摘要
提出了最大乘积型联想记忆网络的一种动态调整学习算法，给出了动态学习算法有效记忆训练规则的一个充要条件。首先给出了一种快速调整学习算法，再进一步发展了一个动态指数细调规则的学习算法，以快速调整学习算法的结果作为连接权矩阵的迭代初值。实验给出了所提算法的两个应用实例。
A dynamic learning algorithm of fuzzy max product associative memory is proposed and a sufficient condition is given for effective memory training rules. First a quick dynamic adjusting algorithm is developed, and then a dynamic exponent learning algorithm for delicate adjustment of inference rules is obtained. The latter algorithm takes the weights of the former as its initial iteration va lue. Experiments demonstrate two applications of the proposed algorithms.
出处
《数据采集与处理》
CSCD
1998年第2期126-130,共5页
Journal of Data Acquisition and Processing
基金
国家攀登计划项目
国家自然科学基金
关键词
神经网络
联想记忆网络
学习算法
图象处理
fuzzy neural network
fuzzy associative memory
learning algorithm
image process
