Two Heads Are Better Than One: Boosting Graph Sparse Training via Semantic and Topological Awareness

TL;DR

This paper introduces Graph Sparse Training (GST), a novel method that dynamically manipulates graph sparsity to preserve topological and semantic information, significantly improving efficiency and robustness in large-scale GNNs.

Contribution

GST is the first approach to dynamically align sparse graphs with a topology & semantic anchor, enhancing sparsity levels while maintaining performance and topological integrity.

Findings

Identifies subgraphs at higher sparsity levels than existing methods

Preserves key spectral properties of graphs

Achieves significant speedup in GNN inference and improves robustness

Abstract

Graph Neural Networks (GNNs) excel in various graph learning tasks but face computational challenges when applied to large-scale graphs. A promising solution is to remove non-essential edges to reduce the computational overheads in GNN. Previous literature generally falls into two categories: topology-guided and semantic-guided. The former maintains certain graph topological properties yet often underperforms on GNNs due to low integration with neural network training. The latter performs well at lower sparsity on GNNs but faces performance collapse at higher sparsity levels. With this in mind, we take the first step to propose a new research line and concept termed Graph Sparse Training (GST), which dynamically manipulates sparsity at the data level. Specifically, GST initially constructs a topology & semantic anchor at a low training cost, followed by performing dynamic sparse training to align the sparse graph with the anchor. We introduce the Equilibria Sparsification Principle to guide this process, effectively balancing the preservation of both topological and semantic information. Ultimately, GST produces a sparse graph with maximum topological integrity and no performance degradation. Extensive experiments on 6 datasets and 5 backbones showcase that GST (I) identifies subgraphs at higher graph sparsity levels (1.67%~15.85% $\uparrow$) than state-of-the-art sparsification methods, (II) preserves more key spectral properties, (III) achieves 1.27-3.42$\times$ speedup in GNN inference and (IV) successfully helps graph adversarial defense and graph lottery tickets.