Improved Kalman filter based differentially private streaming data release in cognitive computing.

Abstract

Cognitive computing works well based on volumes of data, which offers the guarantee of unlocking novel insights and data-driven decisions. Steaming data is a major component of aggregated data, and sharing these real-time aggregated statistics has gained a lot of benefits in decision analysis, such as traffic heat map and disease outbreaks. However, original streaming data sharing will bring users the risk of privacy disclosure. In this paper, differential privacy technology is introduced into cognitive system, and an improved Kalman filter based differentially private streaming data release scheme is proposed for privacy requirement of cognitive computing system. The feasibility of the proposed scheme has been demonstrated through analysis of the utility of sanitized data from four real datasets, and the experimental results show that the proposed scheme outperforms the Kalman filter-based method at the same level of privacy preserving.