|
The linear program problem in dynamic non-hierarchical routing(DNHR) is solved by using Hopfield's nerve net model. The second power function problem is dcalt with by the model. Furthermore, this technique is extended to the convex program and the optimization of the second power function. As a result, thc adaptive filter problem can be solved easily by this technique. 本文利用Hopfield神经网模型,求解电话网动态无级路由选择技术DNHR(Dynamic Non-Hierarchical Routing)问题,并扩展到一切凸规划及二次型优化技术中,用以解决自适应滤波问题。 The author introduces the forward manual nerve net model and makes a comparison with the linear model of traditional statistics. The counter propagation model is improved in order to fit in with the characteristic of time sequence variable. And the author analyses the typical time sequence of the observed value of sunspot moving , examines the forecasting ability of the model and gets good results. 介绍前向人工神经网络模型,并与传统的统计学线性模型进行比较。为了适应时间序列变量的特点,对逆传播模型进行了改进,并对太阳黑子活动的观察值这一典型的时间序列进行了分析,并检验了模型的预测能力,取得了较好的结果 The method of" Nerve net" applies successfully into many fields This paper leads some concetptions of Nerve net and Models to study the realtion of populations in community Our discussion centre on the intensity of touch among populatins . Introduce the method of Nerve net may give an impetus to study of ecology 神经网络作为方法论成功地应用到许多领域中,本文引入神经网络中的一些概念(如学习、稳定性、随机、能量)及其模型对群落内种群关系进行了初步地探讨。并且着重讨论种群间的关联强度,引入神经网络方法,可以大大推进生态学的研究
|