Unraveling Graph Counterfactual Explainability: from Theoretical Foundations to Technical Mastery (ECML-PKDD 2025)
Tutorial at the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD 2025), September 2025, Porto, Portugal.
Graph Neural Networks (GNNs) have proven highly effective in graph-related tasks, including Traffic Modeling, Learning Physical Simulations, Protein Modeling, and Large-scale Recommender Systems. Their ``black-box’’ nature, however, limits interpretability. This half-day tutorial covered Graph Counterfactual Explainability from theoretical foundations to technical mastery.
Speakers
- Bardh Prenkaj (TU Munich)
- Andrea D’Angelo (University of L’Aquila)
- Stratis Limnios
- Mario Alfonso Prado-Romero
- Giovanni Stilo (Luiss University of Rome)
Topics Covered
- Formal definitions and foundations of GCE
- Taxonomy and categorization of methods
- Evaluation metrics and benchmarking with GRETEL
- Hands-on exercises
- Latest advancements and research frontiers
Based on the ACM Computing Survey: A Survey on Graph Counterfactual Explanations: Definitions, Methods, Evaluation, and Research Challenges (ACM CSUR, 2024).