Abstract:Prior work on node classification has shown that Graph Neural Networks (GNNs) can learn representations that transfer across graphs, when underlying graph properties are shared. For a fixed graph, one would then expect GNNs trained for link prediction to learn a representation consistent with that learnt for node classification. We show this intuition does not hold in the general case. Instead, we find popular link prediction models can learn a trivial mini-batch dependent heuristic, enabled by batch-normalisation layers, to solve the edge classification task. When correcting for this, we observe increased alignment of the network representation with node-class relevant features, suggesting the network has learnt a graph representation that better aligns with the underlying graph's properties. Our findings suggest that standard link prediction training may be leading us to overestimate link predictors' ability to learn a generalised representation of a graph that is consistent across tasks.
| Comments: | Accepted at GCLR 2026: the 5th Workshop on Graphs and more Complex Structures For Learning and Reasoning, colocated with AAAI 2026 |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2604.25978 [cs.LG] |
| (or arXiv:2604.25978v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2604.25978 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Kieran Maguire [view email]
[v1]
Tue, 28 Apr 2026 14:03:50 UTC (1,360 KB)