Artificial intelligence

Abstract: The paper is devoted to the problem of development of the clustering methods, which are robust to initialization (number of clusters and initial cluster parameters), to the different cluster volumes, to the outliers. It is proposed a method for estimation of cluster structure and clustering of data, based on the evaluation of similarity measure between data objects in multidimensional space. The proposed method is robust to initialization of clustering parameters, to outliers and allows definition of cluster structure and number of clusters in the data self-organizing process.

Keywords:

References: