EDDPC一种高效分布式密度中心聚类算法.PDFVIP

计算机研究与发展 ISSN 1000-1239/CN 11-1777/TP Journal of Computer Research and Development **(*):****-****, 201* EDDPC ：一种高效的分布式密度中心聚类算法 巩树凤 张岩峰 (东北大学信息科学与工程学院 沈阳 110819) (shidashufeng@163.com) EDDPC: An Efficient Distributed Density Peaks Clustering Algorithm Gong Shufeng and Zhang Yanfeng (Northeastern University College of Information Science and Engineering , Shenyang 110819) Abstract Clustering is a commonly used method for data relationship analytics in data mining. The clustering algorithm divides a set of objects into several groups (clusters), the data objects in the same group are more sim- ilar to each other than to those in other groups. Density peaks clustering is a recently proposed clustering algo- rithm published in Science magazine, which performs clustering in terms of each data object’s ρ value and δ val- ue. It exhibits its superiority over the other traditional clustering algorithms in interactivity, non-iterative process, and non-assumption on data distribution. However, computing each data object’s ρ and δ value requires to meas- 2 ure distance between any pair of objects with high computational cost of O(N ). This limits the practicability of this algorithm when clustering high-volume and high-dimensional data set. In order to improve efficiency and scalability, we propose an efficient distributed density peaks clustering algorithm, EDDPC, which leverages Vo- ronoi diagram and careful data replication/filtering to reduce huge amount of useless distance measurement cost and data shuffle cost. Our results show that our EDDPC algorithm can improve the performance significantly (up to 40x) comparing to naive MapReduce implementation. Key words density peaks; data clustering; Voronoi partition; MapReduce; big data 摘要 聚类分析是数据挖掘中经常用到的一种分析