Local core point clustering algorithm based on parallel natural neighbors

Abstract

The invention discloses a local core point clustering algorithm based on parallel natural neighbors. The local core point clustering algorithm comprises the following steps of S1, carrying out KD-treeconstruction on data set by adopting a rapid sorting method; S2, obtaining neighborhood information of each data object by adopting a parallel natural neighborhood search algorithm; S3, obtaining a local core point by calculating the density of each data object; S4, calculating the distance between the local core points; S5, constructing a decision diagram, and realizing clustering of local corepoints; S6, distributing non-local core points, and achieving clustering of the data set. Through the algorithm, the distance between the local core points based on the shared neighbors is defined, and the efficiency of the clustering algorithm is improved.

Description

technical field [0001] The invention relates to the technical field of data mining, in particular to a local core point clustering algorithm based on parallel natural neighbors. Background technique [0002] Cluster analysis is an important method of data mining. Its purpose is to divide data objects into different classes, so that objects in the same cluster are similar to each other, while objects in different clusters are different from each other. Cluster analysis is widely used in fields such as big data, pattern recognition, image processing and artificial intelligence. Therefore, the research on clustering analysis algorithm is of great significance. The existing clustering algorithms can be roughly divided into partition-based methods, density-based methods, hierarchical-based methods, grid-based methods, and model-based methods. [0003] In recent years, center-based clustering algorithms have gradually become a research hotspot. Partition-based methods, such as ...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the deficiency that the existing DP (Density Peaks) algorithm cannot be applied to complex manifold data sets, the present invention proposes a local core point clustering algorithm based on parallel natural neighbors, which uses local domain information to redefine the distance between local core points , can be better used for 3D point cloud data skeleton extraction

Images