一种卷积神经网络和极限学习机相结合的人脸识别方法

卷积神经网络是一种很好的特征提取器,但却不是最佳的分类器,而极限学习机能够很好地进行分类,却不能学习复杂的特征,根据这两者的优点和缺点,将它们结合起来,提出一种新的人脸识别方法。卷积神经网络提取人脸特征,极限学习机根据这些特征进行识别。本文还提出固定卷积神经网络的部分卷积核以减少训练参 数,从而提高识别精度的方法。在人脸库ORL和XM2VTS上进行测试的结果表明,本文的结合方法能有效提高人脸识别的识别率,而且固定部分卷积核的方式在训练样本少时具有优势。     

基于深度学习网络模型的端到端航迹关联

为提高雷达数据处理中航迹关联的智能性,充分利用目标的特征信息,并简化系统处理流程,提出了一种基于深度学习网络模型的端到端航迹关联算法。首先分析了基于神经网络的航迹关联存在样本细节少、处理流程繁杂的问题,然后提出了端到端的深度学习模型。该模型根据航迹关联数据的处理特征,改进了卷积神经网络结构用于特征提取,充分利用了长短期记忆网络对历史信息和将来信息的处理能力,并分析了前后航迹的关联性。在对原始数据进行卡尔曼滤波后,将全部航迹信息特征作为输入,并由基于卷积神经网络特征提取的长短期记忆深度神经网络模型直接输出航迹关联结果。仿真结果表明,提出的模型可以充分学习推演目标的多个特征信息,具有较高的航迹关联准确率,对航迹关联的智能化分析具有一定的参考价值。     

面向人体行为识别的深度特征学习方法比较*

针对人体行为识别问题,比较了两种基于智能手机惯性加速度传感器数据的深度特征学习方法。与传统的人工特征提取方法相比,基于深度特征学习方法可以实现端到端训练,网络结构简单直观,避免了繁琐的特征工程,通过深度神经网络模型的学习自动获得特征。本文通过对比深度卷积神经网络、长短期记忆网络两种深度学习方法在公开网站UCI的机器学习知识库的人体行为识别数据集上的识别效果,论证了基于Dropout深度卷积神经网络特征学习方法的有效性。     

基于时域建模的自动语音识别

端到端神经网络能够根据特定的任务自动学习从原始数据到特征的变换,解决人工设计的特征与任务不匹配的问题。以往语音识别的端到端网络采用一层时域卷积网络作为特征提取模型,递归神经网络和全连接前馈深度神经网络作为声学模型的方式,在效果和效率两个方面具有一定的局限性。从特征提取模块的效果以及声学模型的训练效率角度,提出多时间频率分辨率卷积网络与带记忆模块的前馈神经网络相结合的端到端语音识别模型。实验结果表明,所提方法语音识别在真实录制数据集上较传统方法字错误率下降10%,训练时间减少80%。     

基于粗糙集-神经网络集成的故障诊断

综合粗糙集和神经网络的优点,提出一种基于粗糙集-神经网络集成的智能故障诊断模型.在数据采集和预处理的基础上,利用粗糙集(RS)理论对原始故障诊断样本进行离散化处理,并根据条件属性(集)对决策属性的正域的大小来选择属性,提取出对诊断故障贡献最大的最小故障特征子集,从而确定神经网络的拓扑结构;通过网络训练建立故障特征与故障之间的映射关系,采用神经网络集成的方法实现故障的诊断.通过热电厂发电机组的故障诊断实例,表明了这种故障诊断方法的工程有效性.     

AKSNet: A novel convolutional neural network with adaptive kernel width and sparse regularization for machinery fault diagnosis

Convolutional kernels have significant affections on feature learning of convolutional neural network (CNN). However, it is still a challenging problem to determine appropriate kernel width. Moreover, some features learned by convolutional layers are still redundant and noisy. Thus, adaptive selection of kernel width and feature selection of feature maps are key techniques to improve feature learning performance of CNNs. In this paper, a new deep neural network (DNN) model, adaptive kernel sparse network (AKSNet) is proposed to extract multi-scale fault features from one-dimensional (1-D) vibration signals. Firstly, an adaptive kernel selection method is developed, where multiple branches with different kernels are used to extract multi-scale features from vibration signals. Channel-wise attention is developed to fuse features generated by these kernels to obtain different informative scales. Secondly, a spatial attention is used for dynamic receptive field to focus on salient region of feature maps. Thirdly, a sparse regularization layer is embedded in the deep network to further filter noise and highlight impaction of the feature maps. Finally, two cases are adopted to verify effectiveness of AKSNet-based feature learning for bearing fault diagnosis. Experimental results show that AKSNet can effectively extract features from multi-channel vibration signals and then improves fault diagnosis performance of the classifier significantly. AKSNet shows better recognition performance in comparison with that of shallow neural networks and other typical DNNs.     

基于一维卷积神经网络的实时道岔故障诊断

针对道岔故障诊断系统实时性要求高、特征提取严重依赖于先验知识的问题,提出了一种基于一维卷积神经网络（1D-CNN）的道岔实时故障诊断方法。以S700k转辙机的功率曲线为例,建立一维卷积神经网络的结构模型,该模型将特征提取与故障分类融合为一体,优化了网络参数,同时使用正则化Dropout提高模型的泛化能力,采用t-SNE可视化方法,来反映模型提取特征的有效性。仿真实验表明：卷积层和池化层对原始时域信号的自适应特征提取,能较好地捕捉信号空间维度信息,降低模型的计算量,提高模型的抗噪性能,实现了端到端的实时故障诊断,并有效地提高道岔故障实时诊断的准确率。     

One-dimensional convolutional auto-encoder-based feature learning for fault diagnosis of multivariate processes

Noise and high-dimension of process signals decrease effectiveness of those regular fault detection and diagnosis models in multivariate processes. Deep learning technique shows very excellent performance in high-level feature learning from image and visual data. However, the large labeled data are required for deep neural networks (DNNs) with supervised learning like convolutional neural network (CNN), which increases the time cost of model construction significantly. A new DNN model, one-dimensional convolutional auto-encoder (1D-CAE) is proposed for fault detection and diagnosis of multivariate processes in this paper. 1D-CAE is utilized to learn hierarchical feature representations through noise reduction of high-dimensional process signals. Auto-encoder integrated with convolutional kernels and pooling units allows feature extraction to be particularly effective, which is of great importance for fault detection and diagnosis in multivariate processes. The comparison between 1D-CAE and other typical DNNs illustrates effectiveness of 1D-CAE for fault detection and diagnosis on Tennessee Eastman Process and Fed-batch fermentation penicillin process. The proposed method provides an effective platform for deep-learning-based process fault detection and diagnosis of multivariate processes.     

采用Transformer网络的视频序列表情识别

目的 相比于静态人脸表情图像识别,视频序列中的各帧人脸表情强度差异较大,并且含有中性表情的帧数较多,然而现有模型无法为视频序列中每帧图像分配合适的权重。为了充分利用视频序列中的时空维度信息和不同帧图像对视频表情识别的作用力差异特点,本文提出一种基于Transformer的视频序列表情识别方法。方法 首先,将一个视频序列分成含有固定帧数的短视频片段,并采用深度残差网络对视频片段中的每帧图像学习出高层次的人脸表情特征,从而生成一个固定维度的视频片段空间特征。然后,通过设计合适的长短时记忆网络（long short-term memory network,LSTM）和Transformer模型分别从该视频片段空间特征序列中进一步学习出高层次的时间维度特征和注意力特征,并进行级联输入到全连接层,从而输出该视频片段的表情分类分数值。最后,将一个视频所有片段的表情分类分数值进行最大池化,实现该视频的最终表情分类任务。结果 在公开的BAUM-1s （Bahcesehir University multimodal）和RML （Ryerson Multimedia Lab）视频情感数据集上的试验结果表明,该方法分别取得了60.72%和75.44%的正确识别率,优于其他对比方法的性能。结论 该方法采用端到端的学习方式,能够有效提升视频序列表情识别性能。     

基于卷积神经网络的机械故障诊断技术综述

针对传统机械故障诊断方法难以解决人工提取不确定性的问题,提出了大量深度学习的特征提取方法,极大地推动了机械故障诊断的发展。作为深度学习的典型代表,卷积神经网络（CNN）在图像分类、目标检测、图像语义分割等领域都取得了重大的发展,在机械故障诊断领域也有大量文献发表。为了进一步了解利用CNN的方法进行机械故障诊断的问题,首先简单介绍了CNN的相关理论,然后从数据输入类型、迁移学习、预测等方面对CNN在机械故障诊断中的应用进行了归纳总结,最后展望了CNN及其在机械故障诊断应用中的发展方向。     

Extracting Rules from Composite Neural Networks for Medical Diagnostic Problems

Recently, neural networks have been applied to many medical diagnostic problems because of their appealing properties, robustness, capability of generalization and fault tolerance. Although the predictive accuracy of neural networks may be higher than that of traditional methods (e.g., statistical methods) or human experts, the lack of explanation from a trained neural network leads to the difficulty that users would hesitate to take the advise of a black box on faith alone. This paper presents a class of composite neural networks which are trained in such a way that the values of the network parameters can be utilized to generate If-Then rules on the basis of preselected meaningful coordinates. The concepts and methods presented in the paper are illustrated through one practical example from medical diagnosis.     

代码特征自动提取方法

神经网络在软件工程中的应用极大程度上缓解了传统的人工提取代码特征的压力。已有的研究往往将代码简化为自然语言或者依赖专家的领域知识来提取代码特征,简化为自然语言的处理方法过于简单,容易造成信息丢失,而引入专家制定启发式规则的模型往往过于复杂,可拓展性以及普适性不强。鉴于以上问题,提出了一种基于卷积和循环神经网络的自动代码特征提取模型,该模型借助代码的抽象语法树(AST)来提取代码特征。为了缓解因AST过于庞大而带来的梯度消失问题,对AST进行切割,转换成一个AST序列再作为模型的输入。该模型利用卷积网络提取代码中的结构信息,利用双向循环神经网络提取代码中的序列信息。整个流程不需要专家的领域知识来指导模型的训练,只需要将标注类别的代码作为模型的输入就可以让模型自动地学习如何提取代码特征。应用训练好的分类编码器,在相似代码搜索任务上进行测试,Top1、NDCG、MRR的值分别能达到0.560、0.679和0.638,对比当下前沿的用于代码特征提取的深度学习模型以及业界常用的代码相似检测工具有显著的优势。     

一种云样本的控制产生及在电路故障诊断中的应用

为了选择电路故障诊断中的特征样本,提出了产生云样本的方法,并用于神经网络的训练和识别.首先采用逆向云理论对初始特征样本进行统计以获取数字特征,其次采用正向云理论产生扩展训练样本,并用新产生的样本训练两种神经网络.仿真结果表明,采用云样本训练的神经网络要比采用常规样本训练的性能稳健,具有较好的抗噪卢性能,在模拟电路故障诊断中达到了较好的诊断效果.     

基于模糊自组织映射神经网络的故障诊断方法

在研究Kohonen自组织映射网络理论的基础上运用模糊理论方法建立了刹车系统模糊故障诊断模型。该模型只需选择足够的具有代表性的故障样本训练神经网络，将代表故障的信息输入给训练好的神经网络，根据神经网络的输出结果，就可以判断发生故障的类型。该模型除能识别已训练过的故障，还能识别未训练过的故障，并且聚类能力强、速度快，因此很符合复杂系统的故障诊断。     

Learning sequential features for cascade outbreak prediction

Information cascades are ubiquitous in various online social networks. Outbreak of cascades could cause huge and unexpected effects. Therefore, predicting the outbreak of cascades at early stage is of vital importance to avoid potential bad effects and take relevant actions. Existing methods either adopt regression or classification technique with exhaustive feature engineering or predict cascade dynamics via modeling the stochastic process of cascades using a hard-coded diffusion–reaction function. One salient issue of these methods is that these methods heavily depend on human-defined knowledge, features or functions. In this paper, we propose to use recurrent neural network with long short-term memory to directly learn sequential patterns from information cascades, working in a fully data-driven manner. With the learned sequential patterns, the outbreak of cascade could be accurately predicted. Extensive experiments on both Twitter and Sina Weibo datasets demonstrate that our method significantly outperforms state-of-the-art methods at the prediction of cascade outbreaks.     

基于多分支协作的行人重识别网络

设计多分支网络以提取分集特征已成为行人重识别领域的重要方向之一.由于单分支学习到的特征表达能力有限,所以文中提出基于多分支协作的行人重识别网络.在局部分支、全局分支、全局对比池化分支、关联分支这4个相互协作的分支上进行特征提取,获得强大的分集行人特征表达能力.文中网络可应用于不同的主干网络.实验中考虑OSNet、ResNet作为文中网络的主干网络进行验证.实验表明,文中网络在行人重识别数据集上均取得Start-of-the-art结果.     

基于卷积神经网络的自适应权重multi-gram语句建模系统

如今信息量呈爆炸式增长,自然语言处理得到了越来越广泛的重视。传统的自然语言处理系统过多地依赖昂贵的人工标注特征和语言分析工具的语法信息,导致预处理中语法信息的错误传递到系统训练和预测过程中。因此,深度学习的应用受到了学者们的关注。因为它能实现端对端预测并尽可能少地 依赖 外部信息。自然语言处理领域流行的深度学习框架为了更好地获取句子信息,采用multi-gram策略。但不同任务和不同数据集的信息分布状况不尽相同,而且这种策略并没有考虑到不同n-gram的重要性分布。针对该问题,提出了一种基于深度学习的自适应学习multi-gram权重的策略,从而根据各n-gram特征的贡献为其分配相应的权重;并且还提出了一种新的multi-gram特征向量结合方法,大大降低了系统复杂度。将该模型应用到电影评论正负倾向判断和关系分类两种分类任务中,实验结果证明采用的自适应multi-gram权重策略能够大大改善模型的分类效果。     

基于模糊神经网络的复杂网络故障诊断设计

针对雷达等复杂大型电子装备网络系统的故障定位难、影响关系不清晰的问题,采用了基于模糊神经网络的故障定位方法,提高了网络故障定位的快速性与准确性：首先介绍了模糊隶属度及模糊神经元等理论,接着基于模糊理论将网络监测信息进行模糊化处理,并利用神经网络模型对模糊后的信息进行训练与学习,参数训练达到设置的期望误差0.01;最后利用训练好的模型对随机抽取的2组网络故障实例进行了验证,软件执行单次诊断耗时3.5s;结果表明采用基于模糊神经网络的诊断方法,能够较好解决网络故障耦合复杂、故障现象与故障原因关系不清晰等难题,对网络故障快速排除与恢复具有重要意义。     

Modeling plasma surface modification of textile fabrics using artificial neural networks

In this paper, a neural network approach is used to understand the effects of fabric features and plasma processing parameters on fabric surface wetting properties. In this approach, fourteen features characterizing woven structures and two plasma parameters are taken as input variables, and the water contact angle cosine and the capillarity height of woven fabrics as output variables. In order to reduce the complexity of the model and effectively learn the network structure from a small number of data, a fuzzy logic based method is used for selecting the most relevant parameters which are taken as input variables of the reduced neural network models. With these relevant parameters, we can effectively control the plasma treatment by selecting the most appropriate fabric materials. Two techniques are used for improving the generalization capability of neural networks: (i) early stopping and (ii) Bayesian regularization. A methodology for optimizing such models is described. The learning abilities and prediction capabilities of the neural net models are compared in terms of different statistical performance criteria. Moreover, a connection weight method is used to determine the relative importance of each input variable in the networks. The obtained results show that neural network models could predict the process performance with reasonable accuracy. However, the neural model trained using Bayesian regularization provides the best results. Thus, it can be concluded that Bayesian network promises to be a valuable quantitative tool to evaluate, understand, and predict woven fabric surface modification by atmospheric air-plasma treatment.     

模拟电路故障特征提取的小波基选取方法研究

小波技术在高维故障特征数据的压缩及敏感信号提取已被广泛应用,但小波基的选取没有一个统一的标准;通过实际采样信号数据的小波分解、特征向量计算、波动性函数比较等技术对小波基函数的选取进行了研究;最后通过综合小波分析、神经网络等技术的模拟电路故障诊断系统的诊断实例验证了所提选取方法的有效性;使用9种常用小波基函数,分别对采样信号进行分解并计算波动性函数,并在模拟电路故障诊断系统进行验证;小波基函数bior2.2的波动较小且与诊断结果一致。