具有反应扩散的神经网络的稳定性分析
Globally Asymptotic Stability of Neural Networks with Reaction-Diffusion
-
摘要: 通过构造适当的平均Lyapunov函数,利用M矩阵理论,研究了一类具有反应扩散的Hopfield神经网络的 全局稳定性.在放松神经网络的激活函数的有界性、单调递增性和可微性的条件下,得到了神经网络的全局渐近 稳定的条件.这些条件适合于神经网络的关联矩阵为对称或非对称矩阵、激活函数为非单调的情况.
-
关键词:
- 神经网络 /
- 反应扩散 /
- 全局渐近稳定 /
- M矩阵 /
- Lyapunov函数
Abstract: Even type Lyapunov functions were constructed based on M-matrix theory to study the globally asymptotic stability of Hopfield networks with reaction-diffusion. These networks are generalized without assuming the boundedness, monotonicity and differenciability of the activate functions. The conditions were obtained for globally asymptotic stability of the generalized Hopfield networks, where the interconnection matrices are symmetric or non-asymmetric and the neural activate functions are non-monotonic.-
Key words:
- neural networks /
- reaction-diffusion /
- global asymptotic stability /
- M-matrix /
- Lyapunov function
点击查看大图
计量
- 文章访问数: 1561
- HTML全文浏览量: 62
- PDF下载量: 86
- 被引次数: 0