search

XiangqianMa / GitTalk

my gittalk.
0 stars 0 forks source link

模型剪枝-基于损失变化的泰勒展开近似的卷积核重要性度量准则 | MXQ #13

Open XiangqianMa opened 4 years ago

XiangqianMa commented 4 years ago

http://activepony.com/2019/09/12/lun-wen-yue-du/pruning-convolutional-neural-networks-for-resource-efficient-inference/

Pruning Convolutional Neural Networks For Resource Efficient Inference 作者：Pavlo Molchanov, Stephen Tyree, Tero Karras, Timo Aila, Jan Kautz 机构：Nvidia 简介在该方法中，使用迭代步骤进行剪枝：在基于准则的贪婪剪枝和使用反向传播进行微调两个步骤之间进行切