MPS-GNN learns predictive meta-paths in relational databases using aggregate statistics over their occurrences, not just existence, and outperforms prior heterogeneous GNNs in experiments.
Reconsidering faithfulness in regular, self-explainable and domain invariant GNN s
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Self-Explainable Heterogeneous GNN for Relational Deep Learning
MPS-GNN learns predictive meta-paths in relational databases using aggregate statistics over their occurrences, not just existence, and outperforms prior heterogeneous GNNs in experiments.