Preserve, Promote, or Attack? GNN Explanation via Topology Perturbation
This work addresses the need for versatile GNN explanation tools for users like non-experts and researchers, though it appears incremental by extending single-use explanations to multiple purposes.
The paper tackles the problem of explaining graph neural network (GNN) predictions by developing a multi-purpose interpretation framework that uses topology perturbations to preserve, promote, or attack predictions, with case studies showing effectiveness in tasks like image classification on MS-COCO, uncovering biases on Pokec, and comparing to baselines on synthetic data.
Prior works on formalizing explanations of a graph neural network (GNN) focus on a single use case - to preserve the prediction results through identifying important edges and nodes. In this paper, we develop a multi-purpose interpretation framework by acquiring a mask that indicates topology perturbations of the input graphs. We pack the framework into an interactive visualization system (GNNViz) which can fulfill multiple purposes: Preserve,Promote, or Attack GNN's predictions. We illustrate our approach's novelty and effectiveness with three case studies: First, GNNViz can assist non expert users to easily explore the relationship between graph topology and GNN's decision (Preserve), or to manipulate the prediction (Promote or Attack) for an image classification task on MS-COCO; Second, on the Pokec social network dataset, our framework can uncover unfairness and demographic biases; Lastly, it compares with state-of-the-art GNN explainer baseline on a synthetic dataset.