计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (13): 172-177.DOI: 10.3778/j.issn.1002-8331.1805-0215

• 模式识别与人工智能 • 上一篇    下一篇

运用PSO算法的自递归RBF网络建模与应用

曾  添,杨德刚   

  1. 重庆师范大学 计算机与信息科学学院,重庆 401331
  • 出版日期:2019-07-01 发布日期:2019-07-01

Modeling and Application of Self Recursive RBF Network with PSO Algorithm

ZENG Tian, YANG Degang   

  1. College of Computer & Information Science, Chongqing Normal University, Chongqing 401331, China
  • Online:2019-07-01 Published:2019-07-01

摘要: 普通三层RBF网络已经是一种较好的神经网络,为了进一步提高RBF网络的性能,在普通三层RBF网络的基础上,构建出一种运用PSO算法的自递归RBF网络。学习算法采用以梯度学习算法配合PSO算法对参数进行调整。与采用动量-梯度学习算法,且为结构为三层的RBF网络相比,提的运用PSO算法的自递归RBF网络可以在神经元较少的情况下,具有更好的泛化能力、鲁棒性和准确性。最后通过仿真实验,对算法的有效性进行了验证。

关键词: 递归子网, 径向基子网络, 深度学习, 梯度学习算法, PSO算法

Abstract: The ordinary three layer RBF network is a kind of superior neural network. In order to further improve the new ability of RBF network, a self recursive RBF network with the PSO algorithm is constructed on the basis of the ordinary three layer RBF network. Learning algorithm adopts gradient learning algorithm and PSO algorithm to adjust parameters. By comparing with RBF network based on the three layer structure and using the momentum gradient learning algorithm, it is found that in proposal way of this paper, the self recursive RBF network with the PSO algorithm has better generalization ability, robustness and accuracy in the case of less neuron. Finally, the effectiveness of the algorithm is verified by simulation experiments.

Key words: recursive subnet, Radial Basis Function(RBF) network, deep learning, gradient learning algorithm, PSO algorithm