摘要
本文针对联想记忆网络学习样本的选择问题,从网络的泛化能力入手,讨论了学习样本的数量、质量和选取方法问题.并通过一个交通标志形状识别系统的实验,给出了如何确定联想记忆网络学习样本的数量、质量和选取方法的建议.
To solve the problem of how to select training samples for associative memory (AM) neural networks, the number and quality of training samples and the way to choose them are discussed in this paper, according to the generalization capability of neural networks, a suggestion is given to define the number and quality of samples of AMNN and the way to choose them, according to an experiment of a traffic sign shape recognition system.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2002年第3期367-371,共5页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金项目(No.60175011)
安徽省自然科学基金(No.01042301)
教育部优秀青年资助计划
关键词
联想记忆
学习样本
泛化能力
识别
聚类
AM
神经网络理论
Associative Memory(AM), Training Sample, Generalization Capability, Recognition, Clustering
