基于倾斜概率的有效数据聚类数学模型
On Efficient Data Clustering Mathematical Model Based on Tilt Probability
-
摘要: 在相似数据聚类过程中，由于数据相似性过高，造成特征冗余干扰，使数据中心很难确定。该文提出了一种基于倾斜概率的有效聚类数学模型，在计算聚类中心的过程中引入倾斜概率计算数据均值。在数据特征存在较大一致性冗余干扰时，计算分配到同一类簇的概率并建立约束，把这种概率约束运用到数据的类间相似性特征聚类中，可以有效地确定相似特征的初始聚类中心。实验证明该文模型能合理地选择出初始聚类中心，改进分类数学模型的分类效果，与 k均值聚类模型相比，聚类结果更加紧致，鲁棒性更强。Abstract: The process of similar data clustering ,and the high data similarity characteristics of redundancy caused by interference ,have made it difficult to determine the data center .It has been proposed that a mathematical model of effective clustering based on tilt probability in the process of calculating clustering center ,introducing tilt data average probability calculation .When there is a greater consistency redundant interference in the data characteristics ,the probability assigned to the same cluster and constraints has been calculated ,and the probability constraints applied to data clustering similarity between the features , can effectively determine the similar characteristics of the initial clustering center .Experiments show that this model can reasonably selecting initial cluster centers ,improve the classification effect ,the mathemati-cal model of classification compared with k-means clustering model ,clustering result is more compact , stronger robustness .
-
-
计量
- 文章访问数: 885
- HTML全文浏览数: 726
- PDF下载数: 0
- 施引文献: 0
下载: