Transferability of Graph Neural Networks using Graphon and Sampling Theories

TL;DR

This paper introduces a graphon-based neural network architecture that guarantees transferability of GNNs across large graphs, addressing challenges of graph size variability and reducing the need for retraining.

Contribution

It presents an explicit two-layer graphon neural network architecture with proven approximation capabilities and transferability across large, convergent graph sequences.

Findings

WNN can approximate bandlimited graphon signals within specified error.

GNN transferability is established for large graphs in convergent sequences.

Addresses issues related to the curse of dimensionality in GNNs.

Abstract

Graph neural networks (GNNs) have become powerful tools for processing graph-based information in various domains. A desirable property of GNNs is transferability, where a trained network can swap in information from a different graph without retraining and retain its accuracy. A recent method of capturing transferability of GNNs is through the use of graphons, which are symmetric, measurable functions representing the limit of large dense graphs. In this work, we contribute to the application of graphons to GNNs by presenting an explicit two-layer graphon neural network (WNN) architecture. We prove its ability to approximate bandlimited graphon signals within a specified error tolerance using a minimal number of network weights. We then leverage this result, to establish the transferability of an explicit two-layer GNN over all sufficiently large graphs in a convergent sequence. Our work addresses transferability between both deterministic weighted graphs and simple random graphs and overcomes issues related to the curse of dimensionality that arise in other GNN results. The proposed WNN and GNN architectures offer practical solutions for handling graph data of varying sizes while maintaining performance guarantees without extensive retraining.