A One-step Pruning-recovery Framework for Acceleration of Convolutional Neural Networks

Dong Wang, Xiao Bai, Lei Zhou, Jun Zhou
2019 2019 IEEE 31st International Conference on Tools with Artificial Intelligence (ICTAI)  
Acceleration of convolutional neural network has received increasing attention during the past several years. Among various acceleration techniques, filter pruning has its inherent merit by effectively reducing the number of convolution filters. However, most filter pruning methods resort to tedious and time-consuming layer-by-layer pruningrecovery strategy to avoid a significant drop of accuracy. In this paper, we present an efficient filter pruning framework to solve this problem. Our method
more » ... roblem. Our method accelerates the network in one-step pruning-recovery manner with a novel optimization objective function, which achieves higher accuracy with much less cost compared with existing pruning methods. Furthermore, our method allows network compression with global filter pruning. Given a global pruning rate, it can adaptively determine the pruning rate for each single convolutional layer, while these rates are often set as hyper-parameters in previous approaches. Evaluated on VGG-16 and ResNet-50 using ImageNet, our approach outperforms several state-of-the-art methods with less accuracy drop under the same and even much fewer floatingpoint operations (FLOPs).
doi:10.1109/ictai.2019.00111 dblp:conf/ictai/WangBZZ19 fatcat:ml3pukmcyrdgllw4lmoipx4kdm