Large Scale Learning on Non-Homophilous Graphs: New Benchmarks and Strong Simple Methods

TL;DR

This paper introduces new large-scale non-homophilous graph datasets, demonstrates the limitations of existing methods, and proposes LINKX, a simple yet effective approach that achieves state-of-the-art results on these challenging datasets.

Contribution

The paper provides diverse large-scale non-homophilous datasets and introduces LINKX, a simple method that outperforms existing approaches on these datasets.

Findings

Existing scalable graph methods perform poorly on non-homophilous data.

LINKX achieves state-of-the-art results on large non-homophilous graphs.

New datasets enable better evaluation of non-homophilous graph learning methods.

Abstract

Many widely used datasets for graph machine learning tasks have generally been homophilous, where nodes with similar labels connect to each other. Recently, new Graph Neural Networks (GNNs) have been developed that move beyond the homophily regime; however, their evaluation has often been conducted on small graphs with limited application domains. We collect and introduce diverse non-homophilous datasets from a variety of application areas that have up to 384x more nodes and 1398x more edges than prior datasets. We further show that existing scalable graph learning and graph minibatching techniques lead to performance degradation on these non-homophilous datasets, thus highlighting the need for further work on scalable non-homophilous methods. To address these concerns, we introduce LINKX -- a strong simple method that admits straightforward minibatch training and inference. Extensive experimental results with representative simple methods and GNNs across our proposed datasets show that LINKX achieves state-of-the-art performance for learning on non-homophilous graphs. Our codes and data are available at https://github.com/CUAI/Non-Homophily-Large-Scale.