Message-passing selection: Towards interpretable GNNs for graph classification
This addresses the need for interpretable GNNs in graph analysis, though it appears incremental as it builds on existing GNN baselines.
The paper tackled the problem of making GNNs interpretable for graph classification by introducing MSInterpreter, a plug-and-play method that selects critical message-passing paths to enable self-explanation, and demonstrated its effectiveness on benchmarks.
In this paper, we strive to develop an interpretable GNNs' inference paradigm, termed MSInterpreter, which can serve as a plug-and-play scheme readily applicable to various GNNs' baselines. Unlike the most existing explanation methods, MSInterpreter provides a Message-passing Selection scheme(MSScheme) to select the critical paths for GNNs' message aggregations, which aims at reaching the self-explaination instead of post-hoc explanations. In detail, the elaborate MSScheme is designed to calculate weight factors of message aggregation paths by considering the vanilla structure and node embedding components, where the structure base aims at weight factors among node-induced substructures; on the other hand, the node embedding base focuses on weight factors via node embeddings obtained by one-layer GNN.Finally, we demonstrate the effectiveness of our approach on graph classification benchmarks.