Reconsidering the Performance of GAE in Link Prediction

TL;DR

This paper demonstrates that a carefully tuned Graph Autoencoder (GAE) can match the performance of more complex models in link prediction tasks, highlighting the importance of proper baselines and hyperparameter tuning.

Contribution

The study systematically applies model-agnostic tricks and hyperparameter tuning to GAEs, showing they can achieve state-of-the-art results and outperform recent sophisticated models in link prediction.

Findings

GAEs can match complex models' performance with proper tuning

Achieved 78.41% Hits@100 on ogbl-ppa dataset

Simple GAE outperforms recent models on certain benchmarks

Abstract

Recent advancements in graph neural networks (GNNs) for link prediction have introduced sophisticated training techniques and model architectures. However, reliance on outdated baselines may exaggerate the benefits of these new approaches. To tackle this issue, we systematically explore Graph Autoencoders (GAEs) by applying model-agnostic tricks in recent methods and tuning hyperparameters. We find that a well-tuned GAE can match the performance of recent sophisticated models while offering superior computational efficiency on widely-used link prediction benchmarks. Our approach delivers substantial performance gains on datasets where structural information dominates and feature data is limited. Specifically, our GAE achieves a state-of-the-art Hits@100 score of 78.41\% on the ogbl-ppa dataset. Furthermore, we examine the impact of various tricks to uncover the reasons behind our success and to guide the design of future methods. Our study emphasizes the critical need to update baselines for a more accurate assessment of progress in GNNs for link prediction. Our code is available at https://github.com/GraphPKU/Refined-GAE.