IEICE Transactions on Information and Systems
Online ISSN : 1745-1361
Print ISSN : 0916-8532
Regular Section
Loss-Driven Channel Pruning of Convolutional Neural Networks
Xin LONGXiangrong ZENGChen CHENHuaxin XIAOMaojun ZHANG
Author information
JOURNAL FREE ACCESS

2020 Volume E103.D Issue 5 Pages 1190-1194

Details
Abstract

The increase in computation cost and storage of convolutional neural networks (CNNs) severely hinders their applications on limited-resources devices in recent years. As a result, there is impending necessity to accelerate the networks by certain methods. In this paper, we propose a loss-driven method to prune redundant channels of CNNs. It identifies unimportant channels by using Taylor expansion technique regarding to scaling and shifting factors, and prunes those channels by fixed percentile threshold. By doing so, we obtain a compact network with less parameters and FLOPs consumption. In experimental section, we evaluate the proposed method in CIFAR datasets with several popular networks, including VGG-19, DenseNet-40 and ResNet-164, and experimental results demonstrate the proposed method is able to prune over 70% channels and parameters with no performance loss. Moreover, iterative pruning could be used to obtain more compact network.

Content from these authors
© 2020 The Institute of Electronics, Information and Communication Engineers
Previous article Next article
feedback
Top