GSTAM: Efficient Graph Distillation with Structural Attention-Matching

TL;DR

GSTAM introduces a novel graph distillation method that uses structural attention matching to efficiently condense large graph datasets into smaller, informative synthetic graphs, significantly improving performance in graph classification tasks.

Contribution

GSTAM is the first to leverage GNN attention maps for structural distillation, enhancing dataset condensation efficiency and effectiveness for graph classification.

Findings

Achieves 0.45% to 6.5% performance improvement over existing methods.

Effectively distills structural information into synthetic graphs.

Outperforms prior approaches in extreme condensation scenarios.

Abstract

Graph distillation has emerged as a solution for reducing large graph datasets to smaller, more manageable, and informative ones. Existing methods primarily target node classification, involve computationally intensive processes, and fail to capture the true distribution of the full graph dataset. To address these issues, we introduce Graph Distillation with Structural Attention Matching (GSTAM), a novel method for condensing graph classification datasets. GSTAM leverages the attention maps of GNNs to distill structural information from the original dataset into synthetic graphs. The structural attention-matching mechanism exploits the areas of the input graph that GNNs prioritize for classification, effectively distilling such information into the synthetic graphs and improving overall distillation performance. Comprehensive experiments demonstrate GSTAM's superiority over existing methods, achieving 0.45% to 6.5% better performance in extreme condensation ratios, highlighting its potential use in advancing distillation for graph classification tasks (Code available at https://github.com/arashrasti96/GSTAM).