2002 年 38 巻 11 号 p. 1003-1005
In this note, we discuss on the stability of neural networks for image restoration. The neural network treated here is described by a system of linear ordinary differential equations. Our aim is to give the convergence rate of the neural network. In particular, it is shown that, in the case where an image consists of a great many pixels, the convergence rate is derived based on the one of the corresponding partial differential equation.