具有不连续神经激励的循环神经网络的周期解存在性(英文)

Existence of Periodic Solutions for Recurrent Neural Networks with Discontinuous Neuron Activations

本文研究了一类可以描述为右端不连续微分方程的循环神经网络模型.在并不要求激励函数连续、有界及单调非减的情况下,通过利用线性矩阵不等式和微分包含中的Cellina近似选择定理,得到了该神经网络模型存在周期解的充分条件.最后,给出了一个数值例子用以说明本文结果的有效性.

In this paper, we consider a class of recurrent neural network models which are described by differential equations with discontinuous right-hand side. Without presuming the activation functions to be continuous, bounded and monotone nondecreasing, by utilizing linear matrix inequality and CeUina approximate selection theory in differential inclusion,a sufficient condition is provided to ensure existence of periodic solutions for that recurrent neural networks. An example is given to illustrate the effectiveness of our results.