arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.04242cs.CV
收藏

加速卷积神经网络的动态群卷积

Dynamic Group Convolution for Accelerating Convolutional Neural Networks

University of Oulu(奥卢大学);South China University of Technology(华南理工大学);National University of Defense Technology(国防科技大学)

Zhuo Su, Linpu Fang, Wenxiong Kang, Dewen Hu, Matti Pietikäinen, Li Liu

摘要:

用群卷积取代常规卷积可以显著提高现代深度卷积网络的计算效率,这在紧凑的网络架构设计中已被广泛采用。然而,现有的群卷积通过永久地切断一些连接来破坏原始网络结构,从而导致显著的精度下降。在本文中,我们提出了动态组卷积(DGC),自适应地选择输入通道的哪一部分在每个组内连接的个别样本的飞行。具体来说,我们为每个组配备了一个小的特征选择器,以自动选择以输入图像为条件的最重要的输入通道。多个组可以自适应地捕获每个输入图像的丰富和互补的视觉/语义特征。DGC保留了原有的网络结构,同时具有与传统群卷积相似的计算效率。在CIFAR-10、CIFAR-100和ImageNet等多个图像分类基准上的大量实验表明,该方法优于现有的组卷积技术和动态执行方法。该代码可在https://github.com/zhuogege1943/dgc上获得。

英文摘要:

Replacing normal convolutions with group convolutions can significantly increase the computational efficiency of modern deep convolutional networks, which has been widely adopted in compact network architecture designs. However, existing group convolutions undermine the original network structures by cutting off some connections permanently resulting in significant accuracy degradation. In this paper, we propose dynamic group convolution (DGC) that adaptively selects which part of input channels to be connected within each group for individual samples on the fly. Specifically, we equip each group with a small feature selector to automatically select the most important input channels conditioned on the input images. Multiple groups can adaptively capture abundant and complementary visual/semantic features for each input image. The DGC preserves the original network structure and has similar computational efficiency as the conventional group convolution simultaneously. Extensive experiments on multiple image classification benchmarks including CIFAR-10, CIFAR-100 and ImageNet demonstrate its superiority over the existing group convolution techniques and dynamic execution methods. The code is available at https://github.com/zhuogege1943/dgc.

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。

↑