Preserve, Promote, or Attack? GNN Explanation via Topology Perturbation

TL;DR

This paper introduces GNNViz, a versatile framework for interpreting GNNs through topology perturbations, enabling users to preserve, promote, or attack predictions, and uncover biases across various datasets.

Contribution

It develops a multi-purpose interpretation framework with an interactive visualization system for GNN explanations via topology perturbation, addressing multiple interpretability needs.

Findings

GNNViz helps non-experts explore GNN decision relationships.

It can manipulate GNN predictions for image classification.

The framework uncovers biases in social network data.

Abstract

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.