摘要
论文针对ART2网络学习与记忆的特点,在原始ART2的基础上提出具有遗忘机制的改进模型,并开发了相应的MATLAB程序。改进模型解决了原始ART2网络权值学习的随机偏移问题,有效地过滤了噪声,提高了分类结果的稳定性,降低了空间存储消耗。文章进一步运用改进模型对典型输入样本进行了分类,得到了理想的结果。
An improved ART2 algorithm is presented in this paper.By introducing the forgetting mechanism of organism,the proposed algorithm not only inherits all advantages of ART2,but also economizes the networks＇ memory capacity and eliminates wrong patterns effectively.The MATLAB program of the new model is developed,and the satisfied results are obtained in the simulations with clustering and classification of data.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第9期60-62,98,共4页
Computer Engineering and Applications
