Network with Sub-Networks trains a base network and its layer-pruned sub-networks together by copying parameters and averaging gradients, so the base model can shed layers at runtime with little accuracy loss.
Dropout: a simple way to prevent neural networks from overfitting
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Network with Sub-Networks
Network with Sub-Networks trains a base network and its layer-pruned sub-networks together by copying parameters and averaging gradients, so the base model can shed layers at runtime with little accuracy loss.