The paper presents a comparative analysis of six state‑of‑the‑art counterfactual explainers for graph neural networks, focusing on methods that can both add and remove edges to alter model predictions. It evaluates these explainers across diverse real‑world and synthetic datasets, covering binary and multi‑class graph and node classification tasks, using a range of quantitative and qualitative metrics. The study highlights the trade‑offs between explanation size, coverage, and quality, aiming to pinpoint each method’s strengths and weaknesses to inform future research.
arXiv:2607. 04600v1 Announce Type: new Abstract: Graph eXplainable AI (G-XAI) is increasingly important for making Graph Neural Networks interpretable and accountable.
Graph eXplainable AI (G-XAI) is increasingly important for making Graph Neural Networks interpretable and accountable. While a growing number of explainers are available, choosing the right method and assessing the trustworthiness of its outputs remains unclear.
arXiv:2606. 05756v1 Announce Type: new Abstract: Graph Neural Networks (GNNs) have demonstrated remarkable performance across a range of applications involving graph-structured data, particularly in high-stakes domains.
Graph Machine Learning as a Service platforms now offer explainability interfaces to satisfy regulatory transparency, but this transparency can be exploited. The paper introduces a novel model extraction attack for graph classification that operates under strict black‑box constraints, using only discrete class labels and binary explanation masks. The method guides Monte Carlo edge sensitivity estimation toward decision boundaries with Hoeffding guarantees and narrows the search space using explanation subgraphs, outperforming comparable baselines on benchmark datasets.
arXiv:2607. 21381v1 Announce Type: new Abstract: Instance-level explanations aim to reveal the rationale behind a model's decisions for a specific graph.
Instance-level explanations aim to reveal the rationale behind a model's decisions for a specific graph. Previous methods explain graph neural networks (GNNs) by selecting important edges to induce subgraphs, where edge importance is assessed by perturbing each edge and observing changes in the model predictions.
arXiv:2606. 03495v1 Announce Type: new Abstract: Heterogeneous graph neural networks (HGNNs) have demonstrated remarkable performance in modeling complex relational data, however their interpretability in high-stakes applications remains a critical challenge.
arXiv:2608.23835v1 Announce Type: new Abstract: Real-world decision-making in public health and social science can greatly benefit from predictive models, yet translating predictions into effective i...
arXiv:2608. 12083v1 Announce Type: cross Abstract: Graph Neural Networks (GNNs) achieve strong predictive performance on graph-structured data across domains such as chemistry, biology, and network analysis, yet they provide no intrinsic explanation of their predictions.
arXiv:2609.37650v1 Announce Type: new Abstract: Counterfactual explanations of graph neural networks identify edge deletions that flip a prediction. On heterogeneous graphs, however, existing methods...
arXiv:2608. 14121v1 Announce Type: cross Abstract: Graph Neural Networks (GNNs) can solve prediction tasks by unintentionally exploiting shortcuts---that is, edges, nodes, and features that correlate with but are not causal for the prediction---which compromise their reliability in out-of-distribution tasks.
The paper introduces SEMGNN, an end‑to‑end self‑explainable multi‑label graph neural network that simultaneously classifies nodes and identifies edges contributing to each predicted label. Unlike post‑hoc explainers, SEMGNN jointly learns a predictor and a sparse edge‑mask explainer, leveraging label‑label correlations to improve classification and generate distinct, coherent explanations for each label. Experiments on synthetic and real‑world networks in social, entertainment, and life‑science domains demonstrate competitive predictive performance and more faithful, compact label‑conditioned explanations.