CSBA reduces inter-class gradient interference via branch-specific attention in multi-branch CNNs, raising minority-class F1 from 0.261 to 0.522 on imbalanced damage data and Macro-F1 from 0.595 to 0.655 on CIFAR-10-LT.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Class-Specific Branch Attention for Mitigating Gradient Interference under Class Imbalance
CSBA reduces inter-class gradient interference via branch-specific attention in multi-branch CNNs, raising minority-class F1 from 0.261 to 0.522 on imbalanced damage data and Macro-F1 from 0.595 to 0.655 on CIFAR-10-LT.