机构地区: 华中理工大学控制科学与工程系
出 处: 《自动化学报》 1999年第6期721-727,共7页
摘 要: 推广了前人关于离散Hopfield 神经网络的稳定性定理及周期为2 极限环的存在定理,并从理论上给出了新的严格的证明.进一步,提出了关于部分变元稳定和部分变元为极限环的新概念,并给出了判别定理.最后给出了几个有趣的例子,揭示这类网络渐近行为的复杂性. In this paper,we present some results for discrete\|time Hopfield neural networks,which extend the stability results and existence criteria of 2\|period limit cycles given by previous authors.These results are proved strictly here.Furthermore,we propose new conceps on the stability and limit cycles of part neurons in a network,and related criteria.Several examples are given to illustrate the complexity of the asymptoticall behavior of this kind of networks.
领 域: [自动化与计算机技术] [自动化与计算机技术]