基于改进MRII的Madaline网络学习算法  

A Madaline Network Learning Algorithm Based on Improved MRII

作　　者：郑星[1] 李宗祥[1] 

机构地区：[1]河海大学计算机与信息学院智能科学与技术研究所,江苏南京211100

出　　处：《计算机与现代化》2016年第2期49-51,57,共4页Computer and Modernization

基　　金：国家自然科学基金资助项目(60971088)

摘　　要：根据Madaline网络工作原理,针对其网络特点和现有算法中存在的缺点,包括存在权值修改公式参数较多不容易协调,经验取值缺乏理论依据不够灵活,按照置信度原则进行翻转神经元会陷入"局部震荡"。提出改进的MRII学习算法,通过建立神经元敏感性替代置信度作为度量隐层神经元翻转的尺度,并采用感知机学习规则,减少权值调整次数,实验结果验证了该算法的优越性。In this paper, by analyzing the Madaline network work principle, according to the network characteristic and the shortcomings of the existing algorithms including the adjust weigh formula used in MRII algorithm with many parameters, but most of which come from experience in practice without theoretical reason, it made networks fall into ＂local recycle＂ with turning over the neuron based on confidence principle. We present an improved learning algorithm based on MRII, by establishing neuron＇ s sensitivity as a tool for measuring the turn of each hidden neuron, which reduces the number of weight adjustment. Some experimental results verify the effectiveness of the proposed algorithm.

关 键 词：神经网络 Madaline 学习算法 敏感性 MRII 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]