ATEX-CF : Attack-Informed Counterfactual Explanations for Graph Neural Networks
Yu Zhang Aalborg University Aalborg, Denmark, 9220 [email protected] Sean Bin Yang Aalborg University Aalborg, Denmark, 9220 [email protected] Arijit Khan Bowling Green State University, USA, 43403 Aalborg University, Denmark, 9220 [email protected] Cuneyt Gurcan Akcora University of Central Florida Or · 2026
Read ATEX-CF : Attack-Informed Counterfactual Explanations for Graph Neural Networks by Yu Zhang Aalborg University Aalborg, Denmark, 9220 [email protected] Sean Bin Yang Aalborg University Aalborg, Denmark, 9220 [email protected] Arijit Khan Bowling Green State University, USA, 43403 Aalborg University, Denmark, 9220 [email protected] Cuneyt Gurcan Akcora University of Central Florida Or (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-2602-06240.txt.
Description
Explaining a graph neural network can mean asking what tiny change would flip its answer for one node. Yu Zhang and 3 co-authors present ATEX-CF, a framework for such counterfactual explanations that borrows tricks from adversarial attacks. Attacks and explanations share the goal of flipping a prediction, but attacks tend to add edges while explanations tend to remove them.
ATEX-CF uses both additions and deletions, guided by theory about which moves matter most. It balances three aims at once: staying faithful to the model, keeping changes few, and remaining plausible, all within a fixed change budget. On synthetic and real node classification data, it produced faithful, compact, and realistic explanations.
Catalog record
| CREATOR | Yu Zhang Aalborg University Aalborg, Denmark, 9220 [email protected] Sean Bin Yang Aalborg University Aalborg, Denmark, 9220 [email protected] Arijit Khan Bowling Green State University, USA, 43403 Aalborg University, Denmark, 9220 [email protected] Cuneyt Gurcan Akcora University of Central Florida Or |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 101 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-06240