Clustering algorithm for quickly searching and finding density peak value for complex data

Abstract

The invention discloses a clustering algorithm for quickly searching and finding a density peak value for complex data. The method includes: defining the eccentricity of the sample through the distance between the sample and the k-nearest neighbor sample based on the local density definition mode of the accumulated neighborhood, determining the local density of the sample through the comparison ofthe eccentricity of the sample and the k-nearest neighbor sample, and enabling the local information of the sample to be fully utilized through the local density; according to the micro-cluster merging and distributing strategy based on graphic degree connection, firstly, using a density peak clustering and distributing strategyto divide samples into a plurality of micro-clusters, then calculating the weighted proximity between the samples, determining the similarity degree between the micro-clusters, merging the micro-clusters with the highest similarity degree in sequence, and forming a final cluste. Experimental results show that the density peak point can be correctly found in a multi-scale, cross-wound and flow pattern complex data set, other samples can be correctly distributed, andthe clustering precision is greatly improved.

Description

technical field [0001] The invention relates to a clustering algorithm in the field of big data analysis, in particular to a clustering algorithm for fast searching and finding density peaks for complex data. Background technique [0002] The density peak clustering algorithm (Clustering by fast search and find of density peaks, DPC) was proposed by Alex Rodriguez and Alessandro Laio in 2014, and the paper was published in Science. Because of its simple algorithm principle, high-efficiency operation, fast finding of density peak points (cluster centers) without iterative calculation of the objective function, and suitable for cluster analysis of large-scale data sets, it has attracted the attention of scholars since it was proposed. It has been widely used in image processing, community network discovery, gene sequence recombination, travel agency problems, etc., but it is difficult for the DPC algorithm to correctly find the density peak point when processing complex data s...

AI Technical Summary

Problems solved by technology

[0003] Purpose of the invention: for the problems existing in the prior art, the present invention provides a clustering algorithm (A Fast Clustering Algorithm for Searching and Finding Density Peaks for Complex Data, FCA-SFDPCD) for complex data-oriented fast searching and finding density peaks, To solve the inability to find the correct density peak point and the distribution method are prone to errors and joint problems when dealing with complex data sets of multi-scale, cross-winding and flow patterns

Images