An Improved CURD Clustering Algorithm Based on Quotient Space

As the data size increases, the efficiency of algorithm and the clustering quality draw more attraction. CURD (clustering using references and density) is a fast clustering algorithm based on reference and density, which can discover clusters with arbitrary shape and has the linear times complexity. However, it still has some shortcomings such as: the efficiency to deal with the high-dimensional data is uncertain, the noise processing is not ideal, besides the number of the clustering results may not satisfy the requirement of the users. According to these deficiencies, this paper introduces a new method to propose the high-dimensional data with information entropy technology and quotient space theory. Additionally it disposes the noise date in two stages. Finally, some improvement are given on the step of sorting the reference points by quotient space theory to produce multi-level clustering results so as to meet the different needs of customers. Experiments show that the improved algorithm not only improves the quality of the clustering algorithm but also maintains the high efficiency.