arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.06776cs.CVcs.AI
收藏

面向网络剪枝的剪枝感知稀疏正则化

Pruning-aware Sparse Regularization for Network Pruning

University of Chinese Academy of Sciences(中国科学院大学)

Nanfei Jiang, Xu Zhao, Chaoyang Zhao, Yongqi An, Ming Tang, Jinqiao Wang

AI总结:

本文提出名为MaskSparsity的剪枝感知稀疏正则化方法,仅对剪枝掩码选定的滤波器施加细粒度稀疏正则化,避免未剪枝通道受不必要限制,在CIFAR-10和ILSVRC-2012上实现高FLOPs减少且精度损失极小。

摘要:

结构化神经网络剪枝旨在通过剪除对最终输出精度重要性较低的滤波器,来移除深度卷积神经网络(CNNs)中的冗余通道。为了减少剪枝后的性能下降,许多方法利用带有稀疏正则化的损失来产生结构化稀疏性。在本文中,我们分析了这些基于稀疏性训练的方法,发现对未剪枝通道的正则化是不必要的。此外,它限制了网络的容量,导致欠拟合。为了解决这个问题,我们提出了一种名为 MaskSparsity 的新颖剪枝方法,采用剪枝感知稀疏正则化。MaskSparsity 对由剪枝掩码选定的特定滤波器施加细粒度稀疏正则化,而不是模型的所有滤波器。在 MaskSparsity 的细粒度稀疏正则化之前,我们可以使用多种方法获取剪枝掩码,例如运行全局稀疏正则化。MaskSparsity 在 ResNet-110 上实现了 63.03%-FLOPs 减少,移除了 60.34% 的参数,在 CIFAR-10 上没有 top-1 精度损失。在 ILSVRC-2012 上,MaskSparsity 在 ResNet-50 上减少了超过 51.07% FLOPs,top-1 精度仅损失 0.76%。代码发布于 https://github.com/CASIA-IVA-Lab/MaskSparsity。此外,我们已将 MaskSparsity 的代码集成到 PyTorch 剪枝工具包 EasyPruner 中,地址为 https://gitee.com/casia_iva_engineer/easypruner。

英文摘要:

Structural neural network pruning aims to remove the redundant channels in the deep convolutional neural networks (CNNs) by pruning the filters of less importance to the final output accuracy. To reduce the degradation of performance after pruning, many methods utilize the loss with sparse regularization to produce structured sparsity. In this paper, we analyze these sparsity-training-based methods and find that the regularization of unpruned channels is unnecessary. Moreover, it restricts the network's capacity, which leads to under-fitting. To solve this problem, we propose a novel pruning method, named MaskSparsity, with pruning-aware sparse regularization. MaskSparsity imposes the fine-grained sparse regularization on the specific filters selected by a pruning mask, rather than all the filters of the model. Before the fine-grained sparse regularization of MaskSparity, we can use many methods to get the pruning mask, such as running the global sparse regularization. MaskSparsity achieves 63.03%-FLOPs reduction on ResNet-110 by removing 60.34% of the parameters, with no top-1 accuracy loss on CIFAR-10. On ILSVRC-2012, MaskSparsity reduces more than 51.07% FLOPs on ResNet-50, with only a loss of 0.76% in the top-1 accuracy. The code is released at https://github.com/CASIA-IVA-Lab/MaskSparsity. Moreover, we have integrated the code of MaskSparity into a PyTorch pruning toolkit, EasyPruner, at https://gitee.com/casia_iva_engineer/easypruner.

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。

↑