期刊
IEEE TRANSACTIONS ON CYBERNETICS
卷 53, 期 5, 页码 3165-3175出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TCYB.2021.3124284
关键词
Frequency-domain analysis; Discrete cosine transforms; Convolution; Visual systems; Quantization (signal); Transforms; Redundancy; Absolutely unimportant filter pruning; frequency-domain transform; learned representation median (RM)
This article proposes a novel filter pruning method for deep learning networks by calculating the learned representation median in the frequency domain. The method emphasizes the removal of absolutely unimportant filters and has been shown to outperform existing pruning methods in terms of accuracy and FLOPs reduction.
In this article, we propose a novel filter pruning method for deep learning networks by calculating the learned representation median (RM) in frequency domain (LRMF). In contrast to the existing filter pruning methods that remove relatively unimportant filters in the spatial domain, our newly proposed approach emphasizes the removal of absolutely unimportant filters in the frequency domain. Through extensive experiments, we observed that the criterion for ``relative unimportance'' cannot be generalized well and that the discrete cosine transform (DCT) domain can eliminate redundancy and emphasize low-frequency representation, which is consistent with the human visual system. Based on these important observations, our LRMF calculates the learned RM in the frequency domain and removes its corresponding filter, since it is absolutely unimportant at each layer. Thanks to this, the time-consuming fine-tuning process is not required in LRMF. The results show that LRMF outperforms state-of-the-art pruning methods. For example, with ResNet110 on CIFAR-10, it achieves a 52.3% FLOPs reduction with an improvement of 0.04% in Top-1 accuracy. With VGG16 on CIFAR-100, it reduces FLOPs by 35.9% while increasing accuracy by 0.5%. On ImageNet, ResNet18 and ResNet50 are accelerated by 53.3% and 52.7% with only 1.76% and 0.8% accuracy loss, respectively. The code is based on PyTorch and is available at https://github.com/zhangxin-xd/LRMF.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据