MODIFIED DOUBLE-GRAPH DECOMPOSITION ANALYSIS FOR FINDING SYMBOLIC NETWORK FUNCTIONS

The concepts of complementary cofactor pairs, normal double-graphs and feasible torn vertex seta are introduced. By using them a decomposition theorem for first-order cofactor C(Y) is derived. Combining it with the modified double-graph method, a new decomposition analysis-modified double-graph decomposition analysis is presented for finding symbolic network functions. Its advantages are that the resultant symbolic expressions are compact and contain no cancellation terms, and its sign evaluation is very simple.

Neural network-based matrix effect correction in EDXRF analysis

In this paper we discuss neural network-based matrix effect correction in energy dispersive X-ray fluorescence (EDXRF) analysis,with detailed algorithm to classify the samples.The method can correct the matrix effect effectively through classifying the samples automatically,and influence of X-ray absorption and enhancement by major elements of the samples is reduced.Experiments for the complex matrix effect correction in EDXRF analysis of samples in Pangang showed improved accuracy of the elemental analysis result.

Network analysis using organizational risk analyzer

The tool system of the organizational risk analyzer (ORA) to study the network of East Turkistan terrorists is selected. The model of the relationships among its personnel, knowledge, resources and task entities is represented by the meta-matrix in ORA, with which to analyze the risks and vulnerabilities of organizational structure quantitatively, and obtain the last vulnerabilities and risks of the organization. Case study in this system shows that it should be a shortcut to destroy effectively the network of terrorists by recognizing the caucus persons of the terrorism organization for the first and eliminating them when strikes the terror organization. It is vital to ensure effective use of the resources and control the risks of terrorist attacks.

Parallel Active Subspace Decomposition for Tensor Robust Principal Component Analysis

Tensor robust principal component analysis has received a substantial amount of attention in various fields.Most existing methods,normally relying on tensor nuclear norm minimization,need to pay an expensive computational cost due to multiple singular value decompositions at each iteration.To overcome the drawback,we propose a scalable and efficient method,named parallel active subspace decomposition,which divides the unfolding along each mode of the tensor into a columnwise orthonormal matrix(active subspace)and another small-size matrix in parallel.Such a transformation leads to a nonconvex optimization problem in which the scale of nuclear norm minimization is generally much smaller than that in the original problem.We solve the optimization problem by an alternating direction method of multipliers and show that the iterates can be convergent within the given stopping criterion and the convergent solution is close to the global optimum solution within the prescribed bound.Experimental results are given to demonstrate that the performance of the proposed model is better than the state-of-the-art methods.

Codimensional matrix pairing perspective of BYY harmony learning:hierarchy of bilinear systems,joint decomposition of data-covariance,and applications of network biology

One paper in a preceding issue of this journal has introduced the Bayesian Ying-Yang(BYY)harmony learning from a perspective of problem solving,parameter learning,and model selection.In a complementary role,the paper provides further insights from another perspective that a co-dimensional matrix pair(shortly co-dim matrix pair)forms a building unit and a hierarchy of such building units sets up the BYY system.The BYY harmony learning is re-examined via exploring the nature of a co-dim matrix pair,which leads to improved learning performance with refined model selection criteria and a modified mechanism that coordinates automatic model selection and sparse learning.Besides updating typical algorithms of factor analysis(FA),binary FA(BFA),binary matrix factorization(BMF),and nonnegative matrix factorization(NMF)to share such a mechanism,we are also led to(a)a new parametrization that embeds a de-noise nature to Gaussian mixture and local FA(LFA);(b)an alternative formulation of graph Laplacian based linear manifold learning;(c)a codecomposition of data and covariance for learning regularization and data integration;and(d)a co-dim matrix pair based generalization of temporal FA and state space model.Moreover,with help of a co-dim matrix pair in Hadamard product,we are led to a semi-supervised formation for regression analysis and a semi-blind learning formation for temporal FA and state space model.Furthermore,we address that these advances provide with new tools for network biology studies,including learning transcriptional regulatory,Protein-Protein Interaction network alignment,and network integration.

An architecture decomposition method of pneumatic catapult system based on OPM and DSM

The architecture strategy of the Unmanned Aerial Vehicle(UAV)pneumatic launch system should continue to evolve to adapt to complex and variable operating environments.Architecture representation,decomposition perspective,and cluster analysis play a vital role in the early phase of system architecture development.In order for the system to emerge anticipated and desirable intrinsic functional properties,an architecture decomposition method based on the ObjectProcess Methodology(OPM)and Design Structure Matrix(DSM)is put forward in this paper.The OPM is proposed to model the UAV launch process formally,and the matrix representation of the architecture of the pneumatic launch system is established.After the extension of the definition and operations of DSM,with the Idicula-Gutierrez-Thebeau Algorithm plus(IGTA+)clustering algorithm,the transformation of the pneumatic launch system architecture from process decomposition to function decomposition is demonstrated in this paper.The analysis shows that the architecture decomposition of the pneumatic launch system meets the functional requirements of stakeholders.

Fused empirical mode decomposition and wavelets for locating combined damage in a truss-type structure through vibration analysis

Structural health monitoring (SHM) is a relevant topic for civil systems and involves the monitoring, data processing and interpretation to evaluate the condition of a structure, in order to detect damage. In real structures, two or more sites or types of damage can be present at the same time. It has been shown that one kind of damaged condition can interfere with the detection of another kind of damage, leading to an incorrect assessment about the structure condition. Identifying combined damage on structures still represents a challenge for condition monitoring, because the reliable identification of a combined damaged condition is a difficult task. Thus, this work presents a fusion of methodologies, where a single wavelet-packet and the empirical mode decomposition (EMD) method are combined with artificial neural networks (ANNs) for the automated and online identification-location of single or multiple-combined damage in a scaled model of a five-bay truss-type structure. Results showed that the proposed methodology is very efficient and reliable for identifying and locating the three kinds of damage, as well as their combinations. Therefore, this methodology could be applied to detection-location of damage in real truss-type structures, which would help to improve the characteristics and life span of real structures.

Nondestructive determination of the freshness change in bighead carp heads under variable temperatures by using excitation-emission matrix fl uorescence and back-propagation neural networks

This study established back-propagation neural networks(BPNNs)for evaluating the freshness of bighead carp(Hypophthalmichthys nobilis)heads during chilled storage via fluorescence spectroscopy using an excitation-emission matrix(EEM).The total volatile basic nitrogen(TVB-N)and total aerobic count(TAC)of fish increased obviously during storage at 0,4,8,12,and 16°C,while sensory scores decreased with increasing storage time.The EEM fluorescence intensity was measured,and its change was correlated with the freshness indicators of the samples.Three characteristic components of EEM data were extracted by parallel factor analysis,and two freshness indicators were used to construct the EEM-BPNNs model.The results demonstrated that the relative errors of the EEM-BPNNs model for TVB-N and TAC were less than 14%.This result indicated that the EEM-BPNNs model could determine the freshness of fish in cold chains in a rapid and nondestructive way.

Modeling the Drilling Process of Aluminum Composites Using Multiple Regression Analysis and Artificial Neural Networks

In recent years, aluminum-matrix composites (AMCs) have been widely used to replace cast iron in aerospace and automotive industries. Machining of these composite materials requires better understanding of cutting processes re- garding accuracy and efficiency. This study addresses the modeling of the machinability of self-lubricated aluminum /alumina/graphite hybrid composites synthesized by the powder metallurgy method. In this study, multiple regression analysis (MRA) and artificial neural networks (ANN) were used to investigate the influence of some parameters on the thrust force and torque in the drilling processes of self-lubricated hybrid composite materials. The models were identi- fied by using cutting speed, feed, and volume fraction of the reinforcement particles as input data and the thrust force and torque as the output data. A comparison between two prediction methods was developed to compare the prediction accuracy. ANNs showed better predictability results compared to MRA due to the nonlinearity nature of ANNs. The statistical analysis accompanied with artificial neural network results showed that Al2O3, Gr and cutting feed (f) were the most significant parameters on the drilling process, while spindle speed seemed insignificant. Since the spindle speed was insignificant, it directed us to set it either at the highest spindle speed to obtain high material removal rate or at the lowest spindle speed to prolong the tool life depending on the need for the application.

白条猪价格预测模型构建

【目的】增强农产品价格预测准确度,为农产品价格的有效预测提供参考。【方法】以河南省白条猪每周平均批发价格为研究对象,提出一种基于序列分解、主成分分析和神经网络(CEEMDAN-PCA-CNN-LSTM)的白条猪价格预测方法。首先,使用自适应白噪声完全集合模态分解方法(CEEMDAN)对白条猪价格序列进行分解;其次,选用皮尔逊相关系数筛选影响价格波动的相关因素;再次,利用主成分分析(PCA)对影响因素及分解得到的子序列降维处理并作为原始价格序列的特征值,并行输入到作为编码器的卷积神经网络(CNN)中进行特征提取;最后,引入长短期记忆网络(LSTM)作为解码器输出得到预测结果。将该方法应用于河南省白条猪每周平均价格数据,与LSTM、门控循环单元(GRU)、CNN、基于卷积的长短期记忆网络(ConvLSTM)模型进行比较。【结果】CEEMDAN-PCA-CNN-LSTM组合模型预测方法得到的平均绝对误差分别降低了44.95%、27.30%、28.13%、43.17%。【结论】CEEMDAN-PCA-CNN-LSTM模型对于河南省白条猪市场价格的预测性能更优,有助于相关部门针对河南省白条猪价格波动做出科学决策。

一种融合节点变化信息的动态社区发现方法

动态社区发现旨在检测动态复杂网络中蕴含的社区结构,对于揭示网络的功能及演化模式具有重要研究价值.由于相邻时刻网络的社区结构具有平滑性,前一时刻网络的社区划分信息可以用于监督当前时刻网络的社区划分过程,但已有方法均难以有效提取这些信息来提高动态社区发现性能.针对该问题,提出一种融合节点变化信息的动态社区发现方法(Semi-supervised Nonnegative Matrix Factorization combining Node Change Information,NCI-SeNMF).NCI-SeNMF首先采用k-core分析方法提取前一时刻社区网络的degeneracy-core,并选取degeneracy-core中的节点构造社区隶属先验信息,然后对相邻时刻网络的节点局部拓扑结构变化程度进行量化,并将其用于进一步修正社区隶属先验信息,最后通过半监督非负矩阵分解模型集成社区隶属先验信息进行动态社区发现.在多个人工合成动态网络和真实世界动态网络上进行大量对比实验,结果表明,NCI-SeNMF比现有动态社区发现方法在主要评价指标上至少提升了4.8%.

DeepSVDNet:A Deep Learning-Based Approach for Detecting and Classifying Vision-Threatening Diabetic Retinopathy in Retinal Fundus Images

Artificial Intelligence(AI)is being increasingly used for diagnosing Vision-Threatening Diabetic Retinopathy(VTDR),which is a leading cause of visual impairment and blindness worldwide.However,previous automated VTDR detection methods have mainly relied on manual feature extraction and classification,leading to errors.This paper proposes a novel VTDR detection and classification model that combines different models through majority voting.Our proposed methodology involves preprocessing,data augmentation,feature extraction,and classification stages.We use a hybrid convolutional neural network-singular value decomposition(CNN-SVD)model for feature extraction and selection and an improved SVM-RBF with a Decision Tree(DT)and K-Nearest Neighbor(KNN)for classification.We tested our model on the IDRiD dataset and achieved an accuracy of 98.06%,a sensitivity of 83.67%,and a specificity of 100%for DR detection and evaluation tests,respectively.Our proposed approach outperforms baseline techniques and provides a more robust and accurate method for VTDR detection.

可对角化矩阵特征值分解扰动问题的快速求解方法

针对特征值扰动计算的传统方法收敛速度慢的问题,提出了一种求解特征值扰动问题的快速迭代算法.首先,通过矩阵变换将初始矩阵的特征值扰动问题转化为对角矩阵的特征值扰动问题.然后,提出了一种快速迭代算法求解扰动参数,同时对算法的收敛性进行分析,并将其与基于摄动级数展开法导出的方法进行对比.再次,采用逐一求解特征值并进行矩阵降阶的策略,有效降低运算量.最后,通过2个算例分别展示算法的计算过程及其在结构模态参数追踪方面的应用效果.

基于复杂网络分析的广东省城镇居民碳足迹研究

应用投入产出模型和复杂网络分析方法,计算了广东省城镇居民直接碳足迹、间接碳足迹,构建了城镇居民间接碳足迹复杂网络,并对碳足迹重点产业部门进行了结构分解分析。研究结果表明:(1)广东省城镇居民直接碳足迹的主要来源是汽油和液化石油气,间接碳足迹网络的核心产业部门是电力、热力的生产和供应业。(2)广东省城镇居民碳足迹的主要来源为间接碳足迹,对所构建的碳足迹网络进行整体结构和个体特征两方面的分析发现,城镇居民间接碳足迹网络的通达性逐年向好,但各产业部门之间的协同碳关联关系的发挥尚有提升空间。(3)根据产业部门在碳足迹网络中的地位和作用,运用相关指标将其划分为碳核心社区、碳中介社区、碳边缘社区,在对碳核心社区的结构分解分析中发现,碳排放强度效应均为减碳因素,消费水平效应、城镇化水平效应、人口效应为增碳因素,且消费水平效应的增碳贡献率最大。

卫生健康标准中关系型数据共现矩阵计算及SAS程序实现

目的设计卫生健康标准中关系型数据共现矩阵的求解路径及其SAS实现方法。方法文章以计算一组标准的起草单位共现矩阵为例,将“标准-起草单位”二维表格导入SAS(版本号:9.4),按照“两两相乘”的计算思路自行设计宏程序计算起草单位之间的共现矩阵。结果以计算一组标准的起草单位共现矩阵为例,采用自行构建的模拟数据进行演示。首先,导入宏循环起始的数据集,并计算共现频次C_(j,k);然后,导出为out数据集,进一步合并数据集,形成最终的共现矩阵。结论SAS宏程序计算标准中关系型数据共现矩阵具有灵活高效的优势,可用于社会网络分析和互动演进规律总结。

宽度-深度融合时频分析的径流智能预测方法

为解决现有基于LSTM的径流预测模型易陷入局部最优的问题,提出了基于VMD-LSTMBLS(variational mode decomposition-LSTM-broad learning system)的径流预测模型。将宽度学习系统与LSTM结合,针对径流序列多噪音特点,采用时频分析方法中的变分模态分解,将径流时间序列的一维时域信号变换到二维时频平面,减少噪声对预测结果的影响。仿真结果表明:与基线模型及现有基于LSTM的径流预测模型相比,该模型的预测精度有较为明显的提高。

基于测点聚类的POD-BPNN风压重构方法

文章提出本征正交分解(POD)与聚类分析结合的结构表面风压测点分类与关键测点布置方法,基于少量测点的风压数据,通过POD与误差反向传播神经网络(BPNN)方法实现方柱结构表面风压时程的重构。机器学习数据集为多风向角均匀来流下单方柱测压风洞试验得到的测点风压时程。将44个测点的风压时程数据POD降维,并采用K-means++聚类分析得到方柱周向轮廓系数分布,并基于轮廓系数的多风向角平均值,得到12、16、20和24个关键测点的轴对称布置方案。以关键测点的风压时程数据为训练集,采用POD-BPNN方法重构方柱表面其余测点所在位置的风压时程,并将风压时程及其统计值同试验结果对比。从12~20测点方案,风压重构精度逐步提升;20测点和24测点方案的重构风压差异较小,二者都能较好地重构方柱表面风压分布,仅在0°风向角方柱脉动风压误差偏大。

基于经验模态分解和深度学习的短期风电功率预测

精准的风电功率预测有利于全网电力平衡、系统安全稳定运行和节能减耗。提出一种基于经验模态分解(empirical mode decomposition, EMD)、核主成分分析(kernel principal component analysis, KPCA)和长短期记忆(long short-term memory, LSTM)神经网络的短期风功率预测模型。采用EMD技术将多维气象序列分解为多个固有模态分量,以挖掘原始数据的主要特征并消除噪声;引入KPCA进行降维处理,提取数据的非线性特征;使用LSTM神经网络对特征提取的序列进行学习并完成预测,获得风电功率预测的最终结果。使用所提出的模型对新疆某一风电场风电功率进行预测,将预测结果与其他模型对比。结果表明,该预测模型能改善预测性能,降低风电功率预测误差。

Windowed SSA （Singular Spectral Analysis） for Geophysical Time Series Analysis

Although the SSA （singular spectral analysis） is a potential tool for analysing time series of different physical processes, the processing of large geophysical data set requires more time and is found to be computationally expansive. In particular for the SVD （singular value decomposition） of large trajectory matrix, the processing units require huge memory and high performance computing system. In the present work, we propose an alternative scheme based on WSSA （windowed singular spectral analysis）, which is robust for analysing long data sets without losing any valuable low-frequency information contained in the data. The underlying scheme reduces the floating point operations in SVD computations as the size of the trajectory matrix is small in windowed processing. In order to test the efficiency, the authors applied the proposed method on two geophysical data sets i.e., the climatic record with 30,000 data points and seismic reflection trace with 8,000 data points. The authors have shown that without distorting any physical information, the low-frequency contents of the data are well preserved after the windowed processing in both the cases.

基于密集连接卷积神经网络的结构损伤识别

提出一种经验模态分解(empirical mode decomposition, EMD)和密集连接卷积神经网络(densely connected convolutional network, DenseNet)相结合的结构损伤识别网络模型(E-DenseNet)。对采集的加速度信号进行EMD得到多个本征模态函数(intrinsic mode function, IMF)分量,接着剔除皮尔逊相关系数绝对值较小的弱相关IMF分量。根据输入数据的组织方式,设定3种E-DenseNet模型:E-DenseNet1利用强相关IMF分量重构信号建立一维单通道输入数据;E-DenseNet2将各强相关IMF分量分别视作一个通道来建立一维多通道输入数据;E-DenseNet3利用所有强相关IMF分量组成二维矩阵来建立二维单通道输入数据。某简支梁算例分析表明:E-DenseNet1计算速度快但识别精度低,E-DenseNet2计算速度快且识别精度高,E-DenseNet3识别精度高但计算速度慢;与一维多通道残差卷积神经网络(residual network, ResNet)及标准卷积神经网络(convolutional neural network, CNN)相比,E-DenseNet2的识别精度明显更优;E-DenseNet2因而具有兼顾计算效率和识别精度的优点。E-DenseNet2可视化分析表明了其识别过程,对于相同工况下的不同样本,输出层越深其输出特征越相似,直至全连接层给出极大相似输出特征。