Estimating Subgraph Importance with Structural Prior Domain Knowledge

TL;DR

This paper introduces a novel subgraph importance estimation method for pretrained GNNs that leverages domain knowledge and outperforms existing baselines without needing target labels.

Contribution

It formulates subgraph importance as a Group Lasso regression problem, independent of GNN output layers, and extends to node importance identification.

Findings

Outperforms existing baselines in real-world datasets

Effectively leverages prior domain knowledge

Does not require ground-truth labels

Abstract

We propose a subgraph importance estimation method for pretrained Graph Neural Networks (GNNs) on graph-level tasks, formulated as a linear Group Lasso regression problem in the embedding space. Our method effectively leverages prior domain knowledge of graph substructures, while remaining independent of the specific form of the output layer or readout function used in the GNN architecture, and it does not require access to ground-truth target labels. Experiments on real-world graph datasets demonstrate that our method consistently outperforms existing baselines in subgraph importance estimation. Furthermore, we extend our method to identify important nodes within the graph.