Expansive entropy information bottleneck principle based clustering method

Abstract

The invention discloses an expansive entropy information bottleneck principle based clustering method, which mainly aims at the problem of unsupervised clustering in data mining and the problem of array set clustering when complicated relationship exists among clustered arrays but cannot be described by statistic probability. The method not only can indicate the relevance among the arrays, but also can indicate the correspondence among positions corresponding to the arrays, and can provide an objective cluster truncation rule so as to effectively avoid the defect of designating the clusteringnumber subjectively. The clustering method can be widely applied to clustering problems in the fields of medicine, intelligent transportation, mode identification and the like.

Description

technical field [0001] The invention relates to the field of data mining, in particular to cluster analysis. Background technique [0002] Cluster analysis is an important research content in the field of data mining, which aims to divide the data set into several classes, so that the differences within the class are small and the differences between classes are large. The distance between data is usually used to describe the similarity, the larger the distance, the smaller the similarity, and vice versa. Cluster analysis is widely used in pattern recognition, data analysis, image processing, intelligent transportation, medicine and other fields. Many effective clustering methods have been formed, such as K-means method, k-center method, self-organizing neural network, Bayesian neural network, Fisher clustering and so on. Different clustering methods have different starting points, which are mainly reflected in the measurement of the distance between samples or variables. ...

AI Technical Summary

Problems solved by technology

Since there is no prior target as a reference, this problem is an unsupervised clustering analysis problem

Images