Redesigning graph filter-based GNNs to relax the homophily assumption

TL;DR

This paper introduces a new GNN architecture that effectively handles both homophilic and heterophilic graph data by reinterpreting graph filters, enhancing expressiveness, and preventing oversmoothing, with proven permutation equivariance.

Contribution

The paper proposes a novel GNN architecture that relaxes the homophily assumption, improving performance on heterophilic graphs and incorporating a stronger inductive bias.

Findings

Outperforms state-of-the-art baselines on various datasets

Handles both homophilic and heterophilic data effectively

Is permutation equivariant

Abstract

Graph neural networks (GNNs) have become a workhorse approach for learning from data defined over irregular domains, typically by implicitly assuming that the data structure is represented by a homophilic graph. However, recent works have revealed that many relevant applications involve heterophilic data where the performance of GNNs can be notably compromised. To address this challenge, we present a simple yet effective architecture designed to mitigate the limitations of the homophily assumption. The proposed architecture reinterprets the role of graph filters in convolutional GNNs, resulting in a more general architecture while incorporating a stronger inductive bias than GNNs based on filter banks. The proposed convolutional layer enhances the expressive capacity of the architecture enabling it to learn from both homophilic and heterophilic data and preventing the issue of oversmoothing. From a theoretical standpoint, we show that the proposed architecture is permutation equivariant. Finally, we show that the proposed GNNs compares favorably relative to several state-of-the-art baselines in both homophilic and heterophilic datasets, showcasing its promising potential.