-
摘要: 讨论了关于改进LVQ聚类网络的理论与算法.为克服LVQ网络聚类算法对初值敏 感的问题广义学习矢量量化(GLVQ)网络算法对LVQ算法进行了改进,但GLVQ算法性能不 稳定.GLVQ-F是对GLVQ网络算法的修改,但GLVQ-F算法仍存在对初值的敏感问题.分 析了GLVQ-F网络算法对初值敏感的原因以及算法不稳定的理论缺陷,改进了算法理论并给 出了一种新的改进的网络算法(MLVQ).实验结果表明新的算法解决了原有算法所存在的问 题,而且性能稳定.Abstract: Theory and algorithms of clustering neural network are discussed. The generalized LVQ neural network was aimed to improve the algorithm of LVQ. However it brought another problem. GLVQ-F algorithm was an improvement to GLVQ, but it was still unstable and left the problem of LVQ unsolved. The defect of GLVQ-F clustering neural network algorithm is theoretically analyzed in this paper, and the modified theory is discussed. Finally, a modified neural network algorithm of competitive learning schemes (MLVQ) is given. Experiment results show the new algorithm is stable and effective.
-
Key words:
- LVQ algorithm /
- cluster analysis /
- MLVQ algorithm
-
点击查看大图
计量
- 文章访问数: 2752
- HTML全文浏览量: 91
- PDF下载量: 1168
- 被引次数: 0
下载:
