OOD Link Prediction Generalization Capabilities of Message-Passing GNNs in Larger Test GraphsDownload PDF

Published: 31 Oct 2022, Last Modified: 14 Dec 2022NeurIPS 2022 AcceptReaders: Everyone
Keywords: OOD, GNNs, link prediction, Message Passing GNNs, random graphs, graphon
Abstract: This work provides the first theoretical study on the ability of graph Message Passing Neural Networks (gMPNNs) ---such as Graph Neural Networks (GNNs)--- to perform inductive out-of-distribution (OOD) link prediction tasks, where deployment (test) graph sizes are larger than training graphs. We first prove non-asymptotic bounds showing that link predictors based on permutation-equivariant (structural) node embeddings obtained by gMPNNs can converge to a random guess as test graphs get larger. We then propose a theoretically-sound gMPNN that outputs structural pairwise (2-node) embeddings and prove non-asymptotic bounds showing that, as test graphs grow, these embeddings converge to embeddings of a continuous function that retains its ability to predict links OOD. Empirical results on random graphs show agreement with our theoretical results.
TL;DR: This work proves bounds on the ability of (structural) node and pairwise message-passing GNNs to inductively predict links OOD when test graphs are larger than training graphs
Supplementary Material: pdf
18 Replies