BatchNorm does not consistently change CNN representation sparsity, but it tends to produce purer class-based clusters than training without it.
Predicting the generalization gap in deep networks with margin distributions
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
-
Impact of Batch Normalization on Convolutional Network Representations
BatchNorm does not consistently change CNN representation sparsity, but it tends to produce purer class-based clusters than training without it.