![[2602.16145] Investigating GNN Convergence on Large Randomly Generated Graphs with Realistic Node Feature Correlations](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16145] Investigating GNN Convergence on Large Randomly Generated Graphs with Realistic Node Feature Correlations
Summary
This paper investigates the convergence behavior of Graph Neural Networks (GNNs) on large random graphs with realistic node feature correlations, proposing a novel method for generating such graphs.
Why It Matters
Understanding GNN convergence is crucial for their application in real-world scenarios. This research addresses limitations in existing studies by incorporating realistic correlations between node features, potentially enhancing the applicability and expressiveness of GNNs in practical settings.
Key Takeaways
- The study introduces a novel method for generating random graphs with correlated node features.
- Incorporating realistic correlations can lead to divergent behavior in GNNs, suggesting greater expressiveness than previously thought.
- The findings challenge existing limitations of GNNs based on traditional random graph models.
- The research validates theoretical analysis through empirical testing on large random graphs.
- This work has implications for the design and application of GNNs in various real-world networks.
Computer Science > Machine Learning arXiv:2602.16145 (cs) [Submitted on 18 Feb 2026] Title:Investigating GNN Convergence on Large Randomly Generated Graphs with Realistic Node Feature Correlations Authors:Mohammed Zain Ali Ahmed View a PDF of the paper titled Investigating GNN Convergence on Large Randomly Generated Graphs with Realistic Node Feature Correlations, by Mohammed Zain Ali Ahmed View PDF HTML (experimental) Abstract:There are a number of existing studies analysing the convergence behaviour of graph neural networks on large random graphs. Unfortunately, the majority of these studies do not model correlations between node features, which would naturally exist in a variety of real-life networks. Consequently, the derived limitations of GNNs, resulting from such convergence behaviour, is not truly reflective of the expressive power of GNNs when applied to realistic graphs. In this paper, we will introduce a novel method to generate random graphs that have correlated node features. The node features will be sampled in such a manner to ensure correlation between neighbouring nodes. As motivation for our choice of sampling scheme, we will appeal to properties exhibited by real-life graphs, particularly properties that are captured by the Barabási-Albert model. A theoretical analysis will strongly indicate that convergence can be avoided in some cases, which we will empirically validate on large random graphs generated using our novel method. The observed divergent beh...
