Efficient and robust unsupervised possibility clustering algorithm

Abstract

The invention discloses an efficient and robust unsupervised possibility clustering algorithm. According to the algorithm, the number of common similar sample points is considered when the relation between two objects is confirmed; and the algorithm uses the following concepts: measurement of similar sample points, links, objective function and similarity, so as to maximize the objective functionto obtain an optimal clustering result, and finally, the sum of links between clusters is the smallest while the sum of links in the clusters is greatest. The efficient and robust unsupervised possibility clustering algorithm is scientific and reasonable in structure, is safe and convenient to use, can obtain a non-spherical clustering result, can well describe the data distribution while the algorithm complexity is also lower than that of a common K-means algorithm; and the clustering algorithm only considers the distance between points, so that the points do not need to be mapped into a vector space, and then the points with the density lower than that of the points around the clustering center are obtained, and meanwhile, the distance between the points and the clustering center is thenearest to that of other clustering centers.

Description

technical field [0001] The invention relates to the technical field of clustering algorithms, in particular to an efficient and robust unsupervised possibility clustering algorithm. Background technique [0002] Cluster analysis originated from taxonomy. In ancient taxonomy, people mainly rely on experience and professional knowledge to achieve classification, and rarely use mathematical tools for quantitative classification. With the development of human science and technology, the requirements for classification are increasing. As a result, sometimes it is difficult to accurately classify only by experience and professional knowledge, so people gradually introduce mathematical tools into taxonomy, forming numerical taxonomy, and then introducing multivariate analysis techniques into numerical taxonomy to form Cluster analysis, the content of cluster analysis is very rich, there are systematic clustering method, ordered sample clustering method, dynamic clustering method, f...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides an efficient and robust unsupervised possibility clustering algorithm, which can effectively solve the above-mentioned background technology. Since each point is assigned to the nearest cluster center, it cannot detect the data distribution of the aspheric category. For clustering of arbitrary shape distributions, a density threshold must be specified to remove data below this density threshold. noise point problem

Images