An improved probability propagation algorithm for density peak clustering based on natural nearest neighborhood

Clustering by fast search and find of density peaks (DPC) (Since, 2014) has been proven to be a promising clustering approach that efficiently discovers the centers clusters finding peaks. The accuracy DPC depends on cutoff distance (dc), cluster number (k) selection clusters. Moreover, final allocation strategy is sensitive poor fault tolerance. shortcomings above make algorithm parameters only applicable for some specific datasets. To overcome limitations DPC, this paper presents an improved probability propagation peak based natural nearest neighborhood (DPC-PPNNN). By introducing idea propagation, DPC-PPNNN realizes nonparametric process makes more complex In experiments several datasets, shown outperform K-means DBSCAN.