保密通信新介质--人工神经网络

根据人工神经网络的结构特点和信息传输与处理特征，提出了利用物理神经网络作为通信信道的观点．在分析了现有通信信道物理模型之后，具体讨论了作为通信信道的神经网络的可能拓朴结构、神经网络学习要点、允许传输信号的模式特征，研究了神经网络信道的传输延时、通信带宽及信噪比，分析了神经网络信道处理多媒体信号的计算潜力，并着重剖析了神经网络信道固有的保密通信特征．最后给出了一个具体的神经网络信道设计实例．

Artificial neural network: a secrecy communication medium

In terms of characters of artificial neural networks and signal transmitting and processing, a point of view on communication channel based on physical neural networks was presented. The existing physical communication channel models were analyzed. The possible topological structures, the learning-training outline and the allowable models of signal transmitting on the neural network communication channel were discussed. The channel transmission time-delay, the bandwidth, the signal-to-noise and the potential for computation on multimedia signal were investigated. The secrecy character of the neural network communication channel was studied. An example on the neural network communication channel was presented.