The authors propose WILT, a weighted Weisfeiler-Leman tree distance that can be learned to approximate MPNN embedding distances, and show that MPNNs align with task-relevant functional distances rather than structural ones.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
WILTing Trees: Interpreting the Distance Between MPNN Embeddings
The authors propose WILT, a weighted Weisfeiler-Leman tree distance that can be learned to approximate MPNN embedding distances, and show that MPNNs align with task-relevant functional distances rather than structural ones.