Improving Graph Neural Networks at Scale: Combining Approximate PageRank and CoreRank

TL;DR

This paper introduces CorePPR, a scalable GNN model that combines approximate PageRank and CoreRank with a dynamic neighbor selection mechanism, improving efficiency and performance on large graphs.

Contribution

The paper proposes CorePPR, a novel scalable GNN approach that integrates approximate PageRank and CoreRank with dynamic neighbor selection for large-scale graph learning.

Findings

CorePPR outperforms PPRGo on large graphs.

Dynamic neighbor selection reduces training time.

CorePPR maintains high performance with improved scalability.

Abstract

Graph Neural Networks (GNNs) have achieved great successes in many learning tasks performed on graph structures. Nonetheless, to propagate information GNNs rely on a message passing scheme which can become prohibitively expensive when working with industrial-scale graphs. Inspired by the PPRGo model, we propose the CorePPR model, a scalable solution that utilises a learnable convex combination of the approximate personalised PageRank and the CoreRank to diffuse multi-hop neighbourhood information in GNNs. Additionally, we incorporate a dynamic mechanism to select the most influential neighbours for a particular node which reduces training time while preserving the performance of the model. Overall, we demonstrate that CorePPR outperforms PPRGo, particularly on large graphs where selecting the most influential nodes is particularly relevant for scalability. Our code is publicly available at: https://github.com/arielramos97/CorePPR.