A Clustering Algorithm Based on Symmetric Neighborhood of Micro-clusters

Abstract

Clustering is an important task in data mining with numerous applications, including minefield detection, seismology, astronomy, etc. At present, the academic communities have introduced various clustering algorithms, and these methods have been widely applied to different fields according to their respective characteristics. In this paper, we propose a novel clustering algorithm based on symmetric neighborhood of micro-clusters in large database. Firstly we use k-means algorithm to produce micro-clusters which are introduced to compress the data, and then calculate both neighbors and reverse neighbors of micro-clusters to estimate their densities distribution, and gain the ultimate clustering result. The algorithm can discover arbitrary shape and different densities, and also it needs fewer input parameters than the existing clustering algorithms, such as, k-means algorithm. The efficiencies and effectiveness of the algorithm are validated through the test of IRIS testing dataset and synthetic dataset.