一种基于高阶矩的神经网络参数估计器

A Cumulant-Based Parameter Estimator Using Neural Networks

本文提出了一种可控学习的两级多层神经网络模型,由此设计出一种基于高阶矩匹配的神经网络参数估计器;并对该神经网络模型的学习算法进行了研究,提出了一种自适应并行学习算法。仿真结果表明,这种利用神经网络进行模型参数估计的方法是可行的。

This paper develops a learning-controllable multilayered neural network to design a kind of parameter estimator based on eumulant matching. The learning of the network is also studied and an adaptive parallel learning algorithm is proposed. Simulations demonstrate the applicability of the neural network approach to parameter estimation.