DBSCAN Clustering Algorithm Based on Locality Sensitive Hashing

Abstract

The traditional DBSCAN clustering algorithm runs less efficiently on large data sets and high dimensional data sets. Aiming at the disadvantages of this, a core point selection algorithm and isolated points detection algorithm based on locality sensitive hashing is proposed. After that a traditional DBSCAN algorithm is run on the core points. Finally, the remaining points are assigned to the same category of the core points which both are in the same sub-cluster. The experiment results show that the proposed algorithm maintains a high correct rate on synthetic and real data sets and the efficiency is greatly improved.