基于改进的Hopfield离散神经网络的模式识别

本文介绍基于改进的Ｈｏｐｆｉｅｌｄ离散神经网络（ＩＨＤＮ）的模式识别。当一个要求存贮在网络中的模式输入时，我们计算它与其它模式的类似测试，并把类似测试在网络综合中加以考虑。当输入任一矢量时，由于网络的联想能力，该矢量将进化到与它距离最近的一个存贮模式中。理论分析和实验仿真结果表明，ＩＨＤＮ比原Ｈｏｐｆｉｅｌｄ离散神经网络（ＨＤＮ）有更大的存贮容量和更强的联想能力。

Pattern Recognition Based on an Improved Hopfield Discrete-time Neural Network

In this paper,we introduce pattern recognition based on an improved Hopfield discrete-time neural network model(IHDN).When an input desired pattern vector is provided,We compute the similarity of this vector to all of the desired patterns,and the similarity measures are taken into account in the synthesis of networks.When an arbitrary input vector is provided,it reaches one of patterns which is nearest from the input pattern in distance measure due to the associative capability of the network.The qualitative analysis andexperimintal simulation results show that the proposed network is of much better storage capacity and the associative capability than those of the original Hopield discrete-time neural network(HDN)