Keep It Simple: Graph Autoencoders Without Graph Convolutional Networks
This work questions the need for complex models in graph representation learning, suggesting simpler methods can be effective for many applications, but it is incremental as it modifies an existing approach without introducing a new paradigm.
The authors tackled the problem of graph autoencoders relying on complex graph convolutional networks (GCN) by proposing a simple linear model based on the adjacency matrix, and found that it consistently achieves competitive performance on tasks like link prediction and node clustering across real-world datasets such as Cora, Citeseer, and Pubmed.
Graph autoencoders (AE) and variational autoencoders (VAE) recently emerged as powerful node embedding methods, with promising performances on challenging tasks such as link prediction and node clustering. Graph AE, VAE and most of their extensions rely on graph convolutional networks (GCN) to learn vector space representations of nodes. In this paper, we propose to replace the GCN encoder by a simple linear model w.r.t. the adjacency matrix of the graph. For the two aforementioned tasks, we empirically show that this approach consistently reaches competitive performances w.r.t. GCN-based models for numerous real-world graphs, including the widely used Cora, Citeseer and Pubmed citation networks that became the de facto benchmark datasets for evaluating graph AE and VAE. This result questions the relevance of repeatedly using these three datasets to compare complex graph AE and VAE models. It also emphasizes the effectiveness of simple node encoding schemes for many real-world applications.