RAN-GNNs: breaking the capacity limits of graph neural networks

TL;DR

This paper introduces randomly wired graph neural network architectures that enhance capacity and performance by effectively merging multi-scale receptive fields, overcoming depth limitations in traditional GNNs.

Contribution

It demonstrates that randomly wired architectures outperform traditional deep GNNs by acting as ensembles of varied receptive fields, with adjustable widths for improved representation.

Findings

Randomly wired GNNs outperform traditional deep GNNs across multiple tasks.

These architectures behave like ensembles of paths with diverse receptive fields.

Experimental results validate the effectiveness of the proposed approach.

Abstract

Graph neural networks have become a staple in problems addressing learning and analysis of data defined over graphs. However, several results suggest an inherent difficulty in extracting better performance by increasing the number of layers. Recent works attribute this to a phenomenon peculiar to the extraction of node features in graph-based tasks, i.e., the need to consider multiple neighborhood sizes at the same time and adaptively tune them. In this paper, we investigate the recently proposed randomly wired architectures in the context of graph neural networks. Instead of building deeper networks by stacking many layers, we prove that employing a randomly-wired architecture can be a more effective way to increase the capacity of the network and obtain richer representations. We show that such architectures behave like an ensemble of paths, which are able to merge contributions from receptive fields of varied size. Moreover, these receptive fields can also be modulated to be wider or narrower through the trainable weights over the paths. We also provide extensive experimental evidence of the superior performance of randomly wired architectures over multiple tasks and four graph convolution definitions, using recent benchmarking frameworks that addresses the reliability of previous testing methodologies.