Bayesian Graph Convolutional Neural Networks using Node Copying
This work addresses uncertainty in graph-based learning for applications like node classification, but it is incremental as it builds on an existing Bayesian framework with a new generative model.
The paper tackles the problem of uncertainty in graph structure for graph convolutional neural networks by introducing a node copying generative model within a Bayesian framework, resulting in favorable performance compared to state-of-the-art methods in benchmark node classification tasks.
Graph convolutional neural networks (GCNN) have numerous applications in different graph based learning tasks. Although the techniques obtain impressive results, they often fall short in accounting for the uncertainty associated with the underlying graph structure. In the recently proposed Bayesian GCNN (BGCN) framework, this issue is tackled by viewing the observed graph as a sample from a parametric random graph model and targeting joint inference of the graph and the GCNN weights. In this paper, we introduce an alternative generative model for graphs based on copying nodes and incorporate it within the BGCN framework. Our approach has the benefit that it uses information provided by the node features and training labels in the graph topology inference. Experiments show that the proposed algorithm compares favorably to the state-of-the-art in benchmark node classification tasks.