Research map / Graph learning / GNN architectures
Graph Neural Networks Can (Often) Count Substructures
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Shows that message passing GNNs can often count substructures despite known expressivity limits, by analyzing when substructure counts are determinable from node or graph representations.
Problem
Message passing GNNs have limited expressive power and are commonly thought unable to detect or count arbitrary graph substructures (subgraph isomorphism), which matters for many graph tasks.
Approach
Observes that GNNs are in fact often able to count substructures in practice, challenging the common belief that their expressivity limits preclude substructure counting.
Surprising findings
- Message passing GNNs are often able to count substructures despite being widely believed unable to detect or count arbitrary substructures.
Closest papers by meaning
- Homomorphism Counts for Graph Neural Networks: All About That Basis (ICML 2024 · similarity 0.93)
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · similarity 0.90)
- Deep Homomorphism Networks (NeurIPS 2024 · similarity 0.90)
- Beyond Message Passing: Neural Graph Pattern Machine (ICML 2025 · similarity 0.89)
- On the Power of the Weisfeiler-Leman Test for Graph Motif Parameters (ICLR 2024 · similarity 0.89)
- Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness (ICLR 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.