A bounded-L1 regularizer combined with exponential gating layers prunes neural network channels to exactly zero during training, compressing standard models by 30 to 75 percent with little accuracy loss.
In: Computer Vision (ICCV), 2017 IEEE International Conference on
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Group Pruning using a Bounded-Lp norm for Group Gating and Regularization
A bounded-L1 regularizer combined with exponential gating layers prunes neural network channels to exactly zero during training, compressing standard models by 30 to 75 percent with little accuracy loss.