基于证据理论的三向密度峰值聚类OA北大核心CSTPCD
THREE-WAY DENSITY PEAK CLUSTERING BASED ON EVIDENCE THEORY
为避免聚类标签错误传播,并且充分挖掘邻域信息,提出一种基于证据理论的三向密度峰聚类方法.在分配非分组点时考虑到K近邻的聚类信息,有利于提升聚类精度;用证据理论来描述和合并这些近邻信息,使建立的三向聚类模型能够将它们分配到最可能的聚类中,从而有效地避免了密度峰值聚类算法中错误标签的传播.在多个数据集上的实验对比结果表明,提出的方法能够有效避免聚类标签错误传播,并且实现了较高的聚类精度.
In order to avoid the error propagation of clustering labels and fully mine the neighborhood information,a three-way density peak clustering method based on evidence theory is proposed.The clustering information of k-nearest neighbor was considered when non grouping points were allocated,which was conducive to improving the clustering accuracy.The evidence theory was used to describe and combine these neighbor information,so that the established three-way clustering model could assign them to the most likely clusters,thus effectively avoiding the propagation of wrong labels in the peak density clustering algorithm.Experimental results on several data sets show that the proposed method can effectively avoid the error propagation of clustering labels and achieve high clustering accuracy.
赵乌吉斯古楞;凃云杰
呼伦贝尔学院计算机学院 内蒙古呼伦贝尔 021008
计算机与自动化
密度峰值聚类标签错误传播三向理论证据理论
Density peak clusteringLabel error propagationThree-way theoryEvidence theory
《计算机应用与软件》 2024 (005)
264-273 / 10
内蒙古自治区自然科学基金项目(2017MS(LH)0682);呼伦贝尔学院科学技术研究项目(2018JYYB06,2019KCZD01).
评论