A-Channel-level-Pruning-Strategy-for-Convolutional-Neural-Networks

This repo shows some code about the pruning strategy in the paper "A Channel-level Pruning Strategy for Convolutional Neural Networks".

• The file "prune.ipynb" contains some code about the pruning process. In this file we use SE-BN-Inception for the example. You can download the network structure file and model parameters file in https://github.com/hujie-frank/SENet. We use Caffe to do this work. This is not a complete code but should be enough to show the pruning process.

Highlights

network pruning based on attention.

Citations

@inproceedings{song2018channel,
  title={A Channel-level Pruning Strategy for Convolutional Layers in CNNs},
  author={Song, Fangzhou and Wang, Ying and Guo, Yao and Zhu, Chuang and Liu, Jun and Jin, Mulan},
  booktitle={2018 International Conference on Network Infrastructure and Digital Content (IC-NIDC)},
  pages={135--139},
  year={2018},
  organization={IEEE}
}