arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.09736cs.CVcs.AI
收藏

压缩CNN的动态结构修剪

Dynamic Structure Pruning for Compressing CNNs

Korea University(高丽大学)

Jun-Hyung Park, Yeachan Kim, Junho Kim, Joon-Young Choi, SangKeun Lee

摘要:

结构剪枝是压缩和加速神经网络的一种有效方法。虽然滤波器和信道修剪在实际加速和硬件兼容性方面优于其他结构修剪方法,但是具有更细粒度的修剪方法(诸如信道内修剪)被期望能够产生更紧凑和计算效率更高的网络。由于大的搜索空间,典型的通道内修剪方法利用静态和手工修剪粒度,这为其修剪性能留下了改进的空间。在这项工作中,我们引入了一种新的结构修剪方法,称为动态结构修剪,以确定最佳的修剪粒度内通道修剪。与现有的通道内剪枝方法相比,所提出的方法在训练深度神经网络的同时自动优化每层的动态剪枝粒度。为了实现这一目标,我们提出了一种可微组学习方法,旨在有效地学习基于过滤器组的梯度学习的修剪粒度。实验结果表明,动态结构修剪实现了最先进的修剪性能和更好的真实感加速在GPU上相比,通道修剪。特别是,它将ResNet50的FLOP降低了71.85%,而不会降低ImageNet数据集的准确性。我们的代码可在www.example.com上获得。

英文摘要:

Structure pruning is an effective method to compress and accelerate neural networks. While filter and channel pruning are preferable to other structure pruning methods in terms of realistic acceleration and hardware compatibility, pruning methods with a finer granularity, such as intra-channel pruning, are expected to be capable of yielding more compact and computationally efficient networks. Typical intra-channel pruning methods utilize a static and hand-crafted pruning granularity due to a large search space, which leaves room for improvement in their pruning performance. In this work, we introduce a novel structure pruning method, termed as dynamic structure pruning, to identify optimal pruning granularities for intra-channel pruning. In contrast to existing intra-channel pruning methods, the proposed method automatically optimizes dynamic pruning granularities in each layer while training deep neural networks. To achieve this, we propose a differentiable group learning method designed to efficiently learn a pruning granularity based on gradient-based learning of filter groups. The experimental results show that dynamic structure pruning achieves state-of-the-art pruning performance and better realistic acceleration on a GPU compared with channel pruning. In particular, it reduces the FLOPs of ResNet50 by 71.85% without accuracy degradation on the ImageNet dataset. Our code is available at https://github.com/irishev/DSP.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑