Beyond Topological Self-Explainable GNNs: A Formal Explainability Perspective
This work addresses the need for more reliable and tractable explanations in graph neural networks, particularly for users in fields like social network analysis or bioinformatics, though it is incremental as it builds on existing SE-GNN frameworks.
The paper tackles the problem of understanding and improving the explanations from Self-Explainable Graph Neural Networks (SE-GNNs) by formalizing their limitations and proposing Dual-Channel GNNs that integrate a rule extractor, resulting in performance on par or better than SE-GNNs while recovering succinct rules.
Self-Explainable Graph Neural Networks (SE-GNNs) are popular explainable-by-design GNNs, but their explanations' properties and limitations are not well understood. Our first contribution fills this gap by formalizing the explanations extracted by some popular SE-GNNs, referred to as Minimal Explanations (MEs), and comparing them to established notions of explanations, namely Prime Implicant (PI) and faithful explanations. Our analysis reveals that MEs match PI explanations for a restricted but significant family of tasks. In general, however, they can be less informative than PI explanations and are surprisingly misaligned with widely accepted notions of faithfulness. Although faithful and PI explanations are informative, they are intractable to find and we show that they can be prohibitively large. Given these observations, a natural choice is to augment SE-GNNs with alternative modalities of explanations taking care of SE-GNNs' limitations. To this end, we propose Dual-Channel GNNs that integrate a white-box rule extractor and a standard SE-GNN, adaptively combining both channels. Our experiments show that even a simple instantiation of Dual-Channel GNNs can recover succinct rules and perform on par or better than widely used SE-GNNs.