面向多中心数据的超图卷积神经网络及应用

近年来,图神经网络在神经性脑疾病诊断中的应用引起了广泛关注。然而,现有研究中使用的图通常只是基于简单的点对点连接,无法反映3个或更多受试者之间的复杂关联,尤其是在多中心数据集中,即由不同医疗机构所使用的不同采集设备和不同受试人群而集成的具有异质性的数据集。为解决医疗影像数据中存在的多中心异质性问题,提出了一种多中心超图数据结构来描述多中心数据之间的关系。这种超图由两种不同的超边构成,一种是描述单个中心内部关系的中心内超边,另一种是描述不同中心之间关系的跨中心超边。另外,还提出了一种超图卷积神经网络来学习节点的特征表示,这种超图卷积由两部分构成,第一部分是超图节点卷积,第二部分是超边卷积。在两个多中心数据集上的实验结果证明了所提方法的有效性。     

基于超图的多模态情绪识别

近年来多模态情绪识别获得广泛关注,模态间的特征融合决定了情绪识别的效果,现有基于图的情绪特征融合方法多基于二元关系图,在处理三种及以上模态数据时难以实现有效的模态间特征融合,限制了多模态情绪识别的效果.为解决该问题,本文提出基于超图的多模态情绪识别模型（Multi-modal Emotion Recognition Based on Hypergraph,MORAH）,引入超图来建立多模态的多元关系,以此替代现有图结构采用的多个二元关系,实现更加充分、高效的多模态特征融合.具体来说,该模型将多模态特征融合分为两个阶段：超边构建阶段和超图学习阶段.在超边构建阶段,通过胶囊网络实现对序列中每个时间步的信息聚合,并建立单模态的图,然后使用图卷积进行第二次信息聚合,并以此作为下一阶段建立超图的基础,得益于图胶囊聚合方法的加入,MORAH可以同时处理对齐数据和未对齐数据,无需手动对齐;在超图学习阶段,模型建立同一样本不同模态节点之间的关联,以及同类样本所有模态之间的关联,同时,在超图卷积过程中,使用分层多级超边来避免过于平滑的节点嵌入,并使用简化的超图卷积方法来融合模型之间的高级特征,以确保所有...     

基于稀疏重构的超图谱聚类方法

超图谱聚类方法由于能很好地描述数据点间的高阶信息,近年来受到了广泛的关注。不同于传统图结构,超图结构中的超边不是两两数据点间的连接,而是一组具有某种相同特性的数据子集。在实际应用中,常用K-近邻来构建超图中的超边,因此,并没有考虑到数据内在的关联性。提出一种新的基于稀疏重构的超图构建方法。对每一样本,用稀疏表示来找到与其最有关联的近邻样本,以此形成基于稀疏重构的超图模型,使得每个超边内的样本都具有很强的关联性。最后通过对超图拉普拉斯矩阵进行谱分解得到聚类结果。在人脸数据库、手写体数据库上的实验结果验证了算法的有效性。     

基于L0稀疏超图半监督学习的异常用电行为识别

对异常用电行为进行自动分析是电力部门十分关心的问题,故提出一种基于L0稀疏超图半监督学习的用户窃电行为识别算法。与经典图模型中两个顶点的边连接方式不同,超图模型将具有相似特征的多个样本组建为超边,有利于表示用户用电数据间的复杂关联关系。然而常用的K近邻构建超边方法存在超边大小固定,无法有效匹配用户用电数据非均匀分布的缺点。为解决该问题,建立基于L0稀疏重构的超图模型,模型针对每一用户数据建立一个超边,通过L0范数约束的稀疏分解自适应选择与当前用户紧密关联的多个样本,更有利于匹配用户数据的分布结构。继而构建超图拉普拉斯正则约束的半监督分类模型,利用少量的标定样本数据,判别用户用电行为是否存在异常。选取某城市8 900余居民300余天的实际用电数据作为测试样本集,实验结果验证了方法的有效性。     

自注意力超图池化网络

近年来,将卷积神经网络推广到图数据上的图卷积神经网络引起了广泛关注,主要包括重新定义图的卷积和池化操作.由于图数据只能表达二元关系的局限性,使其在实际应用中表现欠佳.相比之下,超图能够捕获数据的高阶相关性,利用其灵活的超边易于处理复杂的数据表示.然而,现有的超图卷积神经网络还不够成熟,目前尚无有效的超图池化操作.因此,提出了带有自注意机制的超图池化网络,使用超图结构建模,通过引入自注意力的超图卷积操作学习带有高阶数据信息的节点隐藏层特征,再经过超图池化操作选择并保留在结构和内容上的重要节点,进而得到更准确的超图表示.在文本分类、菜肴分类和蛋白质分类任务上的实验结果表明:与目前多种主流方法相比,该方法均取得了更好的效果.     

基于线-超图神经网络的阿尔兹海默症分类

在阿尔兹海默症分类问题中,超图神经网络可以从被试间的超图关系中提取特征,在表示学习复杂图结构方面具有很好的优势,但大多数模型都直接或间接地将超图所表示的被试间的高阶复杂关系分解,转化为简单的二元关系进行特征学习,没有有效利用超边的高阶信息,因此提出了基于线-超图神经网络(line-hypergraph neural network, L-HGNN)的阿尔兹海默症分类模型,该模型利用稀疏线性回归表征被试间多元相关性,借助超图和线图的转换在神经网络模型中实现节点的高阶邻域信息传递和超边整体结构特征学习,同时,结合注意力机制生成更具区分性的节点嵌入,进而用于阿尔兹海默症的辅助诊断.在ADNI数据上与常用的两种方法比较,实验结果表明,该方法能有效提高分类准确率,在阿尔兹海默症早期诊断上具有重要的应用价值.     

基于概率超图的视频事件语义检测

视频事件类别的归属具有模糊性和不确定性,将超图的点边射入矩阵拓展成概率形式的软超图进行关联关系分析和语义分析,将会更有利于提高多事件检索检测的精准率和召回率。提出基于概率超图模型的视频事件语义检测算法（PHVESD）。 该方法首先将颜色、灰度共生矩阵、Tchebichef矩、局部二值模式（LBP）等四种底层视觉特征进行融合; 然后定义视频段的亲密度函数并利用亲密度的信息构建概率超图模型,其中每条超边对应一种事件语义;采用随机游走过程来预测视频段属于每条超边的概率;最后结合阈值采用条件概率模型对视频段进行事件语义分类。将该方法用于交通突发事件多语义检测中并与其他的识别算法相比较,实验结果表明,与基于超图模型的多标签随机游走算法（MLRW）相比,PHVESD的算法使多语义事件检测的准确率提高了10%,召回率提高了8%。     

基于超边相关性的图像分类方法

传统基于超图的图像分类方法在构建超图时并未考虑各超边之间的关系,导致最终分类效果不理想.文中结合图像视觉信息和标注信息量化超边间相关性,提出一种基于超边相关性的图像分类方法,有效地将图像相关的标注信息作为判定图像类别的指标引入到图像分类中,进而对图像进行更准确的分类.在LabelMe和UIUC数据集上的实验验证该方法的有效性.     

面向异质变分超图自动编码器的超边链接预测模型

如何采用超边建模网络数据中的多元关联关系,实现潜在超边链接关系的预测具有重要的现实意义。现有方法主要集中于研究具有成对关系的网络数据,然而,直接将现有的链接预测方法用于超图网络中的超边链接预测具有一定的局限性。因此,提出基于异质变分超图自动编码器的超边链接预测模型(heterogeneous variational hypergraph autoencoder,HVGAE)。首先,利用超图卷积实现变分超图自动编码器,将超图网络数据转换成一种低维空间表示;其次,加入节点近邻度函数,最大程度地保留其结构信息,从而构建异质超图网络超边链接预测模型。针对三种不同类型的超图网络进行实验,结果表明相比其他的基准方法,HVGAE模型获得了较好的预测结果,说明其能够较好地解决超图网络中的超边链接预测问题。     

面向时间感知的知识超图链接预测

知识超图是一种使用多元关系表示现实世界的异构图,但无论在通用领域还是垂直领域,现有的知识超图普遍存在不完整的情况.因此,如何通过知识超图中已有的链接推理缺失的链接是一个具有挑战性的问题.目前大多数研究使用基于多元关系的知识表示学习方法完成知识超图的链接预测任务,但这些方法仅从时间未知的超边中学习实体与关系的嵌入向量,没有考虑时间因素对事实动态演变的影响,导致在动态环境中的预测性能较差.首先,根据本文首次提出的时序知识超图定义,提出时序知识超图链接预测模型,同时从实体角色、位置和时序超边的时间戳中学习实体的静态表征和动态表征,以一定比例融合后作为实体嵌入向量用于链接预测任务,实现对超边时序信息的充分利用.同时,从理论上证明模型具有完全表达性和线性空间复杂度.此外,通过上市公司的公开经营数据构建时序知识超图数据集CB67,并在该数据集上进行了大量实验评估.实验结果表明:模型能够在时序知识超图数据集上有效地执行链接预测任务.     

Nonlinear Dimensionality Reduction by Topologically Constrained Isometric Embedding

Many manifold learning procedures try to embed a given feature data into a flat space of low dimensionality while preserving as much as possible the metric in the natural feature space. The embedding process usually relies on distances between neighboring features, mainly since distances between features that are far apart from each other often provide an unreliable estimation of the true distance on the feature manifold due to its non-convexity. Distortions resulting from using long geodesics indiscriminately lead to a known limitation of the Isomap algorithm when used to map non-convex manifolds. Presented is a framework for nonlinear dimensionality reduction that uses both local and global distances in order to learn the intrinsic geometry of flat manifolds with boundaries. The resulting algorithm filters out potentially problematic distances between distant feature points based on the properties of the geodesics connecting those points and their relative distance to the boundary of the feature manifold, thus avoiding an inherent limitation of the Isomap algorithm. Since the proposed algorithm matches non-local structures, it is robust to strong noise. We show experimental results demonstrating the advantages of the proposed approach over conventional dimensionality reduction techniques, both global and local in nature.     

A unified framework of image latent feature learning on Sina microblog

Large-scale user-contributed images with texts are rapidly increasing on the social media websites, such as Sina microblog. However, the noise and incomplete correspondence between the images and the texts give rise to the difficulty in precise image retrieval and ranking. In this paper, a hypergraph-based learning framework is proposed for image ranking, which simultaneously utilizes visual feature, textual content and social link information to estimate the relevance between images. Representing each image as a vertex in the hypergraph, complex relationship between images can be reflected exactly. Then updating the weight of hyperedges throughout the hypergraph learning process, the effect of different edges can be adaptively modulated in the constructed hypergraph. Furthermore, the popularity degree of the image is employed to re-rank the retrieval results. Comparative experiments on a large-scale Sina microblog data-set demonstrate the effectiveness of the proposed approach.     

基于属性自表达的低秩超图属性选择算法*

针对高维数据具有低秩形式和属性冗余等特点,提出一种基于属性自表达的无监督超图属性选择算法。具体地,该算法首先利用属性自表达特点用其他属性稀疏地表达每个属性,此自表达形式使用低秩假设寻找高维数据的低秩表示,然后建立超图正则化因子保持高维数据的局部结构,最后利用稀疏正则化因子进行属性选择。属性自表达特性确定属性的重要性,低秩表示相当于考虑数据的全局信息进行子空间学习,超图正则化因子考虑数据的局部结构对数据进行子空间学习。该算法实际上考虑数据全局和局部信息进行子空间学习,更是一种嵌入了子空间学习的属性选择算法。实验结果表明,该算法相比其它对比算法,能更有效地选取属性,并能取得很好的分类效果。     

Latent representation learning based autoencoder for unsupervised feature selection in hyperspectral imagery

In hyperspectral image (HSI) analysis, high-dimensional data may contain noisy, irrelevant and redundant information. To mitigate the negative effect from these information, feature selection is one of the useful solutions. Unsupervised feature selection is a data preprocessing technique for dimensionality reduction, which selects a subset of informative features without using any label information. Different from the linear models, the autoencoder is formulated to nonlinearly select informative features. The adjacency matrix of HSI can be constructed to extract the underlying relationship between each data point, where the latent representation of original data can be obtained via matrix factorization. Besides, a new feature representation can be also learnt from the autoencoder. For a same data matrix, different feature representations should consistently share the potential information. Motivated by these, in this paper, we propose a latent representation learning based autoencoder feature selection (LRLAFS) model, where the latent representation learning is used to steer feature selection for the autoencoder. To solve the proposed model, we advance an alternative optimization algorithm. Experimental results on three HSI datasets confirm the effectiveness of the proposed model.

     

基于图嵌入的自适应多视降维方法

随着监控摄像头的普及和数据采集技术的快速发展,多视数据呈现出规模大、维度高和多源异构的特点,使得数据存储空间大、传输慢、算法复杂度高,造成“有数据、难利用”的困境。到目前为止,国内外在多视降维方面的研究还比较少。针对这一问题,本文提出一种基于图嵌入的自适应多视降维方法。该方法在考虑视角内降维后数据重构原始高维数据的基础上,提出自适应学习相似矩阵来探索不同视角之间降维后数据的关联关系,学习各视数据的正交投影矩阵实现多视降维任务。本文在多个数据集上对降维后的多视数据进行了聚类/识别实验验证,实验结果表明基于图嵌入的自适应多视降维方法优于其他降维方法。     

Clustering ensemble via structured hypergraph learning

Clustering ensemble integrates multiple base clustering results to obtain a consensus result and thus improves the stability and robustness of the single clustering method. Since it is natural to use a hypergraph to represent the multiple base clustering results, where instances are represented by nodes and base clusters are represented by hyperedges, some hypergraph based clustering ensemble methods are proposed. Conventional hypergraph based methods obtain the final consensus result by partitioning a pre-defined static hypergraph. However, since base clusters may be imperfect due to the unreliability of base clustering methods, the pre-defined hypergraph constructed from the base clusters is also unreliable. Therefore, directly obtaining the final clustering result by partitioning the unreliable hypergraph is inappropriate. To tackle this problem, in this paper, we propose a clustering ensemble method via structured hypergraph learning, i.e., instead of being constructed directly, the hypergraph is dynamically learned from base results, which will be more reliable. Moreover, when dynamically learning the hypergraph, we enforce it to have a clear clustering structure, which will be more appropriate for clustering tasks, and thus we do not need to perform any uncertain postprocessing, such as hypergraph partitioning. Extensive experiments show that, our method not only performs better than the conventional hypergraph based ensemble methods, but also outperforms the state-of-the-art clustering ensemble methods.     

无监督混阶栈式稀疏自编码器的图像分类学习

目前多数图像分类的方法是采用监督学习或者半监督学习对图像进行降维,然而监督学习与半监督学习需要图像携带标签信息。针对无标签图像的降维及分类问题,提出采用混阶栈式稀疏自编码器对图像进行无监督降维来实现图像的分类学习。首先,构建一个具有三个隐藏层的串行栈式自编码器网络,对栈式自编码器的每一个隐藏层单独训练,将前一个隐藏层的输出作为后一个隐藏层的输入,对图像数据进行特征提取并实现对数据的降维。其次,将训练好的栈式自编码器的第一个隐藏层和第二个隐藏层的特征进行拼接融合,形成一个包含混阶特征的矩阵。最后,使用支持向量机对降维后的图像特征进行分类,并进行精度评价。在公开的四个图像数据集上将所提方法与七个对比算法进行对比实验,实验结果表明,所提方法能够对无标签图像进行特征提取,实现图像分类学习,减少分类时间,提高图像的分类精度。     

Hypergraph recovery algorithms from a given vector of vertex degrees

Methods of obtaining certain classes of hypergraphs from a given integer vector of vertex degrees are considered. These classes are as follows: hyperedges with unit weight incident upon k vertices; hyperedges with unit weight incident upon k vertices in the case when the vertices may be non-unique; multiple hyperedges incident upon k vertices; and arbitrary hypergraph in which the edges can contain any set of k vertices. For each of these classes, an algorithm is proposed for constructing the hypergraph from an arbitrary vector. If the construction is impossible, the algorithm determines how much the vector should be reduced so that the hypergraph could be constructed.     

Adaptive hypergraph superpixels

Superpixel segmentation, which amounts to partitioning an image into a number of superpixels each of which is a set of pixels sharing common visual meanings, requires specific needs for different computer vision tasks. Graph based methods, as a kind of popular superpixel segmentation method, regard an image as a weighted graph whose nodes correspond to pixels of the image, and partition all pixels into superpixels according to the similarity between pixels over various feature spaces. Despite their improvement of the performance of segmentation, these methods ignore high-order relationship between them incurred from either locally neighboring pixels or structured layout of the image. Moreover, they measure the similarity of pairwise pixels using Gaussian kernel where a robust radius parameter is difficult to find for pixels which exhibit multiple features (e.g., texture, color, brightness). In this paper, we propose an adaptive hypergraph superpixel segmentation (AHS) of intensity images for solving both issues. AHS constructs a hypergraph by building the hyperedges with an adaptive neighborhood scheme, which explores an intrinsic relationship of pixels. Afterwards, AHS encodes the relationship between pairwise pixels using characteristics of current two pixels as well as their neighboring pixels defined by hyperedges. Essentially, AHS models the relationship of pairwise pixels in a high-order group fashion while graph based methods evaluate it in a one-vs-one fashion. Experiments on four datasets demonstrate that AHS achieves higher or comparable performance compared with state-of-the-art methods.