Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Invariant-Based Weight Sharing for Message Passing

About

Message-passing neural networks (MPNNs) are a powerful framework for learning representations of graph-structured domains. However, weights in MPNNs act on features only, limiting their ability to capture structural patterns. We introduce a novel structure-aware weight sharing principle that explicitly incorporates information inherent to the graph structure. Weights are indexed directly by user-chosen graph invariants, i.e., functions preserved under node permutations, enabling systematic reuse across structurally equivalent subgraphs. We present ShareGNNs, which instantiate this principle within a simple encoder-decoder architecture, resulting in an MPNN with learnable adjacency and transformer-like connectivity. We show that their expressivity is at least as strong as the discriminative power of the chosen invariants, providing explicit control over the model complexity. Experiments on synthetic and real-world data, as well as subgraph counting tasks, demonstrate consistent improvements over standard MPNNs, competitive expressivity beyond the 1-WL test, and scalability to large datasets.

Florian Seiffarth• 2026

Related benchmarks

TaskDatasetResultRank
Graph ClassificationNCI1
Accuracy86.1
658
Graph ClassificationIMDB-M
Accuracy53.1
425
Graph ClassificationNCI109
Accuracy86.8
267
Graph ClassificationIMDB-B
Mean Accuracy77.7
159
Graph RegressionZINC-12K
MAE0.107
57
Graph ClassificationNCI1 Fair setup
Accuracy85.4
8
Graph ClassificationIMDB-Multi Fair setup
Accuracy51.6
8
Graph ClassificationIMDB-Binary Fair setup
Accuracy75.9
8
Graph ClassificationNCI109 Fair setup
Accuracy85.4
6
Graph ClassificationMutagenicity Fair setup
Accuracy83.3
6
Showing 10 of 18 rows

Other info

Follow for update