Graph Neural Networks on Graph Databases

TL;DR

This paper presents a novel method for training graph neural networks directly on graph databases, leveraging native query engines to improve resource efficiency and scalability for large datasets.

Contribution

It introduces a new approach that combines GNN training with graph database query engines, enabling efficient data retrieval and sampling for large-scale graph learning.

Findings

Resource advantages in single-machine training

Effective sampling using query engines

Scalability improvements for large datasets

Abstract

Training graph neural networks on large datasets has long been a challenge. Traditional approaches include efficiently representing the whole graph in-memory, designing parameter efficient and sampling-based models, and graph partitioning in a distributed setup. Separately, graph databases with native graph storage and query engines have been developed, which enable time and resource efficient graph analytics workloads. We show how to directly train a GNN on a graph DB, by retrieving minimal data into memory and sampling using the query engine. Our experiments show resource advantages for single-machine and distributed training. Our approach opens up a new way of scaling GNNs as well as a new application area for graph DBs.