mlsanityData to 7 Oct 2026

Concept Graph Convolutions: Message Passing in the Concept Space

arXiv 2604.20082 ↗First posted 22 Apr 2026cs.LGcited by 0

On the map: LLM reasoning → Knowledge- and tool-augmented reasoning

Abstract (arXiv)

The trust in the predictions of Graph Neural Networks is limited by their opaque reasoning process. Prior methods have tried to explain graph networks via concept-based explanations extracted from the latent representations obtained after message passing. However, these explanations fall short of explaining the message passing process itself. To this aim, we propose the Concept Graph Convolution, the first graph convolution designed to operate on node-level concepts for improved interpretability. The proposed convolutional layer performs message passing on a combination of raw and concept representations using structural and attention-based edge weights. We also propose a pure variant of the convolution, only operating in the concept space. Our results show that the Concept Graph Convolution allows to obtain competitive task accuracy, while enabling an increased insight into the evolution of concepts across convolutional steps.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Knowledge- and tool-augmented reasoning, one of 17 directions in LLM reasoning. Builds on 37 of its 48 references that are in the map. First posted on arXiv 22 Apr 2026.

Builds on

37 of 48 references.

References · 37

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  2. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,282
  3. 2017Graph Attention Networkscited by 8,854
  4. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,690
  5. 2017Inductive Representation Learning on Large Graphscited by 6,284
  6. 2018How Powerful are Graph Neural Networks?cited by 4,355
  7. 2015Learning Deep Features for Discriminative Localizationcited by 2,430
  8. 2016Convolutional Neural Networks on Graphs with Fast Localized Spectral Filteringcited by 2,085
  9. 2021Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gaugescited by 1,131
  10. 2019Attention is not Explanationcited by 916
  11. 2021How Attentive are Graph Attention Networks?cited by 818
  12. 2019GNNExplainer: Generating Explanations for Graph Neural Networkscited by 735
25 more
  1. 2020TUDataset: A collection of benchmark datasets for learning with graphscited by 718
  2. 2019Attention is not not Explanationcited by 629
  3. 2020Parameterized Explainer for Graph Neural Networkcited by 423
  4. 2019On Completeness-aware Concept-Based Explanations in Deep Neural Networkscited by 239
  5. 2020PGM-Explainer: Probabilistic Graphical Model Explanations for Graph Neural Networkscited by 210
  6. 2022Interpretable and Generalizable Graph Learning via Stochastic Attention Mechanismcited by 171
  7. 2015Stochastic modified equations and adaptive stochastic gradient algorithmscited by 141
  8. 2022How to Find Your Friendly Neighborhood: Graph Attention Design with Self-Supervisioncited by 102
  9. 2018Attention-based Graph Neural Network for Semi-supervised Learningcited by 100
  10. 2021ProtGNN: Towards Self-Explaining Graph Neural Networkscited by 96
  11. 2022Global Explainability of GNNs via Logic Combination of Learned Conceptscited by 64
  12. 2020Revisiting Over-smoothing in Deep GCNscited by 55
  13. 2020Why Attentions May Not Be Interpretable?cited by 41
  14. 2021GCExplainer: Human-in-the-Loop Concept-based Explanations for Graph Neural Networkscited by 35
  15. 2022Explaining the Explainers in Graph Neural Networks: a Comparative Studycited by 34
  16. 2022Global Concept-Based Interpretability for Graph Neural Networks via Neuron Analysiscited by 32
  17. 2024Causal Concept Graph Models: Beyond Causal Opacity in Deep Learningcited by 28
  18. 2025A Comprehensive Survey on Self-Interpretable Neural Networkscited by 18
  19. 2023Relational Concept Bottleneck Modelscited by 15
  20. 2022MEGAN: Multi-Explanation Graph Attention Networkcited by 7
  21. 2023Efficient GNN Explanation via Learning Removal-based Attributioncited by 6
  22. 2022Attention cannot be an Explanationcited by 4
  23. 2025Explaining GNN Explanations with Edge Gradientscited by 1
  24. 2026Beyond Message Passing: A Symbolic Alternative for Expressive and Interpretable Graph Learningcited by 1
  25. 2026End-to-end Feature Alignment: A Simple CNN with Intrinsic Class Attributioncited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Subgraph Concept Networks: Concept Levels in Graph Classification
  2. 2026Revealing Combinatorial Reasoning of GNNs via Graph Concept Bottleneck Layer
  3. 2026When Vision Meets Graphs: A Survey on Graph Reasoning and Learning
  4. 2026Rethinking Message Passing as Retrieval for Text-Attributed Graph Learning
  5. 2026TAG-DLM: Diffusion Language Models for Text-Attributed Graph Learning
  6. 2025Graph Learning
  7. 2024Towards Graph Foundation Models: Training on Knowledge Graphs Enables Transferability to General Graphs
  8. 2024Verbalized Graph Representation Learning: A Fully Interpretable Graph Model Based on Large Language Models Throughout the Entire Process