Graph transformer has been proven as an effective graph learning method for its adoption of attention mechanism that is capable of capturing expressive representations from complex topological and feature information of graphs. Graph transformer conventionally performs dense attention (or global attention) for every pair of nodes to learn node representation vectors, resulting in quadratic computational costs that are unaffordable for large-scale graph data. Therefore, mini-batch training for graph transformers is a promising direction, but limited samples in each mini-batch can not support effective dense attention to encode informative representations. Facing this bottleneck, (1) we start by assigning each node a token list that is sampled by personalized PageRank (PPR) and then apply standard multi-head self-attention only on this list to compute its node representations. This PPR tokenization method decouples model training from complex graph topological information and makes heavy feature engineering offline and independent, such that mini-batch training of graph transformers is possible by loading each node's token list in batches. We further prove this PPR tokenization is viable as a graph convolution network with a fixed polynomial filter and jumping knowledge. However, only using personalized PageRank may limit information carried by a token list, which could not support different graph inductive biases for model training. To this end, (2) we rewire graphs by introducing multiple types of virtual connections through structure- and content-based super nodes that enable PPR tokenization to encode local and global contexts, long-range interaction, and heterophilous information into each node's token list, and then formalize our Virtual Connection Ranking based Graph Transformer (VCR-Graphormer).
In graph machine learning, data collection, sharing, and analysis often involve multiple parties, each of which may require varying levels of data security and privacy. To this end, preserving privacy is of great importance in protecting sensitive information. In the era of big data, the relationships among data entities have become unprecedentedly complex, and more applications utilize advanced data structures (i.e., graphs) that can support network structures and relevant attribute information. To date, many graph-based AI models have been proposed (e.g., graph neural networks) for various domain tasks, like computer vision and natural language processing. In this paper, we focus on reviewing privacy-preserving techniques of graph machine learning. We systematically review related works from the data to the computational aspects. We first review methods for generating privacy-preserving graph data. Then we describe methods for transmitting privacy-preserved information (e.g., graph model parameters) to realize the optimization-based computation when data sharing among multiple parties is risky or impossible. In addition to discussing relevant theoretical methodology and software tools, we also discuss current challenges and highlight several possible future research opportunities for privacy-preserving graph machine learning. Finally, we envision a unified and comprehensive secure graph machine learning system.
Graph pre-training strategies have been attracting a surge of attention in the graph mining community, due to their flexibility in parameterizing graph neural networks (GNNs) without any label information. The key idea lies in encoding valuable information into the backbone GNNs, by predicting the masked graph signals extracted from the input graphs. In order to balance the importance of diverse graph signals (e.g., nodes, edges, subgraphs), the existing approaches are mostly hand-engineered by introducing hyperparameters to re-weight the importance of graph signals. However, human interventions with sub-optimal hyperparameters often inject additional bias and deteriorate the generalization performance in the downstream applications. This paper addresses these limitations from a new perspective, i.e., deriving curriculum for pre-training GNNs. We propose an end-to-end model named MentorGNN that aims to supervise the pre-training process of GNNs across graphs with diverse structures and disparate feature spaces. To comprehend heterogeneous graph signals at different granularities, we propose a curriculum learning paradigm that automatically re-weighs graph signals in order to ensure a good generalization in the target domain. Moreover, we shed new light on the problem of domain adaption on relational data (i.e., graphs) by deriving a natural and interpretable upper bound on the generalization error of the pre-trained GNNs. Extensive experiments on a wealth of real graphs validate and verify the performance of MentorGNN.
Directly motivated by security-related applications from the Homeland Security Enterprise, we focus on the privacy-preserving analysis of graph data, which provides the crucial capacity to represent rich attributes and relationships. In particular, we discuss two directions, namely privacy-preserving graph generation and federated graph learning, which can jointly enable the collaboration among multiple parties each possessing private graph data. For each direction, we identify both "quick wins" and "hard problems". Towards the end, we demonstrate a user interface that can facilitate model explanation, interpretation, and visualization. We believe that the techniques developed in these directions will significantly enhance the capabilities of the Homeland Security Enterprise to tackle and mitigate the various security risks.
Disinformation refers to false information deliberately spread to influence the general public, and the negative impact of disinformation on society can be observed for numerous issues, such as political agendas and manipulating financial markets. In this paper, we identify prevalent challenges and advances related to automated disinformation detection from multiple aspects, and propose a comprehensive and explainable disinformation detection framework called DISCO. It leverages the heterogeneity of disinformation and addresses the prediction opaqueness. Then we provide a demonstration of DISCO on a real-world fake news detection task with satisfactory detection accuracy and explanation. The demo video and source code of DISCO is now publicly available. We expect that our demo could pave the way for addressing the limitations of identification, comprehension, and explainability as a whole.
Graph neural networks (GNNs) integrate the comprehensive relation of graph data and the representation learning capability of neural networks, which is one of the most popular deep learning methods and achieves state-of-the-art performance in many applications, such as natural language processing and computer vision. In real-world scenarios, increasing the depth (i.e., the number of layers) of GNNs is sometimes necessary to capture more latent knowledge of the input data to mitigate the uncertainty caused by missing values. However, involving more complex structures and more parameters will decrease the performance of GNN models. One reason called oversmoothing is recently introduced but the relevant research remains nascent. In general, oversmoothing makes the final representations of nodes indiscriminative, thus deteriorating the node classification and link prediction performance. In this paper, we first survey the current de-oversmoothing methods and propose three major metrics to evaluate a de-oversmoothing method, i.e., constant divergence indicator, easy-to-determine divergence indicator, and model-agnostic strategy. Then, we propose the Topology-guided Graph Contrastive Layer, named TGCL, which is the first de-oversmoothing method maintaining all three mentioned metrics. With the contrastive learning manner, we provide the theoretical analysis of the effectiveness of the proposed TGCL. Last but not least, we design extensive experiments to illustrate the empirical performance of TGCL comparing with state-of-the-art baselines.
Nowadays, many network representation learning algorithms and downstream network mining tasks have already paid attention to dynamic networks or temporal networks, which are more suitable for real-world complex scenarios by modeling evolving patterns and temporal dependencies between node interactions. Moreover, representing and mining temporal networks have a wide range of applications, such as fraud detection, social network analysis, and drug discovery. To contribute to the network representation learning and network mining research community, in this paper, we generate a new biological dataset of dynamic protein-protein interaction networks (i.e., DPPIN), which consists of twelve dynamic protein-level interaction networks of yeast cells at different scales. We first introduce the generation process of DPPIN. To demonstrate the value of our published dataset DPPIN, we then list the potential applications that would be benefited. Furthermore, we design dynamic local clustering, dynamic spectral clustering, dynamic subgraph matching, dynamic node classification, and dynamic graph classification experiments, where DPPIN indicates future research opportunities for some tasks by presenting challenges on state-of-the-art baseline algorithms. Finally, we identify future directions for improving this dataset utility and welcome inputs from the community. All resources of this work are deployed and publicly available at https://github.com/DongqiFu/DPPIN.