IceBerg uses pseudo-label-based double balancing and decoupled propagation to improve graph neural networks on class-imbalanced and few-shot node classification.
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
-
IceBerg: Debiased Self-Training for Class-Imbalanced Node Classification
IceBerg uses pseudo-label-based double balancing and decoupled propagation to improve graph neural networks on class-imbalanced and few-shot node classification.