Temporal Sheaf Neural Networks with Dynamic Orthogonal Transport
About
We introduce Temporal Sheaf Neural Networks (TSNN), a temporal link prediction framework that equips each node with a time-varying orthogonal frame and compares node states only after explicit transport between local coordinate systems. In contrast to existing continuous-time graph models that operate in a shared global embedding space, TSNN models node-specific and evolving interaction semantics through dynamic local frames. The model parameterizes per-node frames via efficient low-rank Householder products, preserves stored hidden states exactly under frame updates, and uses a geometric-residual decoder that anchors predictions on transported distances while learning residual corrections. All computations are strictly causal and use only the pre-event history. We show that the symmetric degree-normalized sheaf Laplacian is orthogonally similar to the symmetric normalized graph Laplacian, with the random-walk normalized form similar in the corresponding degree metric; the full-active, feature-scaled diffusion used by TSNN is exactly a metric-gradient step on the combinatorial sheaf Dirichlet energy, with a degree-free monotone-descent and non-expansiveness guarantee. Frame drift perturbs updates only linearly. Across TGB v2 link-prediction and temporal-heterogeneous leaderboards, together with the DGB benchmark suite, TSNN matches or surpasses the strongest prior methods on most benchmarks, with the largest improvements on graphs exhibiting strong node-role heterogeneity. Ablations confirm the distinct benefit of dynamic frames, orthogonal transport, and geometric-residual decoding.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Future Link Prediction | SocialEvolution inductive | AP97.23 | 89 | |
| Inductive dynamic link prediction | MOOC (inductive) | AP99.43 | 42 | |
| Temporal Link Prediction | Wikipedia (inductive) | AUC-ROC99.1 | 42 | |
| Temporal Link Prediction | Reddit (inductive) | AUC-ROC99.32 | 42 | |
| Dynamic Link Prediction | tgbl-wiki (test) | MRR0.8282 | 19 | |
| Link-property prediction | tgbl-review v2 (test) | MRR50.28 | 18 | |
| Temporal Link Prediction | Wikipedia DGB | AP99.41 | 13 | |
| Temporal Link Prediction | Reddit DGB | AP99.66 | 13 | |
| Temporal Link Prediction | MOOC DGB | AP99.36 | 13 | |
| Temporal Link Prediction | LastFM DGB | Average Precision (AP)94.51 | 13 |