mlsanityData to 14 Oct 2026

Reconsidering Faithfulness in Regular, Self-Explainable and Domain Invariant GNNs

arXiv 2406.15156 ↗First posted 21 Jun 2024cs.LG cs.AIcited by 16

Not on the map. Why?

Abstract (arXiv)

As Graph Neural Networks (GNNs) become more pervasive, it becomes paramount to build reliable tools for explaining their predictions. A core desideratum is that explanations are textitfaithful, ie that they portray an accurate picture of the GNN's reasoning process. However, a number of different faithfulness metrics exist, begging the question of what is faithfulness exactly and how to achieve it. We make three key contributions. We begin by showing that textitexisting metrics are not interchangeable -- ie explanations attaining high faithfulness according to one metric may be unfaithful according to others -- and can systematically ignore important properties of explanations. We proceed to show that, surprisingly, textitoptimizing for faithfulness is not always a sensible design goal. Specifically, we prove that for injective regular GNN architectures, perfectly faithful explanations are completely uninformative. This does not apply to modular GNNs, such as self-explainable and domain-invariant architectures, prompting us to study the relationship between architectural choices and faithfulness. Finally, we show that textitfaithfulness is tightly linked to out-of-distribution generalization, in that simply ensuring that a GNN can correctly recognize the domain-invariant subgraph, as prescribed by the literature, does not guarantee that it is invariant unless this subgraph is also faithful.The code is publicly available on GitHub

Key facts. Cited by 16 papers in the map (data through 14 Oct 2026). Builds on 51 of its 73 references that are in the map. First posted on arXiv 21 Jun 2024.

Builds on

51 of 73 references.

References · 51

  1. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,361
  2. 2017Graph Attention Networkscited by 8,923
  3. 2018How Powerful are Graph Neural Networks?cited by 4,393
  4. 2016Categorical Reparameterization with Gumbel-Softmaxcited by 2,739
  5. 2020Open Graph Benchmark: Datasets for Machine Learning on Graphscited by 2,163
  6. 2021Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gaugescited by 1,167
  7. 2017MoleculeNet: A Benchmark for Molecular Machine Learningcited by 1,051
  8. 2020Concept Bottleneck Modelscited by 1,032
  9. 2016Geometric deep learning on graphs and manifolds using mixture model CNNscited by 468
  10. 2020Explainability in Graph Neural Networks: A Taxonomic Surveycited by 384
  11. 2022Discovering Invariant Rationales for Graph Neural Networkscited by 188
  12. 2022Interpretable and Generalizable Graph Learning via Stochastic Attention Mechanismcited by 171
38 more
  1. 2022Learning Causally Invariant Representations for Out-of-Distribution Generalization on Graphscited by 144
  2. 2022GOOD: A Graph Out-of-Distribution Benchmarkcited by 141
  3. 2020Graph Information Bottleneck for Subgraph Recognitioncited by 119
  4. 2020A survey on domain adaptation theory: learning bounds and theoretical guaranteescited by 109
  5. 2018Robustly Disentangled Causal Mechanisms: Validating Deep Representations for Interventional Robustnesscited by 108
  6. 2019A Gentle Introduction to Deep Learning for Graphscited by 108
  7. 2021Causal Attention for Interpretable and Generalizable Graph Classificationcited by 108
  8. 2021ProtGNN: Towards Self-Explaining Graph Neural Networkscited by 96
  9. 2022Evaluating Explainability for Graph Neural Networkscited by 87
  10. 2022Debiasing Graph Neural Networks via Learning Disentangled Causal Substructurecited by 83
  11. 2022Learning and Evaluating Graph Neural Network Explanations based on Counterfactual and Factual Reasoningcited by 79
  12. 2021Towards Self-Explainable Graph Neural Networkcited by 75
  13. 2022GlanceNets: Interpretabile, Leak-proof Concept-based Modelscited by 73
  14. 2022Global Explainability of GNNs via Logic Combination of Learned Conceptscited by 65
  15. 2022GraphFramEx: Towards Systematic Evaluation of Explainability Methods for Graph Neural Networkscited by 58
  16. 2022Post-Hoc Explanations Fail to Achieve their Purpose in Adversarial Contextscited by 53
  17. 2022Leveraging Explanations in Interactive Machine Learning: An Overviewcited by 53
  18. 2021Local Explanations via Necessity and Sufficiency: Unifying Theory and Practicecited by 45
  19. 2021Improving Subgraph Recognition with Variational Graph Information Bottleneckcited by 45
  20. 2022Explaining the Explainers in Graph Neural Networks: a Comparative Studycited by 35
  21. 2023The expressive power of pooling in Graph Neural Networkscited by 35
  22. 2022Causal Explanations and XAIcited by 30
  23. 2023Joint Learning of Label and Environment Causal Independence for Graph Out-of-Distribution Generalizationcited by 29
  24. 2023A Survey on Explainability of Graph Neural Networkscited by 29
  25. 2023Learning on Graphs with Out-of-Distribution Nodescited by 29
  26. 2023Adversarial Training for Graph Neural Networks: Pitfalls, Solutions, and New Directionscited by 28
  27. 2022Interpretable Geometric Deep Learning via Learnable Randomness Injectioncited by 27
  28. 2022Encoding Concepts in Graph Neural Networkscited by 18
  29. 2023Towards Robust Fidelity for Evaluating Explainability of Graph Neural Networkscited by 17
  30. 2024VN-EGNN: E(3)-Equivariant Graph Neural Networks with Virtual Nodes Enhance Protein Binding Site Identificationcited by 15
  31. 2022Towards Prototype-Based Self-Explainable Graph Neural Networkcited by 14
  32. 2020Graph Neural Networks Including Sparse Interpretabilitycited by 13
  33. 2022Towards Faithful and Consistent Explanations for Graph Neural Networkscited by 13
  34. 2023GInX-Eval: Towards In-Distribution Evaluation of Graph Neural Network Explanationscited by 9
  35. 2022L2XGNN: Learning to Explain Graph Neural Networkscited by 7
  36. 2022Towards Training GNNs using Explanation Directed Message Passingcited by 7
  37. 2023How Faithful are Self-Explainable GNNs?cited by 5
  38. 2023Robust Ante-hoc Graph Explainer using Bilevel Optimizationcited by 4

50 most cited of 51 All 51 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 3; 2025: 5; 2026: 8.

16 papers.

Sort:

By area:

  1. 2025Beyond Topological Self-Explainable GNNs: A Formal Explainability Perspectivecited by 11
  2. 2024Benchmarking XAI Explanations with Human-Aligned Evaluationscited by 8
  3. 2024A Self-Explainable Heterogeneous GNN for Relational Deep Learningcited by 7
  4. 2026GNN Explanations that do not Explain and How to find Themcited by 5
  5. 2025Boosting Team Modeling through Tempo-Relational Representation Learningcited by 3
  6. 2024Interpreting Temporal Graph Neural Networks with Koopman Theorycited by 2
  7. 2025Knowing What You Cannot Explain: Learning to Reject Low-Quality Explanationscited by 2
  8. 2025Attribution Explanations for Deep Neural Networks: A Theoretical Perspectivecited by 2
  9. 2025Database Views as Explanations for Relational Deep Learningcited by 1
  10. 2026A Generalized Tikhonov Layer for Interpretable-by-design Graph Neural Networkscited by 1
  11. 2026Why Self-Inconsistency Arises in GNN Explanations and How to Exploit Itcited by 0
  12. 2026Automorphism-Induced Non-Canonicity in Top-k Explanations of Graph Neural Networkscited by 0
4 more
  1. 2026Overcoming Shortcut Learning in Graph Neural Networks through Active Explanation Guidancecited by 0
  2. 2026From Multimodal Observation to Interpretable Suggestions: Counterfactual Time-Expanded Relational Modeling of Surgical Teamscited by 0
  3. 2026Representable but Unlearned: Encoding Rank and the Interaction-Prediction Floorcited by 0
  4. 2026Beyond Trained Models: Compiling GNNs for a Sound Explainer Benchmarkcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024GNNAnatomy: Rethinking Model-Level Explanations for Graph Neural Networks
  2. 2025Forget and Explain: Transparent Verification of GNN Unlearning
  3. 2023Distill n' Explain: explaining graph neural networks using simple surrogates
  4. 2024Systematic Relational Reasoning With Epistemic Graph Neural Networks
  5. 2026Revealing Combinatorial Reasoning of GNNs via Graph Concept Bottleneck Layer
  6. 2026When the Tool Decides: LLM Agents Defer Blindly to Graph Neural Network Tools, and Stronger Backbones Defer More
  7. 2023Benchmarking Faithfulness: Towards Accurate Natural Language Explanations in Vision-Language Tasks
  8. 2022Toward Multiple Specialty Learners for Explaining GNNs via Online Knowledge Distillation