利用EM算法估计隐含观测量的回馈神经网络学习新方法  

A New Method for Training RNN via Hidden Representation Estimated by EM Algorithm

作　　者：戴宪华[1] 

机构地区：[1]汕头大学电子工程系,汕头515063

出　　处：《电子学报》2000年第10期133-137,共5页Acta Electronica Sinica

基　　金：国家自然科学基金!(No.69872 0 2 1 );广东省自然科学基金!(No .980 4 38)

摘　　要：研究回馈神经网络 (RNN)参数估计的新方法 .利用隐含观测量 ,将复杂RNN的训练分解为线性输出层和多个单隐元的参数估计 .基于每个隐元激励函数的多点线性近似 ,RNN可利用统计混合专家网络模型 (ME)描述 ,从而将RNN的参数估计转化为包含隐含观测量的线性系统的最大似然估计问题 ,最后利用期望最大化 (EM)算法获得RNN的隐含观测量及其参数估计 .A new method for training recurrent neural network (RNN) has been proposed.By introducing the hidden representation or hidden variables into RNN,training the complicated RNN is decomposed into training a set of single neurons and a linear output layer.Based on linear approximation of RNN hidden units,RNN is remodeled with a “mixture of experts”(ME) model.Morever,training RNN is also changed into a maximum likelihood estimation of the linear systems with hidden variables.Finally,training RNN is fulfilled with the expectation maximization (EM) algorithm.

关 键 词：回馈神经网络 EM算法 隐含观测量 学习算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]