Runtime Network Routing for Efficient Image Classification

Runtime Network Routing for Efficient Image Classification
复制标题

用于高效图像分类的运行时网络路由

DOI:
10.1109/tpami.2018.2878258
复制
发表时间:
2019-10
影响因子:
23.6
通讯作者:
Zhou Jie
Zhou Jie
中科院分区:
计算机科学1区
文献类型:
--
作者:
Rao Yongming;Lu Jiwen;Lin Ji;Zhou Jie

文献摘要

参考文献

相似文献

在本文中,我们提出了一个通用的运行时网络路由(RNR)框架,用于有效的图像分类,该框架在网络中选择最优路径。与现有的静态神经网络加速方法不同,我们的方法保留了原始大型网络的全部能力,并根据输入图像和当前特征图在运行时进行动态路由。路由以自下而上、逐层的方式执行,我们将其建模为马尔可夫决策过程,并使用强化学习进行训练。agent确定每个子路径的估计奖励,并根据不同的样本进行路由,当图像更容易完成任务时,选择更快的路径。由于充分保留了网络的能力,因此很容易根据可用资源调整平衡点。我们在多路径残差网络和增量卷积通道剪枝上测试了我们的方法,并表明在CIFAR和ImageNet数据集上,在相同的计算复杂度下,RNR始终优于静态方法。我们的方法也可以应用于现成的神经网络结构,并且很容易扩展到其他应用场景。
In this paper, we propose a generic Runtime Network Routing (RNR) framework for efficient image classification, which selects an optimal path inside the network. Unlike existing static neural network acceleration methods, our method preserves the full ability of the original large network and conducts dynamic routing at runtime according to the input image and current feature maps. The routing is performed in a bottom-up, layer-by-layer manner, where we model it as a Markov decision process and use reinforcement learning for training. The agent determines the estimated reward of each sub-path and conducts routing conditioned on different samples, where a faster path is taken when the image is easier for the task. Since the ability of network is fully preserved, the balance point is easily adjustable according to the available resources. We test our method on both multi-path residual networks and incremental convolutional channel pruning, and show that RNR consistently outperforms static methods at the same computation complexity on both the CIFAR and ImageNet datasets. Our method can also be applied to off-the-shelf neural network structures and easily extended to other application scenarios.
DOI: --
发表时间: 2016-02
期刊: ArXiv
影响因子: --
作者:
F. Iandola;Matthew W. Moskewicz;Khalid Ashraf;Song Han;W. Dally;K. Keutzer
通讯作者: F. Iandola;Matthew W. Moskewicz;Khalid Ashraf;Song Han;W. Dally;K. Keutzer
DOI: --
发表时间: 2015-11
期刊: --
影响因子: --
作者:
Amjad Almahairi;Nicolas Ballas;Tim Cooijmans;Yin Zheng;H. Larochelle;Aaron C. Courville
通讯作者: Amjad Almahairi;Nicolas Ballas;Tim Cooijmans;Yin Zheng;H. Larochelle;Aaron C. Courville
DOI: --
发表时间: --
期刊: --
影响因子: --
作者:
Neeraj Kumar;A. Berg;P. Belhumeur;S. Nayar
通讯作者: Neeraj Kumar;A. Berg;P. Belhumeur;S. Nayar
DOI: 10.1609/aaai.v32i1.11630
发表时间: 2017-01
期刊: ArXiv
影响因子: --
作者:
Lanlan Liu;Jia Deng
通讯作者: Lanlan Liu;Jia Deng
DOI: 10.1007/s11263-015-0816-y
发表时间: 2015-12-01
影响因子: 19.5
作者:
Russakovsky, Olga;Deng, Jia;Fei-Fei, Li
通讯作者: Fei-Fei, Li