A privacy-preserving k-means clustering method

Abstract

The invention relates to a privacy protection k-means clustering algorithm based on a negative database, comprising: converting each piece of data of X in a database that needs to be clustered into a binary string, and performing a K-hidden algorithm for each binary string Generate the corresponding negative database; randomly generate k different binary strings as the initial cluster center; calculate the Euclidean distance from the negative database to each cluster center for each negative database, and divide the negative database to the cluster with the smallest Euclidean distance Class center; for each cluster, recalculate the cluster center; repeat the iteration until the cluster center no longer changes. The invention can improve the clustering accuracy of the existing k-means algorithm based on the negative database, and the proposed Euclidean distance estimation method can also be used in other data mining algorithms to protect data privacy.

Description

technical field [0001] The invention belongs to the field of privacy protection and security, and in particular relates to a privacy-protected k-means clustering method formed by using a negative database to protect data privacy of a k-means algorithm. Background technique [0002] With the rapid development of computer technology, a large amount of data is produced, which also makes the data mining technology develop rapidly. Many data mining algorithms have been proposed, such as k-means algorithm, k-nearest neighbor algorithm and so on. These data mining algorithms do not protect data security during operation, which may cause users' private information to be leaked. As people pay more and more attention to private information, data mining for privacy protection becomes particularly important. [0003] The negative database (negative database, NDB) is inspired by the artificial immune system. Unlike the traditional database, the negative database stores information that...

AI Technical Summary

Problems solved by technology

NDB also supports distance metrics, but there is little work applying NDB to privacy-preserving data mining

Existing work has proved that negative databases can be used for clustering and classification to protect the privacy of the original data, but the existing privacy-preserving clustering algorithms are based on the Hamming distance, which is not applicable to most databases, and now Many clustering algorithms are based on Euclidean distance, therefore, it is necessary to propose a privacy-preserving clustering algorithm based on Euclidean distance on the negative database

Images