The authors propose DCN and PDCN, new GNN architectures using causal graph filters for convolutional learning on DAGs, with established equivariance properties and competitive empirical performance.
A reduction of a graph to a canonical form and an algebra arising during this reduction,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SP 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Directed Acyclic Graph Convolutional Networks
The authors propose DCN and PDCN, new GNN architectures using causal graph filters for convolutional learning on DAGs, with established equivariance properties and competitive empirical performance.