0 引言在神经网络的研究工作中,Hopfield模型的建立被称为是突破性的,但用以解优化问题有其局限性,即被优化的目标函数只能是二次函数.本文给出了一种推广的Hopfield模型,可用于解决一大类优化问题,只要该优化问题能表示成某一目标函数,且该函数能满足对各变量的一价偏导数存在且连续的条件即可. 1 神经网络的结构 0 Introduction In the research work of neural networks, the establishment of Hopfield model is called a breakthrough, but there are limitations to solve the optimization problem, that is, the objective function to be optimized can only be a quadratic function. A generalized Hopfield model can be used to solve a large class of optimization problems as long as the optimization problem can be expressed as an objective function and the function can satisfy the existence and continuity of the one-valent partial derivatives of the variables. Structure of neural networks