Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
10226804 | Physica A: Statistical Mechanics and its Applications | 2019 | 22 Pages |
Abstract
The Density Peaks Clustering (DPC) algorithm, published in Science, is a novel density-based clustering approach. Gravitation-based Density Peaks Clustering (GDPC) algorithm, inherited the advantages of DPC, is an improved algorithm. GDPC is able to detect outliers and to find the number of clusters correctly. However, it still has some problems in: (1) processing some data sets of varying densities; (2) processing some data sets of irregular shapes. An improved density clustering algorithm, named as DPC-LG, is proposed to overcome some weakness of GDPC. It can be seen from experimental results that the DPC-LG algorithm is more feasible and effective, compared with AP, DPC and GDPC.
Related Topics
Physical Sciences and Engineering
Mathematics
Mathematical Physics
Authors
Jianhua Jiang, Yujun Chen, Dehao Hao, Keqin Li,