摘要
针对聚类隐私保护算法中隐私开销随数据维度增加而提升的问题,提出了一种基于随机映射的隐私保护算法(RPPP)。该算法首先利用对称不确定性方法筛选相关特征,并通过独立同分布的高斯序列生成随机矩阵。为增强距离保持特性,随机矩阵经Gram-Schmidt正交化处理确保正交性,随后分解成多个独立子矩阵,对降维特征进行映射,生成特征匹配域和加噪干扰域。为进一步增强隐私保护性能,在干扰域中注入随机噪声。实验分析表明,RPPP在隐私攻击防御方面具有较强能力。通过Cancer和Diabetes数据集实验验证,结果显示RPPP在隐私保护性和聚类效率上均优于传统算法,与UPA、GCCG和AKA相比,聚类效率分别提升约16.34%、23.44%和32.94%。综合来看,RPPP算法在提升隐私保护性的同时显著提高了聚类效率,验证了其有效性。
To address the challenge of increasing privacy costs with rising data dimensions in clustering privacy protection algorithms,this paper proposed a random projection-based privacy preserving algorithm(RPPP).RPPP selected relevant features using the symmetrical uncertainty method and generated random matrices through independently and identically distributed Gaussian sequences.To strengthen distance-preserving properties,it applied Gram-Schmidt orthogonalization to ensure the orthogonality of the random matrices.These matrices were decomposed into multiple independent sub-matrices to map the reduced-dimensional features,and created a feature-matching domain and a noise-perturbed domain.To further enhance privacy protection,the algorithm injected random noise into the noise-perturbed domain.Experimental results demonstrate that RPPP effectively defends against privacy attacks.Tests conducted on the Cancer and Diabetes datasets show that RPPP outperforms traditional algorithms in both privacy protection and clustering efficiency.Specifically,RPPP improves clustering efficiency by approximately 16.34%,23.44%,and 32.94%compared with UPA,GCCG,and AKA algorithms,respectively.Overall,RPPP significantly enhances privacy protection while boosting clustering efficiency,confirming its effectiveness and practical applicability.
作者
何丽丽
张成林
曹明增
张磊
He Lili;Zhang Chenglin;Cao Mingzeng;Zhang Lei(School of Information and Electronic Technology,Jiamusi University,Jiamusi Heilongjiang 154007,China;Heilongjiang Provincial Key Laboratory of Autonomous Intelligence&Information Processing,Jiamusi University,Jiamusi Heilongjiang 154007,China;Jiamusi Key Laboratory of Satellite Navigation Technology&Equipment Engineering Technology,Jiamusi University,Jiamusi Heilongjiang 154007,China)
出处
《计算机应用研究》
北大核心
2025年第8期2511-2517,共7页
Application Research of Computers
基金
黑龙江省哲学社会科学研究规划资助项目(23GLD033)
黑龙江省自然科学基金联合引导项目(LH2021F054)
黑龙江省省属高等学校基本科研业务费优秀创新团队建设项目(2022-KYYWF-0654)
黑龙江省自主智能与信息处理重点实验室开放课题(ZZXC202302)
佳木斯大学国家基金培育项目(JMSUGPZR2022-014)
黑龙江省高等教育教学改革研究项目(SJGY20210873)。
关键词
高维数据
隐私保护
聚类
随机映射
K-MEANS
high-dimensional data
privacy protection
clustering
random projection
K-means
