基于区域分解的快速卷积神经网络学习策略研究
作者:
作者单位:

作者简介:

通讯作者:

基金项目:

国家自然科学基金项目(81661168015, 81871447);深圳市科学技术创新委员会项目(ZDSYS201703031711426)

伦理声明:



A Study on Domain Decomposition Inspired Fast Convolutional Neural Network Learning Strategy
Author:
Ethical statement:

Affiliation:

Funding:

National Natural Science Foundation of China(81661168015, 81871447); Shenzhen Science and Technology Innovation Commission Project (ZDSYS201703031711426)

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    为加快卷积神经网络的训练,该研究提出一种受区域分解方法启发的新型学习策略。将该方 法应用于残差网络(ResNet)进行图像分类时,使用 ResNet32 可获得最佳结果。进一步地,将 ResNet32 分成 4 个子网络,其中每个子网具有 0.47 M 参数,此为原始 ResNet32 的 1/16,从而简化了学习过程。 此外,由于可以并行训练子网络,因此在使用 CIFAR-10 数据集进行分类任务时,计算时间可以从 8.53 h (通过常规学习策略)减少到 5.65 h,分类准确性从 92.82% 提高到 94.09%。CIFAR-100 和 Food-101 数 据集也实现了类似的改进。实验结果显示,所提出的学习策略可以大大减少计算时间,并提高分类的 准确性。这表明所提出的策略可以潜在地应用于训练带有大量参数的卷积神经网络。

    Abstract:

    We propose a novel learning strategy inspired by domain decomposition methods to accelerate the training of convolutional neural network (CNN). The proposed method is applied to residual networks (ResNet) for image classification tasks. The best result is achieved with ResNet32. In this case, we split ResNet32 into 4 sub-networks. Each sub-network has 0.47 M parameters which is 1/16 of the original ResNet32, thereby facilitating the learning process. Moreover, because the sub-networks can be trained in parallel, the computational time can therefore be reduced to 5.65 h from 8.53 h (by the conventional learning strategy) in the classification task with the CIFAR-10 dataset. We also find that the accuracy of the classification is improved to 94.09% from 92.82%. Similar improvements are also achieved with the CIFAR-100 and Food-101 datasets. In conclusion, the proposed learning strategy can reduce the computational time substantially with improved accuracy in classification. The results suggest that the proposed strategy can potentially be applied to train CNN with a large amount of parameters.

    参考文献
    相似文献
    引证文献
引用本文

引文格式
张 卫,古林燕,刘 嘉.基于区域分解的快速卷积神经网络学习策略研究 [J].集成技术,2020,9(6):48-58

Citing format
ZHANG Wei, GU Linyan, LIU Jia. A Study on Domain Decomposition Inspired Fast Convolutional Neural Network Learning Strategy[J]. Journal of Integration Technology,2020,9(6):48-58

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2020-11-24
  • 出版日期: