MetaPruning: Meta Learning for Automatic Neural Network Channel Pruning

Zechun Liu, Haoyuan Mu, Xiangyu Zhang, Zichao Guo, Xin Yang, Kwang-Ting Cheng, Jian Sun; The IEEE International Conference on Computer Vision (ICCV), 2019, pp. 3296-3305

Abstract


In this paper, we propose a novel meta learning approach for automatic channel pruning of very deep neural networks. We first train a PruningNet, a kind of meta network, which is able to generate weight parameters for any pruned structure given the target network. We use a simple stochastic structure sampling method for training the PruningNet. Then, we apply an evolutionary procedure to search for good-performing pruned networks. The search is highly efficient because the weights are directly generated by the trained PruningNet and we do not need any finetuning at search time. With a single PruningNet trained for the target network, we can search for various Pruned Networks under different constraints with little human participation. Compared to the state-of-the-art pruning methods, we have demonstrated superior performances on MobileNet V1/V2 and ResNet. Codes are available on https://github.com/liuzechun/MetaPruning.

Related Material


[pdf]
[bibtex]
@InProceedings{Liu_2019_ICCV,
author = {Liu, Zechun and Mu, Haoyuan and Zhang, Xiangyu and Guo, Zichao and Yang, Xin and Cheng, Kwang-Ting and Sun, Jian},
title = {MetaPruning: Meta Learning for Automatic Neural Network Channel Pruning},
booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
month = {October},
year = {2019}
}