Tag
This paper introduces a logic-based framework for extracting global logical rules from Simple Graph Convolution networks using minimal abductive explanations, aiming to provide compact and faithful explanations for node classification in graph neural networks.