摘要
现有的隐私保护异常检测算法大多是基于距离的,针对垂直划分的数据库提出一种隐私保护密度异常检测算法VPPDBOM。该算法基于密度异常检测算法(DBOM)的思想实现异常检测,利用垂直划分数据库中对象的邻域是其局部邻域交集的子集的特征,提高了DBOM算法中对象邻域的计算效率。同时基于半诚实模型,应用安全多方计算技术的安全和协议、安全交集协议实现隐私保护。理论分析和实验结果表明,该算法既保护了隐私信息又保证了性能。
For existing privacy-preserving distributed outlier detection algorithms are mainly distance based,a density based outlier detection algorithm for privacy-preserving DBOM outlier detecting over vertically partitioned data are proposed.By using the feature that neighborhood is the subset of all local neighborhoods＇ intersection over vertically partitioned data,VPPDBOM improved the computational efficiency of neighborhoods in DBOM;meanwhile,under the semi-honest model,the algorithm used secure sum protocol and secure intersection protocol of secure multi-party computation for protecting privacy information.Theoretical analysis and experimental results show that algorithm maintains privacy of the data sets of each party and keeps communication and computation cost low.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第23期4960-4962,5023,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(60773049)
江苏大学高级人才启动基金项目(09JDG041)
关键词
隐私保护
分布式数据
基于密度
安全多方计算
异常检测
privacy-preserving
partitioned data
density based
secure multi-party computation
outlier detection
