Cloud Computing, Smart Grid and Innovative Frontiers in Telecommunications. 9th EAI International Conference, CloudComp 2019, and 4th EAI International Conference, SmartGIFT 2019, Beijing, China, December 4-5, 2019, and December 21-22, 2019

Research Article

A Lightweight Neural Network Combining Dilated Convolution and Depthwise Separable Convolution

Download
229 downloads
  • @INPROCEEDINGS{10.1007/978-3-030-48513-9_17,
        author={Wei Sun and Xijie Zhou and Xiaorui Zhang and Xiaozheng He},
        title={A Lightweight Neural Network Combining Dilated Convolution and Depthwise Separable Convolution},
        proceedings={Cloud Computing, Smart Grid and Innovative Frontiers in Telecommunications. 9th EAI International Conference, CloudComp 2019, and 4th EAI International Conference, SmartGIFT 2019, Beijing, China, December 4-5, 2019, and December 21-22, 2019},
        proceedings_a={CLOUDCOMP},
        year={2020},
        month={6},
        keywords={Lightweight neural network Dilated convolution Depthwise separable convolution Classification accuracy Cloud computing},
        doi={10.1007/978-3-030-48513-9_17}
    }
    
  • Wei Sun
    Xijie Zhou
    Xiaorui Zhang
    Xiaozheng He
    Year: 2020
    A Lightweight Neural Network Combining Dilated Convolution and Depthwise Separable Convolution
    CLOUDCOMP
    Springer
    DOI: 10.1007/978-3-030-48513-9_17
Wei Sun,*, Xijie Zhou1,*, Xiaorui Zhang, Xiaozheng He2
  • 1: Nanjing University of Information Science and Technology
  • 2: Rensselaer Polytechnic Institute
*Contact email: sunw0125@163.com, cfcfwyjd@163.com

Abstract

Aimed to reduce the excessive cost of neural network, this paper proposes a lightweight neural network combining dilated convolution and depthwise separable convolution. Firstly, the dilated convolution is used to expand the receptive field during the convolution process while maintaining the number of convolution parameters, which can extract more high-level global semantic features and improve the classification accuracy of the network. Second, the use of the depthwise separable convolution reduces the network parameters and computational complexity in convolution operations. The experimental results on the CIFAR-10 dataset show that the proposed method improves the classification accuracy of the network while effectively compressing the network size.