Neural Sheaf Diffusion: A Topological Perspective on Heterophily and Oversmoothing in GNNsDownload PDF

Published: 31 Oct 2022, Last Modified: 06 Jan 2023NeurIPS 2022 AcceptReaders: Everyone
Keywords: sheaf, graph, neural network, diffusion, heterophily, oversmoothing, algebraic topology
TL;DR: We study heterophily and oversmoothing in graph neural networks through the lens of sheaf theory and propose Sheaf Neural Networks to address these two problems.
Abstract: Cellular sheaves equip graphs with a ``geometrical'' structure by assigning vector spaces and linear maps to nodes and edges. Graph Neural Networks (GNNs) implicitly assume a graph with a trivial underlying sheaf. This choice is reflected in the structure of the graph Laplacian operator, the properties of the associated diffusion equation, and the characteristics of the convolutional models that discretise this equation. In this paper, we use cellular sheaf theory to show that the underlying geometry of the graph is deeply linked with the performance of GNNs in heterophilic settings and their oversmoothing behaviour. By considering a hierarchy of increasingly general sheaves, we study how the ability of the sheaf diffusion process to achieve linear separation of the classes in the infinite time limit expands. At the same time, we prove that when the sheaf is non-trivial, discretised parametric diffusion processes have greater control than GNNs over their asymptotic behaviour. On the practical side, we study how sheaves can be learned from data. The resulting sheaf diffusion models have many desirable properties that address the limitations of classical graph diffusion equations (and corresponding GNN models) and obtain competitive results in heterophilic settings. Overall, our work provides new connections between GNNs and algebraic topology and would be of interest to both fields.
Supplementary Material: pdf
17 Replies