Beyond Edge Deletion: A Comprehensive Approach to Counterfactual Explanation in Graph Neural Networks DOI: XXXXXXX.XXXXXXX Conference: KDD’26; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06

Matteo De Sanctis * , Riccardo De Sanctis, Stefano Faralli, Paola Velardi, Bardh Prenkaj · 2026

Read Beyond Edge Deletion: A Comprehensive Approach to Counterfactual Explanation in Graph Neural Networks DOI: XXXXXXX.XXXXXXX Conference: KDD’26; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 by Matteo De Sanctis * , Riccardo De Sanctis, Stefano Faralli, Paola Velardi, Bardh Prenkaj (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-04209.txt.

Description

De Sanctis and others go beyond edge deletion for counterfactual explanation in graph neural networks. Counterfactual explanations, minimal changes flipping a model's output, give a clear window onto network conduct. The authors present XPlore, broadening the counterfactual search space through gradient-guided changes to adjacency and node feature matrices, done together within one gradient-based method. Tests on 13 real-world plus 5 synthetic benchmarks show large validity and fidelity gains over leading rivals while keeping runtime competitive.

Catalog record

CREATORMatteo De Sanctis * , Riccardo De Sanctis, Stefano Faralli, Paola Velardi, Bardh Prenkaj
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE82128
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.04209

Browse the full library · About the Tower of Babel