离散时间细胞神经网络的收敛性条件

Study on the Convergence of Discrete-Time Cellular Neural Networks

神经网络的收敛性是网络各种应用的基础。主要研究了离散细胞神经网络的收敛性 ,并给出了几个新的网络收敛性条件。如果细胞网络的模板不是互补的 ,则给出一个网络在细胞格子非相互作用演化方式下的收敛性结果 ,所获结果推广了已有的结论。如果模板是互补的 ,且是行占优的 ,则网络按细胞格子行方式进行演化是收敛的。如果模板是互补的 ,且是列占优的 ,则网络按细胞格子列方式进行演化是收敛的。

The convergence of neural networks is known as the basis of various applications. In this paper, convergence of discrete-time cellular neural networks is studied and some new convergence conditions of the networks are given. If the template of the network is not reciprocal, a new convergence result of the networks with non-interacting evolution mode is given and the obtained result generalizes the existing result. If the template is reciprocal and row-dominant, the network is convergent in a row-wise evolution of the cell-grid. If the template is reciprocal and column-dominant, the network is convergent in a column-wise evolution of the cell-grid.;