Open Your Eyes: Vision Enhances Message Passing Neural Networks in Link Prediction

TL;DR

This paper introduces Graph Vision Network (GVN), a novel framework that incorporates visual perception into message-passing neural networks to improve link prediction accuracy across diverse datasets.

Contribution

It pioneers the integration of vision-based structural awareness into MPNNs, achieving state-of-the-art results and demonstrating the effectiveness of visual perception in link prediction tasks.

Findings

GVN outperforms existing methods on seven datasets.

GVN achieves new state-of-the-art results.

Vision enhancement benefits are consistent across large-scale graphs.

Abstract

Message-passing graph neural networks (MPNNs) and structural features (SFs) are cornerstones for the link prediction task. However, as a common and intuitive mode of understanding, the potential of visual perception has been overlooked in the MPNN community. For the first time, we equip MPNNs with vision structural awareness by proposing an effective framework called Graph Vision Network (GVN), along with a more efficient variant (E-GVN). Extensive empirical results demonstrate that with the proposed frameworks, GVN consistently benefits from the vision enhancement across seven link prediction datasets, including challenging large-scale graphs. Such improvements are compatible with existing state-of-the-art (SOTA) methods and GVNs achieve new SOTA results, thereby underscoring a promising novel direction for link prediction.