Adversarially Regularized Graph Autoencoder for Graph Embedding

Abstract

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.

References in corpus (12)

• Efficient Estimation of Word Representations in Vector Space
• Semi-Supervised Classification with Graph Convolutional Networks
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Graph Embedding Techniques, Applications, and Performance: A Survey
• Energy-based Generative Adversarial Network
• Variational Graph Auto-Encoders
• Network Embedding as Matrix Factorization: Unifying DeepWalk, LINE, PTE, and node2vec
• Adversarial Feature Learning
• Adversarially Learned Inference
• Adversarial Network Embedding
• A Survey on Network Embedding
• Network Representation Learning: A Survey

Cited by in corpus (27)

• Fast Graph Representation Learning with PyTorch Geometric
• Variational Graph Normalized Auto-Encoders
• Graph Autoencoders for Embedding Learning in Brain Networks and Major Depressive Disorder Identification
• A Comparative Study for Unsupervised Network Representation Learning
• AEGCN: An Autoencoder-Constrained Graph Convolutional Network
• Learning Graph Neural Networks with Positive and Unlabeled Nodes
• Implementing graph neural networks with TensorFlow-Keras
• Attributed Graph Clustering via Adaptive Graph Convolution
• Node Embeddings and Exact Low-Rank Representations of Complex Networks
• Neural Link Prediction with Walk Pooling
• FedGL: Federated Graph Learning Framework with Global Self-Supervision
• ANAE: Learning Node Context Representation for Attributed Network Embedding
• Adversarial Graph Disentanglement
• Distribution-induced Bidirectional Generative Adversarial Network for Graph Representation Learning
• Feature Interaction-aware Graph Neural Networks
• Node Copying: A Random Graph Model for Effective Graph Sampling
• Edge: Enriching Knowledge Graph Embeddings with External Text
• Permutation-Invariant Variational Autoencoder for Graph-Level Representation Learning
• Learning Features of Network Structures Using Graphlets
• Bipartite Link Prediction based on Topological Features via 2-hop Path
• Memory-Associated Differential Learning
• Barlow Graph Auto-Encoder for Unsupervised Network Embedding
• Brain Graph Super-Resolution Using Adversarial Graph Neural Network with Application to Functional Brain Connectivity
• Comparison of Syntactic and Semantic Representations of Programs in Neural Embeddings
• Multi-Hot Compact Network Embedding
• Anomaly Detection for Aggregated Data Using Multi-Graph Autoencoder
• Label Embedding with Partial Heterogeneous Contexts