特征加权距离的半监督模糊子空间聚类算法
发布时间：2019-06-15 03:03
【摘要】：针对已有的基于特征加权距离的双指数模糊子空间聚类算法(DI-FSC)没有充分利用数据集中的已知信息,提出利用少量监督信息辅助聚类过程的特征加权距离的半监督模糊子空间聚类算法(SS-FSC).通过在该算法中加入特征加权距离来改善传统聚类中利用欧氏距离计算数据点之间差异的不足;同时向约束条件中引入指数r和β,增加了算法的灵活性.实验表明,所提出的算法在少量监督信息的辅助下,在真实数据集上有较好的聚类效果.
[Abstract]:In view of the fact that the existing double exponential fuzzy subspace clustering algorithm (DI-FSC) based on feature weighted distance does not make full use of the known information in the data set, a semi-supervised fuzzy subspace clustering algorithm (SS-FSC) is proposed, which uses a small amount of supervised information to assist the feature weighted distance of the clustering process. By adding feature weighted distance to the algorithm, the shortage of using Euclidean distance to calculate the difference between data points in traditional clustering is improved, and the index r and 尾 are introduced into the constraint condition, which increases the flexibility of the algorithm. The experimental results show that the proposed algorithm has a good clustering effect on the real data set with the assistance of a small amount of supervision information.
【作者单位】： 江南大学数字媒体学院;
【基金】：国家自然科学基金项目(61272210)资助
【分类号】：TP181
本文编号：2499915
[Abstract]:In view of the fact that the existing double exponential fuzzy subspace clustering algorithm (DI-FSC) based on feature weighted distance does not make full use of the known information in the data set, a semi-supervised fuzzy subspace clustering algorithm (SS-FSC) is proposed, which uses a small amount of supervised information to assist the feature weighted distance of the clustering process. By adding feature weighted distance to the algorithm, the shortage of using Euclidean distance to calculate the difference between data points in traditional clustering is improved, and the index r and 尾 are introduced into the constraint condition, which increases the flexibility of the algorithm. The experimental results show that the proposed algorithm has a good clustering effect on the real data set with the assistance of a small amount of supervision information.
【作者单位】： 江南大学数字媒体学院;
【基金】：国家自然科学基金项目(61272210)资助
【分类号】：TP181
【相似文献】
相关期刊论文 前6条
1 章照止;神经网络与最小加权距离译码[J];电子学报;1992年10期
2 周立前;杨碧波;;一种构建细小病毒进化树的加权距离方法[J];湖南工业大学学报;2012年02期
3 王骏;王士同;王晓明;;基于特征加权距离的双指数模糊子空间聚类算法[J];控制与决策;2010年08期
4 韦立庆;陈秀宏;;基于Cam加权距离的增量拉普拉斯方法[J];计算机工程;2011年22期
5 鲁宇;范希鲁;;模糊加权距离及其合理性讨论[J];北方交通大学学报;1990年02期
6 ;[J];;年期
，本文编号：2499915
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2499915.html
