SpLPG reduces the communication cost of distributed GNN link prediction training by up to 80% by using sparsified remote subgraphs for negative sampling while keeping full neighbor information locally.
Link prediction in complex networks: A survey,
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
-
Demystifying Distributed Training of Graph Neural Networks for Link Prediction
SpLPG reduces the communication cost of distributed GNN link prediction training by up to 80% by using sparsified remote subgraphs for negative sampling while keeping full neighbor information locally.