Adversarially Regularized Graph Autoencoder for Graph Embedding

13 Feb 2018  ·  Shirui Pan, Ruiqi Hu, Guodong Long, Jing Jiang, Lina Yao, Chengqi Zhang ·

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.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Graph Clustering Citeseer ARGE ARI 34.1 # 4
F1 54.6 # 2
NMI 0.35 # 8
Precision 57.3 # 1
ACC 57.3 # 6
Graph Clustering Citeseer ARVGE ARI 24.5 # 5
F1 52.9 # 3
NMI 26.1 # 7
Precision 54.9 # 2
ACC 54.4 # 7
Link Prediction Citeseer ARGE AUC 91.9 # 10
AP 93 # 9
Graph Clustering Cora ARVGE ARI 37.4 # 4
F1 62.7 # 2
NMI 45 # 7
Precision 62.4 # 2
ACC 63.8 # 7
Link Prediction Cora ARGE AUC 92.4% # 11
AP 93.2% # 10
Graph Clustering Cora ARGE ARI 35.2 # 5
F1 61.9 # 3
NMI 0.449 # 8
Precision 64.6 # 1
ACC 64 # 6
Link Prediction Pubmed ARGE AUC 96.8% # 5
AP 97.1% # 5

Methods