Redesigning graph filter-based GNNs to relax the homophily assumption
This addresses a key limitation for users of GNNs in applications with heterophilic data, though it appears incremental as it builds on existing filter-based methods.
The paper tackles the problem of graph neural networks (GNNs) performing poorly on heterophilic data by redesigning graph filter-based GNNs to relax the homophily assumption, resulting in a more general architecture that compares favorably to state-of-the-art baselines on both homophilic and heterophilic datasets.
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.