search

liuzechun / MetaPruning

MetaPruning: Meta Learning for Automatic Neural Network Channel Pruning. In ICCV 2019.
MIT License
351 stars 74 forks source link

Why use meta learning to predict weights #18

Closed zeyu-liu closed 4 years ago

zeyu-liu commented 5 years ago

Why use an independent meta networks to generate the weights for each block, in stead of using the weights of the original networks and update them directly?

liuzechun commented 4 years ago

Because the accuracy is low due to the excessive weight sharing. You can refer to figure.8 in the paper.