论文部分内容阅读
本文提出一类可用于模式识别的联想神经网络的综合方法,这类网络结构不受对称联接的限制,网络保证了要求的M类模式的稳定形成,且网络的容量远远超过Hopfield的联想神经网络,网络渐近稳定平衡点的吸引特性使受噪声污染的模式能得以正确恢复,体现了神经网络的非线性滤波性质。文中给出了综合一个这类联想网络计算机模拟以及模式识别的例子。
In this paper, we propose an integrated approach that can be used for pattern recognition in the context of associative neural networks that are not constrained by symmetric linking. The network guarantees the stable formation of the required M-class patterns and the capacity of the network far exceeds that of Hopfield’s associative nerves The attracting characteristics of networks and networks asymptotically stable equilibrium point make the mode of noise pollution recover correctly, which reflects the nonlinear filtering property of neural network. The paper gives an example of synthesizing computer simulation and pattern recognition of such a network.