Hopfield连续神经网络平衡态局部指数稳定的充要条件

Local Stability of Hopfield Associative Memory Neural Networks

研究Ｈｏｐｆｉｅｌｄ连续联想记忆神经网络平衡点的局部指数稳定性，给出了Ｈｏｐｆｉｅｌｄ连续联想记忆神经网络的平衡点局部指数稳定的充要条件．通过引入特征函数的方法，得到了各记忆模式的指数吸引域的估计，以此可用来评价Ｈｏｐｆｉｅｌｄ连续反馈联想记忆的容错能力和模式恢复能力，且可用于综合更为实用的连续联想记忆神经网络．

A sufficient and necessary condition is established for Hopfield continuous time neural network of associative memory with local exponential stability. A function is presented to quantify the character of the domain of attraction of equilibrium point.These results can be used not only for evalua ting error correction and pattern recovery, but also for synthesis of neural network with associate memories.