Acta mathematica scientia,Series A ›› 2019, Vol. 39 ›› Issue (5): 1192-1204.

Previous Articles     Next Articles

Global Exponential Periodicity of Complex-Valued Neural Networks with Discontinuous Activation Functions

Yao Zou1(),Chunna Zeng2(),Jin Hu1,*()   

  1. 1 School of Mathematics and Statistics, Chongqing Jiaotong University, Chongqing 400074
    2 School of Mathematical Sciences, Chongqing Normal University, Chongqing 401331
  • Received:2018-10-31 Online:2019-10-26 Published:2019-11-08
  • Contact: Jin Hu E-mail:18323190385@163.com;zengchn@163.com;jhu@cqjtu.edu.cn
  • Supported by:
    the NSFC(61773004);the NSFC(11801048);the Program of Chongqing Innovation Team Project in University(CXTDX201601022);the Natural Scinece Foundation Project of CQ CSTC(cstc2017jcyjAX0172);the Natural Scinece Foundation Project of CQ CSTC(cstc2017jcyjAX0022);the Natural Scinece Foundation Project of CQ CSTC(cstc2017jcyjAX0082);the Natural Scinece Foundation Project of CQ CSTC(cstc2018jcyjAX0606);the Technology Research Foundation of Chongqing Educational Committee(KJ1705118);the Technology Research Foundation of Chongqing Educational Committee(KJQN201800740);the Venture & Innovation Support Program for Chongqing Overseas Returnees(cx2018034)

Abstract:

In this paper, we investigate a type of complex-valued neural networks with discontinuous activation functions. By using Filippov differential inclusion theory, Leray-Schauder alternative theorem and Lyapunov function, we obtain the sufficiet conditions for the global exponential periodicity of the neural network. The simulation shows the effectiveness of the results.

Key words: Complex-valued neural networks, Discontinuous activation functions, Global exponential periodicity

CLC Number: 

  • O175.13
Trendmd