SuperMeshNet achieves lower RMSE than fully supervised baselines while using 90% less high-resolution training data by jointly training two complementary MPNN models on paired and unpaired low-resolution meshes.
The interpolated value at the target nodey 0 is then obtained as y0 = Pk i=1 wiyi Pk i=1 wi
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Semi-Supervised Neural Super-Resolution for Mesh-Based Simulations
SuperMeshNet achieves lower RMSE than fully supervised baselines while using 90% less high-resolution training data by jointly training two complementary MPNN models on paired and unpaired low-resolution meshes.