Algorithm Research on Improving Activation Function of Convolutional Neural Networks

被引：0
作者：
Guo, Yanhua [1 ]
Sun, Lei [1 ]
Zhang, Zhihong [2 ]
He, Hong [1 ]
机构：
[1] Tianjin Univ Technol, Sch Elect Engn & Automat, Tianjin Key Lab Control Theory & Applicat Complic, Tianjin 300384, Peoples R China
[2] Tianjin Inst Elect Technol, Tianjin 300232, Peoples R China
来源：
PROCEEDINGS OF THE 2019 31ST CHINESE CONTROL AND DECISION CONFERENCE (CCDC 2019) | 2019年
关键词：
Deep Learning; Convolutional Neural Networks; Activation function; Image classification;
D O I：
10.1109/ccdc.2019.8833156
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
Aiming at the slow convergence of t he activation function of Sigmoid, Tanh, ReLu and Softplus as the model and the non-convergence caused by gradient dispersion, this paper proposes an algorithm to improve the activation function of convolutional neural network. UsingR-SReLU as the activation function of the neural network, the convergence speed of various excitation functions to the network and the accuracy of image recognition are analyzed. The experimental data shows that the improved activation function R-SReLU not only has a fast convergence speed, but also has a small error rate, and can improve the accuracy of classification more effectively. The maximum recognition accuracy reaches 88.03%.
引用
收藏
页码：3582 / 3586
页数：5
相关论文
共 13 条
  • [1] Convolutional Sparse Feature Descriptor for Object Recognition in CIFAR-10
    Carvalho, Edigleison Francelino
    Engel, Paulo Martins
    [J]. 2013 BRAZILIAN CONFERENCE ON INTELLIGENT SYSTEMS (BRACIS), 2013, : 131 - 135
  • [2] Gan Z, 2015, ARTIFICIAL INTELLIGE
  • [3] Harley AW, 2015, PROC INT CONF DOC, P991, DOI 10.1109/ICDAR.2015.7333910
  • [4] Khan NI, 2018, ARTIFICIAL NEURAL NE
  • [5] Kiran J S, 2015, J MATER CHEM, V42
  • [6] Liang M, 2015, PROC CVPR IEEE, P3367, DOI 10.1109/CVPR.2015.7298958
  • [7] Mccarthy J., 2016, COMMUN ACM, P1
  • [8] Qiu-Yunfl Qiu X, 2014, J SIGNAL PROCESSING
  • [9] Deep learning in neural networks: An overview
    Schmidhuber, Juergen
    [J]. NEURAL NETWORKS, 2015, 61 : 85 - 117
  • [10] Senior Andrew, 2014, 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), P7644, DOI 10.1109/ICASSP.2014.6855087
12