAP）、98.42%的平均召回率均值（MAR）和98.20%的综合评价指标。实验结果表明,该模型在交通监控下的车型分类上具有更好的效果。     

多分支结构强化表征能力的CapsNet方法

CapsNet是一种新的目标识别模型,通过动态路由和capsule识别已知目标的新形态.针对CapsNet的解码器输入层规模随类别数增加而增加,可延展性较弱的问题,本文提出多分支自编码器模型.该模型将各个类别的编码分别传递给解码器,使解码器规模独立于类别数,增强了模型的可延展性.针对单类别图像训练多类别图像识别任务,本文增加新的优化目标降低非标签类别的编码向量对解码器的激励,强化了模型的表征能力.MNIST数据集的实验结果表明,多分支自编码器具有良好的识别能力且重构能力明显优于CapsNet,因而具有更全面的表征能力.     

Cervical image classification based on image segmentation preprocessing and a CapsNet network model

Cervical cancer is one of the most common gynecological malignancies, and when detected and treated at an early stage, the cure rate is almost 100%. Colposcopy can be used to diagnose cervical lesions by direct observation of the surface of the cervix using microscopic biopsy and pathological examination, which can improve the diagnosis rate and ensure that patients receive fast and effective treatment. Digital colposcopy and automatic image analysis can reduce the work burden on doctors, improve work efficiency, and help healthcare institutions to make better treatment decisions in underdeveloped areas. The present study used a deep-learning model to classify the images of cervical lesions. Clinicians could determine patient treatment based on the type of cervix, which greatly improved the diagnostic efficiency and accuracy. The present study was divided into two parts. First, convolutional neural networks were used to segment the lesions in the cervical images; and second, a neural network model similar to CapsNet was used to identify and classify the cervical images. Finally, the training set accuracy of our model was 99%, the test set accuracy was 80.1%, it obtained better results than other classification methods, and it realized rapid classification and prediction of mass image data.     

基于特征扩展CapsNet的轴承故障诊断模型

基于深度网络的轴承故障诊断模型,其深层次的特征提取往往需要大量的训练样本,诊断模型应用于小样本故障信号时受到限制。利用胶囊网络(CapsNet)在处理小样本数据上的优势,通过扩展故障特征向量和改进预测胶囊迭代约束条件,使网络适用于噪声工况下的轴承故障诊断,提高预测精度。在CapsNet的初级胶囊层引入特征图跨通道约束关系,由原来的通道内约束改进为通道内-通道间混合约束,扩展胶囊特征向量,使其能描述故障信号时间序列中与远距离序列点相关的几何特性,为数字胶囊层预测网络提供更完备的故障特征。同时,利用余弦相似度作为特征向量的度量并为数字胶囊层的迭代筛选提供依据,避免向量模值造成的分类误差。凯斯西储大学(CWRU)轴承数据集实验结果表明,改进的iCapsNet诊断模型泛化性能得到明显提升,在信噪比为0 dB时,预测精度可达到90.9%,相比原CapsNet模型提高了44.8%。     

基于随机域名检测和主动防御的用户站安全防护

电力监控系统是电力行业最重要的生产管理系统. 作为电力监控系统的重要组成部分, 缺少电网约束力的用户站将会成为网络攻击的重要目标. 为及时感知用户站侧网络攻击事件, 提出了一种结合用户站侧随机域名实时检测和主动防御的方法. 使用胶囊网络(CapsNet)结合长短期记忆网络(LSTM)对流量数据中提取的域名进行二分类, 当检测到随机域名时, 通过远程终端协议(Telnet)对路由器和交换机下发指令更新其安全策略或关闭路由器和交换机的业务接口以阻断网络攻击. 实验结果表明, 使用CapsNet结合LSTM分类算法在随机域名检测中准确率达到99.16%, 召回率达到98%, 通过Telnet协议可以联动路由器和交换机在不中断业务的情况下做出主动防御.     

Emotion Recognition with Capsule Neural Network

For human-machine communication to be as effective as human-to-human communication, research on speech emotion recognition is essential. Among the models and the classifiers used to recognize emotions, neural networks appear to be promising due to the network’s ability to learn and the diversity in configuration. Following the convolutional neural network, a capsule neural network (CapsNet) with inputs and outputs that are not scalar quantities but vectors allows the network to determine the part-whole relationships that are specific 6 for an object. This paper performs speech emotion recognition based on CapsNet. The corpora for speech emotion recognition have been augmented by adding white noise and changing voices. The feature parameters of the recognition system input are mel spectrum images along with the characteristics of the sound source, vocal tract and prosody. For the German emotional corpus EMO-DB, the average accuracy score for 4 emotions, neutral, boredom, anger and happiness, is 99.69%. For Vietnamese emotional corpus BKEmo, this score is 94.23% for 4 emotions, neutral, sadness, anger and happiness. The accuracy score is highest when combining all the above feature parameters, and this score increases significantly when combining mel spectrum images with the features directly related to the fundamental frequency.     

基于粒子群优化以及深度胶囊网络的轴承故障诊断

针对发动机轴承损坏情况严重以及基于模型方法预测精度不稳定的问题,提出一种基于深度胶囊网络和粒子群优化算法的轴承故障预测方法。通过将观测振动信号自适应降噪后,基于粒子群优化算法进行稀疏盲分离,得到轴承振动信号,通过S变换获取时域图以及轴承振动特征,其次将时域图经由卷积层卷积,输入到胶囊层进行预测。将高低胶囊层之间的算法转化为数学优化问题,提升传输效率,最后得出高层胶囊的预测向量。结合具体轴承监测数据进行实例分析,与基于数据的浅卷积网络以及经验模态分解预测相比,算法体现了更稳定更精确的预测性能。