Picking Winning Tickets Before Training by Preserving Gradient Flow

ICLR, 2020.

Cited by: 3|Views73
EI
Weibo:
We propose Gradient Signal Preservation, a pruning criterion motivated by preserving the gradient flow through the network after pruning

Abstract:

Overparameterization has been shown to benefit both the optimization and generalization of neural networks, but large networks are resource hungry at both training and test time. Network pruning can reduce test-time resource requirements, but is typically applied to trained networks and therefore cannot avoid the expensive training proce...More

Code:

Data:

Your rating :
0

 

Tags
Comments