Deep neural network compression through interpretability-based filter pruning

作者：

Highlights：

• Filters are visualized by the activation maximization to explain functions of filters.

• DNNs are compressed based on the visualization results.

• The redundant filters are measured based on the color and texture similarities.

• The repetitive and invalid filters can be pruned by optimization.

摘要

•Filters are visualized by the activation maximization to explain functions of filters.•DNNs are compressed based on the visualization results.•The redundant filters are measured based on the color and texture similarities.•The repetitive and invalid filters can be pruned by optimization.

论文关键词：Deep neural network (DNN),Convolutional neural network (CNN),Visualization,Compression

论文评审过程：Received 24 October 2020, Revised 6 April 2021, Accepted 19 May 2021, Available online 28 May 2021, Version of Record 12 June 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108056