华中科技大学学报（自然科学版）2025，Vol.53Issue(5)：143-149,7.DOI:10.13245/j.hust.250649
高效的云外包隐私保护K-means聚类研究
Research on efficient cloud outsourcing privacy-preserving K-means clustering
摘要
Abstract
To improve the clustering efficiency of the cloud-outsourced privacy-preserving K-means algorithm and calculate the ciphertext data from multiple users,a cloud-outsourced privacy-preserving K-means clustering scheme was proposed,which could efficiently calculate the multi-party ciphertext.First,the non-negative matrix factorization algorithm based on sparse constraints realized the low-dimensional representation of high-dimensional data,which effectively improved the clustering effect of K-means clustering algorithm under high-dimensional data.Then,the multi-key fully homomorphic encryption technology based on the common key was used to solve the problem with complex homomorphic operations in K-means clustering of cloud servers.Finally,with the help of the triangle inequality theorem,K-means clustering algorithm was optimized,which reduced redundant distance calculations and improved the efficiency of clustering.Experimental results show that the proposed scheme has great efficiency of clustering in dealing with high-dimensional data,and the accuracy of the proposed scheme is similar to that of the plaintext K-means clustering.关键词
K-means算法/多密钥全同态加密/云外包/隐私保护/高维数据Key words
K-means algorithm/multi-key fully homomorphic encryption/cloud outsourcing/privacy preserving/high-dimensional data分类
信息技术与安全科学引用本文复制引用
曹来成,靳娜维,冯涛,郭显..高效的云外包隐私保护K-means聚类研究[J].华中科技大学学报（自然科学版）,2025,53(5):143-149,7.基金项目
国家自然科学基金资助项目(61562059,62162039) （61562059,62162039）
甘肃省自然科学基金资助项目(20JR5RA467). （20JR5RA467）
