Existing methods for interpreting predictions from Graph Neural Networks (GNNs) have primarily focused on generating subgraphs that are especially relevant for a particular prediction. However, such methods do not provide a clear opportunity for recourse: given a prediction, we want to understand how the prediction can be changed in order to achieve a more desirable outcome. In this work, we propose a method for generating counterfac- tual (CF) explanations for GNNs: the mini- mal perturbation to the input (graph) data such that the prediction changes. Using only edge deletions, we find that our method, CF- GNNExplainer, can generate CF explana- tions for the majority of instances across three widely used datasets for GNN explanations, while removing less than 3 edges on average, with at least 94% accuracy. This indicates that CF-GNNExplainer primarily removes edges that are crucial for the original predic- tions, resulting in minimal CF explanations.
2022, The 25th International Conference on Artificial Intelligence and Statistics, Pages -
CF-GNNExplainer: Counterfactual Explanations for Graph Neural Networks (04b Atto di convegno in volume)
Lucic Ana, ter Hoeve Maartje, Tolomei Gabriele, de Rijke Maarten, Silvestri Fabrizio