About this document

Supervised Hybrid Explainer for GNNs by f20220183 is a document available to read on EtoBox.

This document presents a project focused on enhancing post-hoc explainability techniques for graph neural networks (GNNs) using a hybrid connectivity module applied to the BA-Shapes dataset. The proposed Supervised Hybrid Connectivity Explainer outperforms the baseline PGExplainer by producing coherent, connected subgraphs while maintaining fidelity and sparsity. Evaluation metrics demonstrate significant improvements in explanation quality, with future work aimed at extending these methods to real-world da

Author

f20220183

Language

EN