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 1902.08192cs.CVcs.LGcs.NE
收藏

对偶空间-Winograd域上的联合稀疏卷积神经网络

Jointly Sparse Convolutional Neural Networks in Dual Spatial-Winograd Domains

Samsung Semiconductor Inc.(三星半导体公司)

Yoojin Choi, Mostafa El-Khamy, Jungwon Lee

摘要:

我们考虑深度卷积神经网络(CNN)的优化,使得它们提供良好的性能,同时如果部署在具有空间域卷积的传统系统或为Winograd卷积设计的较低复杂性系统上,则具有降低的复杂性。所提出的框架产生一个压缩模型,其卷积滤波器可以在空间域或Winograd域中稀疏。因此,压缩模型可以在任何平台上通用部署,而无需在部署的平台上重新训练。为了获得更好的压缩比,稀疏模型在具有较少数量的参数的空间域中被压缩。从我们的实验中,我们获得了在ImageNet数据集上训练的ResNet-18和AlexNet的$24.2\times$和$47.7\times$压缩模型,同时它们的计算成本也分别被$4.5\times$和$5.1\times$降低。

英文摘要:

We consider the optimization of deep convolutional neural networks (CNNs) such that they provide good performance while having reduced complexity if deployed on either conventional systems with spatial-domain convolution or lower-complexity systems designed for Winograd convolution. The proposed framework produces one compressed model whose convolutional filters can be made sparse either in the spatial domain or in the Winograd domain. Hence, the compressed model can be deployed universally on any platform, without need for re-training on the deployed platform. To get a better compression ratio, the sparse model is compressed in the spatial domain that has a fewer number of parameters. From our experiments, we obtain $24.2\times$ and $47.7\times$ compressed models for ResNet-18 and AlexNet trained on the ImageNet dataset, while their computational cost is also reduced by $4.5\times$ and $5.1\times$, respectively.

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。

↑