广义Hopfield神经网络模型

A GENERALIZED HOPFIELD NEURAL NETWORK MODEL

本文在Ｈｏｐｆｉｅｌｄ神经网络模型中通过充分考虑稳定性条件，引入节点偏置和引入高阶相互作用项使其存储容量远高于传统的Ｈｏｐｆｉｅｌｄ神经网络模型，并能存储相关图样。给出了理论推导和分析，对由Ｎ个神经元组成的神经网络系统进行计算机仿真，仿真结果证实了上述结论。新模型的上述特点使其具有良好的应用前景。

This paper investigates a generalized opfield neural network model by con-sidering stability condition,introducing node bias and higher-order interaction.Its storage capacity is much higher than that of Hopfield model,and can storecoherent patterns. The theoretical analyses are given.Calculations of computersimulation to neural network svstems with N neurons are carried out.The re-sults confirm the above conclusion. All these distinguishing features enable thenew model to be of good prospects in application.