This paper explores the possibility of using multiplicative random projection matrices for privacy preserving distributed data mining. It specifically considers the problem of computing statistical aggregates like the inner product matrix, correlation coefficient matrix, and Euclidean distance matrix from distributed privacy sensitive data possibly owned by multiple parties. This class of problems is directly related to many other data-mining problems such as clustering, principal component analysis, and classification. This paper makes primary contributions on two different grounds. First, it explores independent component analysis as a possible tool for breaching privacy in deterministic multiplicative perturbation-based models such as random orthogonal transformation and random rotation. Then, it proposes an approximate random projection-based technique to improve the level of privacy protection while still preserving certain statistical characteristics of the data. The paper presents extensive theoretical analysis and experimental results. Experiments demonstrate that the proposed technique is effective and can be successfully used for different types of privacy-preserving data mining applications.

Defending the safety of an open society from terrorism or other similar threats requires intelligent but careful ways to monitor different types of activities and transactions in the electronic media. Data mining techniques are playing an increasingly important role in sifting through large amount of data in search of useful patterns that might help us in securing our safety. Although the objective of this class of data mining applications is very well justified, they also open up the possibility of misusing personal information by malicious people with access to the sensitive data. This brings up the following question: Can we design data mining techniques that are sensitive to privacy? Several researchers are currently working on a class of data mining algorithms that work without directly accessing the sensitive data in their original form. This paper considers the problem of mining distributed data in a privacy-sensitive manner. It first points out the problems of some of the existing privacy-sensitive data mining techniques that make use of additive random noise to hide sensitive information. Next it briefly reviews some new approaches that make use of random projection matrices for computing statistical aggregates from sensitive data.