Graph convolutional autoencoders with co-learning of graph structure and node attributes

CGMAE: Self-supervised Masked Auto-Encoder with Cross-Graph node alignment for node classification

Graph embedding is an effective method to represent graph data in a low dimensional space for graph analytics. Most existing embedding algorithms typically focus on preserving the topological structure or minimizing the reconstruction errors of graph data, but they have mostly ignored the data distribution of the latent codes from the graphs, which often results in inferior embedding in real-world graph data. In this paper, we propose a novel adversarial graph embedding framework for graph data. The framework encodes the topological structure and node content in a graph to a compact representation, on which a decoder is trained to reconstruct the graph structure. Furthermore, the latent representation is enforced to match a prior distribution via an adversarial training scheme. To learn a robust embedding, two variants of adversarial approaches, adversarially regularized graph autoencoder (ARGA) and adversarially regularized variational graph autoencoder (ARVGA), are developed. Experimental studies on real-world graphs validate our design and demonstrate that our algorithms outperform baselines by a wide margin in link prediction, graph clustering, and graph visualization tasks.

Graph embedding is of great significance for the research and analysis of graphs. Graph embedding aims to map nodes in the network to low-dimensional vectors while preserving information in the original graph of nodes. In recent years, the appearance of graph neural networks has significantly improved the accuracy of graph embedding. However, the influence of clusters was not considered in existing graph neural network (GNN)-based methods, so this paper proposes a new method to incorporate the influence of clusters into the generation of graph embedding. We use the attention mechanism to pass the message of the cluster pooled result and integrate the whole process into the graph autoencoder as the third layer of the encoder. The experimental results show that our model has made great improvement over the baseline methods in the node clustering and link prediction tasks, demonstrating that the embeddings generated by our model have excellent expressiveness.

Link prediction, which utilizes the information of endpoint and network structure to predict the unknown links between two nodes, has attracted much attention in recent years. The network topological attributes contain the structure attributes and node attributes. However, some existing methods focus on the node attributes, while others focus on the structure attributes. To solve this problem, we propose a prediction method based on attributes fusion which combines node attributes and structure attributes. In our proposed method, we first analyze the structural attributes based on common neighbors in directed networks and define the structural attribute similarity. Then the similarity contribution of the influence of the common neighbors to the predicted nonadjacent nodes is analyzed. Experimental results on 9 directed networks show that our proposed index achieves higher performance than existing mainstream baselines under the precision evaluation.

In this paper, we propose a model using a graph convolution network for predicting the accuracy of the automatically generated convolution neural network (CNN). In recent years, there have been many excellent results in the automatic designing of deep learning network structures and their hyperparameters. These works mainly use evolutionary algorithms and reinforcement learning methods. But to verify the accuracy of newly generated neural network samples, it requires a lot of computational time and large scale of hardware. Therefore, how to reduce training time and improve the utilization of samples has become an important problem of automatic design for deep learning networks. On the other hand, the effectiveness of the graph convolution network (GCN) has been shown in processing graph data on the tasks of graph node classification, link prediction, and clustering. We are considering using GCN to predict the accuracy of the automatically designed neural networks. We compare GCN, a multi-layer neural network(NN), and CNN on the accuracy prediction task of automatically generated CNN for image classification tasks on the CIFAR-10 dataset. Our experiment demonstrates the advantages of GCN on small datasets compared to the multi-layer neural network and CNN, while GCN achieves the equivalent accuracy to the other methods on large datasets.

Recent works show that Graph Neural Networks (GNNs) are highly non-robust with respect to adversarial attacks on both the graph structure and the node attributes, making their outcomes unreliable. We propose the first method for certifiable (non-)robustness of graph convolutional networks with respect to perturbations of the node attributes. We consider the case of binary node attributes (e.g. bag-of-words) and perturbations that are L_0-bounded. If a node has been certified with our method, it is guaranteed to be robust under any possible perturbation given the attack model. Likewise, we can certify non-robustness. Finally, we propose a robust semi-supervised training procedure that treats the labeled and unlabeled nodes jointly. As shown in our experimental evaluation, our method significantly improves the robustness of the GNN with only minimal effect on the predictive accuracy.

Graph anonymization is a necessary process before any publication of data from a social network in order to preserve the privacy of individuals, especially when these networks contain sensitive or personally identifiable information, such as social contacts, personal opinions and recordings of private communications. So, this publication can therefore constitute a threat to privacy. Generally, this simple procedure is not sufficient, some inference attacks on the published graph can lead to de-anonymize certain nodes, to learn the existence of a social relation between two nodes or to use the structure of the graph itself to deduce the value of certain sensitive attributes. For this reason, several anonymization techniques were proposed in the literature to combat this problem, they modify the structure of the social graph by adding and/or removing nodes and/or edges, these methods can be classified as Graph Modification, Generalization or Clustering and Differential Privacy Models. In this paper, we present a brief yet systematic review of the existing anonymization techniques for privacy preserving publishing of social network data. We identify the best graph anonymization method that ensure a high confidentiality level after making a clear comparison between them. Finally, we present our perspective or our proposition that still under development which is based on differential privacy method and attacking the graph processing in different ways at the same time, graph structure and node attributes.

Neuroimaging plays an significant role in diagnosing and pathological study of brain diseases. Considering that both functional and structural abnormalities may lead to brain dis-eases and disorders, single modal neuroimaging approach may not fully characterize brain activities and working modes. Fusion of multimodal neuroimaging data is expected to provide more comprehensive characterization of brain diseases, given that the different modalities contain more complementary information. Recently, Graph Convolutional Networks (GCNs) is shown to have powerful capacity in representation learning for graph-structure data, which is considered to integrate both graph se-mantic structure and node information. Therefore, in this paper, we propose the Weighted Graph AutoEncoder (WGAE), a GCN- driven multimodal fusion model, to learn the combinational latent node representation of fMRI and DTI neuroimaging data, which are used as node features and graph structure respectively in the graph in unsupervised manner. Experimental results on two real-world datasets show the superiority of the proposed model over other existing single-modal or multi-modal methods in learning representations for disease prediction as the downstream task. Furthermore, ablation experiments also show the collaborative contribution of multimodal neuroimaging fusion in the proposed model, and also show the feasibility of assessing the respective importance of the two modalities during the disease prediction.

Network embedding aims to learn distributed vector representations of nodes in a network. The problem of network embedding is fundamentally important. It plays crucial roles in many applications, such as node classification, link prediction, and so on. As the real-world networks are often sparse with few observed links, many recent works have utilized the local and global network structure proximity with shallow models for better network embedding. In reality, each node is usually associated with rich attributes. Some attributed network embedding models leveraged the node attributes in these shallow network embedding models to alleviate the data sparsity issue. Nevertheless, the underlying structure of the network is complex. What is more, the connection between the network structure and node attributes is also hidden. Thus, these previous shallow models fail to capture the nonlinear deep information embedded in the attributed network, resulting in the suboptimal embedding results. In this paper, we propose a deep attributed network embedding framework to capture the complex structure and attribute information. Specifically, we first adopt a personalized random walk-based model to capture the interaction between network structure and node attributes from various degrees of proximity. After that, we construct an enhanced matrix representation of the attributed network by summarizing the various degrees of proximity. Then, we design a deep neural network to exploit the nonlinear complex information in the enhanced matrix for network embedding. Thus, the proposed framework could capture the complex attributed network structure by preserving both the various degrees of network structure and node attributes in a unified framework. Finally, empirical experiments show the effectiveness of our proposed framework on a variety of network embedding-based tasks.

Multi-view attribute learning and context relationship encoding enhanced segmentation of lung tumors from CT images

The vanilla Graph Convolutional Network (GCN) uses a low-pass filter to extract low-frequency signals from graph topology, which may lead to the over-smoothing problem when GCN goes deep. To this end, various methods have been proposed to create an adaptive filter by incorporating an extra filter (e.g., a high-pass filter) extracted from the graph topology. However, these methods heavily rely on topological information and ignore the node attribute space, which severely sacrifices the expressive power of the deep GCNs, especially when dealing with disassortative graphs. In this paper, we propose a cross-space adaptive filter, called CSF, to produce the adaptive-frequency information extracted from both the topology and attribute spaces. Specifically, we first derive a tailored attribute-based high-pass filter that can be interpreted theoretically as a minimizer for semi-supervised kernel ridge regression. Then, we cast the topology-based low-pass filter as a Mercer's kernel within the context of GCNs. This serves as a foundation for combining it with the attribute-based filter to capture the adaptive-frequency information. Finally, we derive the cross-space filter via an effective multiple-kernel learning strategy, which unifies the attribute-based high-pass filter and the topology-based low-pass filter. This helps to address the over-smoothing problem while maintaining effectiveness. Extensive experiments demonstrate that CSF not only successfully alleviates the over-smoothing problem but also promotes the effectiveness of the node classification task. Our code is available at https://github.com/huangzichun/Cross-Space-Adaptive-Filter.

Attributed network embedding aims to learn low-dimensional vector representations for nodes in a network, where each node contains rich attributes/features describing node content. Because network topology structure and node attributes often exhibit high correlation, incorporating node attribute proximity into network embedding is beneficial for learning good vector representations. In reality, large-scale networks often have incomplete/missing node content or linkages, yet existing attributed network embedding algorithms all operate under the assumption that networks are complete. Thus, their performance is vulnerable to missing data and suffers from poor scalability. In this paper, we propose a Scalable Incomplete Network Embedding (SINE) algorithm for learning node representations from incomplete graphs. SINE formulates a probabilistic learning framework that separately models pairs of node-context and node-attribute relationships. Different from existing attributed network embedding algorithms, SINE provides greater flexibility to make the best of useful information and mitigate negative effects of missing information on representation learning. A stochastic gradient descent based online algorithm is derived to learn node representations, allowing SINE to scale up to large-scale networks with high learning efficiency. We evaluate the effectiveness and efficiency of SINE through extensive experiments on real-world networks. Experimental results confirm that SINE outperforms state-of-the-art baselines in various tasks, including node classification, node clustering, and link prediction, under settings with missing links and node attributes. SINE is also shown to be scalable and efficient on large-scale networks with millions of nodes/edges and high-dimensional node features. The source code of this paper is available at https://github.com/daokunzhang/SINE.

Drug repositioning framework using embedding drug-protein-disease similarities with graph convolution network and ensemble learning

Despite the recent success of graph convolutional networks (GCNs) in modeling graph structured data, its vulnerability to adversarial attacks has been revealed and attacks on both node feature and graph structure have been designed. Direct extension of defense algorithms based on adversarial samples meets with immediate challenge because computing the adversarial network costs substantially. We propose addressing this issue by perturbing the latent representations in GCNs, which not only dispenses with generating adversarial networks, but also attains improved robustness and accuracy by respecting the latent manifold of the data. This new framework of latent adversarial training on graphs is applied to node classification, link prediction, and recommender systems. Our empirical experimental results confirm the superior robustness performance over strong baselines.

IntroductionComplex regulatory networks controlled by transcription factor (TF)–gene interactions are involved in inflammatory bone diseases, such as periodontitis. Understanding these networks is crucial for identifying master regulators and potential treatment targets. Current models frequently use correlation-based or black-box machine learning techniques, which are not structurally accurate or biologically interpretable. Moreover, most frameworks do not utilize the representational power of quantum-derived data features. This study overcomes these constraints by combining quantum-enhanced graph neural networks to decode TF-gene regulatory networks implicated in periodontal bone inflammation.MethodsWe constructed a directed transcription factor (TF)- gene regulatory network using 1207 carefully selected interactions from the TRRUST v2 human database, which encompassed 231 transcription factors and 536 target genes. One-hot encoded node features were used to train the Graph Autoencoder (GAE) and Graph Generative Adversarial Network (Graph GAN) architectures. We applied quantum data feature extraction to enhance node representation using variational quantum circuits constructed in PennyLane, where classical embeddings were encoded into qubit rotations and entangled states. New quantum features were created by measuring the expectation values of Pauli-Z operators. Distribution divergence measures (KL, JS, Wasserstein, MMD), embedding quality metrics (silhouette score, centrality correlation), and link prediction metrics (AUC, Average Precision) were used to assess performance.ResultsOn every metric, GAE performed noticeably better than Graph GAN. It performed better in clustering (silhouette score = 0.272 vs. 0.107 for GAN) and link prediction accuracy (AUC = 0.997, AP = 0.994). While GAN embeddings displayed little structural alignment, GAE-generated embeddings strongly correlated with network centrality measures, emphasizing biological interpretability. Quantum-enhanced features revealed distinct regulatory modules associated with inflammation and bone resorption pathways, and they maintained the network topology more effectively. We found central regulators with high embedding scores, including NF-κB and STAT3. Distributional analyses validated the fundamental differences between GAE and GAN embeddings with a symmetric KL divergence of 6.76 and a Jensen-Shannon distance of 0.47.ConclusionOur results demonstrate that Graph Autoencoders provide a reliable and comprehensible framework for simulating TF-gene regulatory networks, particularly when combined with quantum-derived feature extraction. The GAE is ideally suited to elucidating the molecular underpinnings of periodontal bone inflammation due to its ability to maintain biological structure, pinpoint important regulatory hubs, and enhance downstream analyses, such as clustering. This method enables the prioritization of periodontitis regulatory targets for upcoming treatment advancements. This integrated computational approach lays the foundation for more biologically based and quantum-aware modelling of intricate regulatory systems in inflammation-related diseases.