A chaos-based multiplicative perturbation scheme for privacy preserving data mining

The multiplicative perturbation is a popular scheme for privacy preserving data mining. It transforms the original data with the projection matrix. The security of projection matrix is a main concern in the multiplicative perturbation scheme. In this paper, we propose a novel multiplicative perturbation scheme which has a large key space. And we utilize the special property of chaotic systems, i.e., sensitivity to the initial condition and parameter, to design a new projection matrix generation algorithm. The experiment results show that the proposed scheme can preserve the privacy and maintain the utility for data miming.