作者: Yiguang Liu , Zhisheng You
DOI: 10.1016/J.CHAOS.2006.01.081
关键词: Bounded function 、 Activation function 、 Cellular neural network 、 Monotonic function 、 Recurrent neural network 、 Mathematics 、 Lipschitz continuity 、 Stochastic neural network 、 Applied mathematics 、 Mathematical analysis 、 Artificial neural network 、 General Mathematics
摘要: Abstract This paper studies multi-stability, existence of almost periodic solutions a class recurrent neural networks with bounded activation functions. After introducing sufficient condition insuring many criteria guaranteeing are derived using Mawhin’s coincidence degree theory. All the constructed without assuming functions smooth, monotonic or Lipschitz continuous, and that contains variables (such as coefficients, inputs functions), so all can be easily extended to fit concrete forms such Hopfield networks, cellular etc. Finally, kinds simulations employed illustrate criteria.