Boost then Convolve: Gradient Boosting Meets Graph Neural Networks

TL;DR

This paper introduces a novel joint architecture combining gradient boosted decision trees and graph neural networks to effectively handle graphs with heterogeneous tabular node features, outperforming existing models.

Contribution

The authors propose a new end-to-end trainable model that integrates GBDT and GNN, specifically designed for graphs with heterogeneous features, which is a novel approach in this domain.

Findings

Significant performance improvements over existing GBDT and GNN models.

Effective handling of heterogeneous tabular features in graph data.

End-to-end training enhances model synergy and accuracy.

Abstract

Graph neural networks (GNNs) are powerful models that have been successful in various graph representation learning tasks. Whereas gradient boosted decision trees (GBDT) often outperform other machine learning methods when faced with heterogeneous tabular data. But what approach should be used for graphs with tabular node features? Previous GNN models have mostly focused on networks with homogeneous sparse features and, as we show, are suboptimal in the heterogeneous setting. In this work, we propose a novel architecture that trains GBDT and GNN jointly to get the best of both worlds: the GBDT model deals with heterogeneous features, while GNN accounts for the graph structure. Our model benefits from end-to-end optimization by allowing new trees to fit the gradient updates of GNN. With an extensive experimental comparison to the leading GBDT and GNN models, we demonstrate a significant increase in performance on a variety of graphs with tabular features. The code is available: https://github.com/nd7141/bgnn.