Clustering method and device

Abstract

The invention provides a clustering method and device, and the method comprises the steps: carrying out the feature extraction of a to-be-clustered object, and obtaining a feature vector; according tothe number of to-be-clustered objects, determining repeated clustering times RT and an iteration times list {Ri} corresponding to a random graph clustering algorithm, the iteration times list {Ri} being composed of iteration times of each time of clustering; a threshold list {Ti} is determined according to the feature vectors, and the threshold list {Ti} is composed of thresholds for constructinga weighted graph during each time of clustering; and clustering the feature vectors according to the repeated clustering times RT, the iteration times list {Ri} and the threshold list {Ti} corresponding to the random graph clustering algorithm. According to the method, the number of iterations of each time of clustering can be determined according to the scale of the to-be-clustered object and the number of repeated clustering times, and for later several times of clustering with a small scale of the to-be-clustered object and a relatively stable clustering result, the small number of iterations can be used, so that the clustering efficiency is improved.

Description

technical field [0001] The invention relates to the technical field of machine learning, in particular to a clustering method and device. Background technique [0002] With the rapid development of machine learning technology and Internet technology, there are more and more scenarios that need to cluster high-dimensional data, such as clustering faces in image data, and clustering text in the field of natural language processing. data clustering etc. In the face of massive data in the Internet, how to realize fast and effective clustering of large-scale high-dimensional data has great practical value. [0003] At present, for high-dimensional data, such as face images and text data clustering methods, feature extraction algorithms are mainly used to extract eigenvectors (or eigenvalues) that can effectively represent data from high-dimensional data, and then through K-means Algorithm, K-means improved algorithm, grid-based clustering algorithm, clustering feature vectors. ...

AI Technical Summary

Problems solved by technology

In addition, according to different thresholds in the threshold list, the feature vectors are clustered multiple times, which can improve the instability of the clustering results caused by the random graph clustering algorithm, and is used to solve the poor clustering effect and clustering problems in the prior art. Technical issues with lower accuracy of class results

Images