基于特征子图的异构信息网络节点相似性度量

为解决异构信息网络相似性度量的问题,提出了基于节点特征子图的节点相似性度量算法,通过节点特征子图的最大公共子图与最小公共超图之间的差异性,进行节点间的相似性度量。该算法以图理论为基础,根据连边的不同类型设定不同权值,在考虑节点信息相似的同时,加入节点在网络中的结构信息,最大程度地利用了异构信息网络所富含的信息。实验结果表明,提出的算法具有较好的性能和有效性。

一种基于特征子图的不确定图分类算法

采用频繁子图作为特征子图,对不确定图进行分类.提出AGF频繁子图挖掘算法,该算法将频繁子图挖掘问题转换为频繁项挖掘问题,可有效提高频繁子图生成效率.利用频繁子图构造分类模型,首次应用于不确定图,通过实验证明,给出的分类算法具有良好的分类正确率.

基于特征子图位平面切割和卷积神经网络的青光眼图像分类

青光眼是常见的眼科疾病之一,不及时治疗则会导致眼睛失明,而早期青光眼的诊断主要依赖于眼科医生的经验判断,容易出现误诊或漏诊的情况。为了降低该类情况发生的可能性,本文提出了一种利用位平面切割和特征子图技术来提取图像特征,并且结合卷积神经网络(CNN)分类器来提高青光眼识别率的方法。同时为了验证上述方法对提高图像分类准确率的有效性和对防御对抗样本攻击的有效性,本文首先采用快速梯度符号方法(Fast Gradient Sign Method,FGSM)对原始数据集进行攻击,并利用VGG16网络对受到攻击后的图像提取特征子图,然后将特征子图分解为8个位平面图像,最后将位平面图像输入CNN分类器。实验结果表明,运用该方法提取图像特征后,其对应位平面图像的分类准确率相较于受攻击后的图像有显著提高,准确率最高可从66.5%提升到97.4%,提高了30.9%。本文所提出的特征提取方法能有效提高分类准确率,并且具有抵抗对抗样本攻击的能力。

融合路径与子图特征的知识图谱多跳推理模型

针对知识推理模型在捕获实体之间的复杂语义特征方面难以捕捉多层次语义信息,同时未考虑单一路径的可解释性对正确答案的影响权重不同等问题,提出一种融合路径与子图特征的知识图谱(KG)多跳推理模型PSHAM(Hierarchical Attention Model fusing Path-Subgraph features)。PS-HAM将实体邻域信息与连接路径信息进行融合,并针对不同路径探索多粒度的特征。首先,使用路径级特征提取模块提取每个实体对之间的连接路径,并采用分层注意力机制捕获不同粒度的信息,且将这些信息作为路径级的表示;其次,使用子图特征提取模块通过关系图卷积网络(RGCN)聚合实体的邻域信息;最后,使用路径-子图特征融合模块对路径级与子图级特征向量进行融合,以实现融合推理。在两个公开数据集上进行实验的结果表明,PS-HAM在指标平均倒数秩(MRR)和Hit@k(k=1,3,10)上的性能均存在有效提升。对于指标MRR,与MemoryPath模型相比,PS-HAM在FB15k-237和WN18RR数据集上分别提升了1.5和1.2个百分点。同时,对子图跳数进行的参数验证的结果表明,PS-HAM在两个数据集上都在子图跳数在3时推理效果达到最佳。

量子衍生图像分解和边缘检测

针对数字图像的分解问题,研究了一种基于量子理论的新分解方法,该方法借鉴量子叠加态建立图像像素之间的关联,将一幅数字图像分解为一系列特征子图的叠加。分析了特征子图的意义,并在此基础上给出了一种边缘检测算子的生成规则。利用该规则构造的一种新的图像边缘检测算法被提出,计算机仿真结果验证了该规则的有效性和提出的边缘检测算法的优越性。

面向STEP文件基于属性邻接图的加工特征识别研究

研究了从中性文件STEPAP203文件中得到加工几何特征的方法。首先分析了STEPAP203文件的组织结构,然后从中提取有关面和面之间的连接信息,用属性邻接图表示出来。讨论了基于特征子图匹配的特征识别方法以及属性邻接图的分解。用此特征识别方法实现了中等复杂程度机械零件的特征识别。

基于属性邻接图的加工特征识别研究

从零件的CAD模型中获取工艺设计所需的加工特征信息是CAPP的基础,也是实现CAPP与CAD集成的关键。研究旨在探讨中等复杂程度机械零件的加工特征识别方法。采取方法是从三维设计软件中得到CAD模型,对该模型输出的STEP AP203文件进行处理,获得面特征及其连接边的凹凸性,得到属性邻接图,将其进行处理得到全凹(或全凸)连接的子图,然后将子图中所有代表毛坯面的节点去掉,剩下的就是各特征对应的特征子图。再按照特征子图匹配法,将每个子图与预先定义的特征图进行匹配,就得到了所需要识别的特征。实例结果显示该方法是简便有效的,大大减少了图匹配的计算量。

基于子图特征的节点排序算法

复杂网络理论已被广泛应用于各个领域,节点重要性排序研究是复杂网络领域的重要分支。复杂网络中节点重要性排序及重要节点挖掘对分析和理解复杂网络的结构与功能具有重要意义。众多学者针对复杂网络的关键节点识别和节点重要性排序问题进行了深入研究,取得了大量研究成果。但随着人工智能的发展和数据体量的飞速增长,复杂网络规模呈指数级增长,传统算法的准确性和泛化性已经无法满足现实需求。文中基于节点的二阶邻域信息,提出了一种基于子图特征的节点重要性排序(Subgraph Feature Extraction Rank,SFE Rank)的机器学习模型。利用二阶邻域信息,建立局部子图的含权邻接矩阵,通过矩阵特征分解提取能够有效反映节点局部特征信息的向量表征,在此基础上,建立机器学习模型,用于学习节点子图特征向量和节点重要性的关联关系。在9个真实网络上进行实验,结果表明,相比已有的节点重要性排序方法,所提方法具有更优的排序效果和更好的泛化性能。

完全t部图K(n-k,n,…,n)的色唯一性

设P(G,λ)是图G的色多项式.如果对任意使P(G,λ)=P(H,λ)的图H都与G同构,则称图G是色唯一图.通过比较图的特征子图的个数,讨论了由文献[Koh K M,Teo K L.The search for chromatically unique graphs.Graphs and Combinatorics,1999,6:259-285]中提出的猜想(若n≥k+2,则完全三部图K(n-k,n,n)是色唯一图);推广了文献[Liu Ru-yin,Zhao Hai-xing,Ye Cheng-fu.A complete solution to a conjecture on chromatic unique of complete tripartite graphs.Discrete Mathematics,2004,289:175-179]中的结果(若n≥k+2≥4,则K(n-k,n,n)是色唯一图;若n≥2k≥4,则K(n-k,n-1,n)是色唯一图);证明了若n≥k+2≥4,则K(n-k,n,…,n)是色唯一图,若n≥k+2≥4,则K(n-k,n-1,n,…,n)是色唯一图.

关于完全t部图K(n_1,n_2,…,n_t)的色唯一性

设P(G,λ)是图G的色多项式,如果对任意使P(G,λ)=P(H,λ)的图H都与G同构,则称G是色唯一图。这里通过比较图的特征子图的个数,讨论了由Koh和Teo在文献[1]中提出的问题(若|ni-nj|≤2,1≤i,j≤t且min{n1,n2,…,nt}充分大,K(n1,n2,…,nt)是否为色唯一图?)。证明了,若|ni-nj|≤2且∑ti=1ni>t22+t t-1,则K(n1,n2,…,nt)是色唯一图;若ai=0或k,∑ti=1n+ai>t28k2+|t2k|t-1,则K(n+a1,n+a2,…,n+at)是色唯一图。其条件比文献[4]中的条件较好一些。

基于STEP文件的图匹配方法研究

从零件的CAD模型中获取工艺设计所需的加工特征信息是CAPP的基础,也是实现CAPP与CAD集成的关键。研究旨在探讨中等复杂程度的机械零件加工特征识别方法,采取其方法从三维设计软件中得到CAD模型。对该模型输出的STEP AP203文件进行处理,获得面特征及其连接边的凹凸性,得到属性邻接图,将其进行处理得到全凹(或全凸)连接的子图,然后将子图中所有代表毛坯面的节点去掉,剩下的就是各特征对应的特征子图。再按照特征子图匹配法,将每个子图与预先定义的特征图进行匹配,就得到了所需要识别的特征。实例结果显示,该方法简便有效,大大减少了图匹配的计算量。

基于迭代分解的特征化概率论辩语义求解方法

抽象概率论辩是用于建模不完备、不确定信息推理的一种重要逻辑工具。关于概率论辩语义的求解,现有的基于特征化子图的方法可以在很大程度上提高算法的计算效率,但在枚举特征子图时仍存在一定的盲目性。针对该问题,本文在现有方法的基础上提出了一种基于迭代分解的改进方法,通过对概率论辩框架进行迭代分解,减小了枚举域的大小,降低了计算量。实验结果表明,改进方法在计算速度上较原算法有所提高,提升幅度约为2个数量级。

基于脑功能超网络的多特征融合分类方法

针对在超网络上提取局部脑区指标作为特征,忽视了全局的拓扑信息,继而影响网络拓扑的评估,降低分类器性能的问题,提出了一种基于脑功能超网络的多特征融合分类方法,该方法首先在抑郁症数据集上构建超网络,其次将局部脑区特征和子图特征进行融合。最后采用基于多核的SVM分类器进行分类。为了验证所提方法的有效性,选取28例正常被试和38例抑郁症患者进行实验,结果表明,该方法获得了令人满意的分类准确率,平均可达91.60%。获得的异常区域包括左侧舌回、左侧尾状核、左侧丘脑等重要的抑郁症病发区域。故而该基于脑功能超网络的多特征融合分类方法可以有效地用于分类正常人和抑郁症患者。

Characteristic atom occupation patterns of Au_3Cu,AuCu_3,AuCuI and AuCuII based on experimental data of disordered alloys

The potential energies, volumes and electronic structures of characteristic atoms coordinated by neighboring configurations were obtained from the experimental heats of formation and lattice parameters of disordered Au1-xCux alloys. From characteristic atom occupation （CAO） patterns of L12-Au3Cu, L12-AuCu3 and Llo-AuCu compounds, their electronic structures, volumetric and energetic properties were calculated. The CAO pattern of Johasson-Linde（J-L） model shows that the transition AuCuI→AuCulI is an exothermic and volume contraction reaction, which is opposite from experimental phenomena. According to CAO pattern of Guymont-Feutelais-Legendre（G-F-L） model, the AuCulI cell consists of two periodic antidirection （PAD） AuCuI regions and two PAD boundary regions. The equations derived from CAO pattern of G-F-L model can be used to calculate energetic properties, volumetric properties and ordering degrees of the PAD AuCuI region and PAD boundary region, as well as corresponding average properties of the AuCulI phase. The results are consistent with experimental phenomena.

Product image sentence annotation based on kernel descriptors and tag-rank

Dealing with issues such as too simple image features and word noise inference in product image sentence anmotation, a product image sentence annotation model focusing on image feature learning and key words summarization is described. Three kernel descriptors such as gradient, shape, and color are extracted, respectively. Feature late-fusion is executed in turn by the multiple kernel learning model to obtain more discriminant image features. Absolute rank and relative rank of the tag-rank model are used to boost the key words＇ weights. A new word integration algorithm named word sequence blocks building （WSBB） is designed to create N-gram word sequences. Sentences are generated according to the N-gram word sequences and predefined templates. Experimental results show that both the BLEU-1 scores and BLEU-2 scores of the sentences are superior to those of the state-of-art baselines.

New learning subspace method for image feature extraction

A new method of Windows Minimum/Maximum Module Learning Subspace Algorithm(WMMLSA) for image feature extraction is presented.The WMMLSM is insensitive to the order of the training samples and can regulate effectively the radical vectors of an image feature subspace through selecting the study samples for subspace iterative learning algorithm,so it can improve the robustness and generalization capacity of a pattern subspace and enhance the recognition rate of a classifier.At the same time,a pattern subspace is built by the PCA method.The classifier based on WMMLSM is successfully applied to recognize the pressed characters on the gray-scale images.The results indicate that the correct recognition rate on WMMLSM is higher than that on Average Learning Subspace Method,and that the training speed and the classification speed are both improved.The new method is more applicable and efficient.

FACE RECOGNITION USING TWO DIMENSIONAL LAPLACIAN EIGENMAP

Recently, some research efforts have shown that face images possibly reside on a nonlinear sub-manifold. Though Laplacianfaees method considered the manifold structures of the face images, it has limits to solve face recognition problem. This paper proposes a new feature extraction method, Two Dimensional Laplacian EigenMap （2DLEM）, which especially considers the manifold structures of the face images, and extracts the proper features from face image matrix directly by using a linear transformation. As opposed to Laplacianfaces, 2DLEM extracts features directly from 2D images without a vectorization preprocessing. To test 2DLEM and evaluate its performance, a series of ex- periments are performed on the ORL database and the Yale database. Moreover, several experiments are performed to compare the performance of three 2D methods. The experiments show that 2DLEM achieves the best performance.

Turbulent Characteristic of Liquid Around a Chain of Bubbles in Non-Newtonian Fluid

The turbulence behavior of gas-liquid two-phase flow plays an important role in heat transfer and mass transfer in many chemical processes. In this work, a 2D particle image velocimetry (PIV) was used to investigate the turbulent characteristic of fluid induced by a chain of bubbles rising in Newtonian and non-Newtonian fluids. The instantaneous flow field, turbulent kinetic energy (TKE) and TKE dissipation rate were measured. The results demonstrated that the TKE profiles were almost symmetrical along the column center and showed higher values in the central region of the column. The TKE was enhanced with the increase of gas flow and decrease of liquid viscosity. The maximum TKE dissipation rate appeared on both sides of the bubble chain, and increased with the increase of gas flow rate or liquid viscosity. These results provide an understanding for gas-liquid mass transfer in non-Newtonian fluids.

Hydrodynamics features of dispersed bubbles in the ventilated wake flow of a cylinder

An experimental study was conducted to investigate the 2 D bubbly flow downstream of a cylinder. Sparsely distributed bubbles were produced using the ventilation method. The carrier flow was measured using the particle image velocimetry(PIV) technique. The shadow imaging technique was used to capture instantaneous bubbly flow images. An image-processing code was compiled to identify bubbles in acquired image, calculate the bubble equivalent diameter and the bubble velocity. The effects of Reynolds number and the flow rate of the injected air were considered. The result indicates that the carrier flow is featured by distinct flow structures and the wake region is suppressed as the upstream velocity increases. Regarding the bubbles trapped in the wake flow, the number of small bubbles increases with the upstream velocity. On the whole, the bubble velocity is slightly lower than that of the carrier flow. The consistency between small bubbles and the carrier flow is high in terms of velocity magnitude, which is justified near the wake edge. The difference between the bubble velocity and the carrier flow velocity is remarkable near the wake centerline. For certain Reynolds number, with the increase in the air flow rate, the bubble equivalent diameter increases and the bubble void fraction is elevated.

Some Remarks for Discrete Versions of Nodal Domain Theorems

In this paper, an error is firstly pointed out in the proof of the main theorems (Theorem 4 and Theorem 6) in [1]. Then the error is corrected and the right proof is given.