Fisher-Bures Adversary Graph Convolutional Networks

Code

• github.com/stellargraph/FisherGCN
  OfficialIn papertf★ 0

Abstract

In a graph convolutional network, we assume that the graph G is generated wrt some observation noise. During learning, we make small random perturbations G of the graph and try to improve generalization. Based on quantum information geometry, G can be characterized by the eigendecomposition of the graph Laplacian matrix. We try to minimize the loss wrt the perturbed G+G while making G to be effective in terms of the Fisher information of the neural network. Our proposed model can consistently improve graph convolutional networks on semi-supervised node classification tasks with reasonable computational overhead. We present three different geometries on the manifold of graphs: the intrinsic geometry measures the information theoretic dynamics of a graph; the extrinsic geometry characterizes how such dynamics can affect externally a graph neural network; the embedding geometry is for measuring node embeddings. These new analytical tools are useful in developing a good understanding of graph neural networks and fostering new techniques.

Tasks

Reproductions