School of Computer Science and Technology, Donghua University, ShangHai, China.

---

Abstract—Data release is likely to result in privacy disclosure, so appropriate privacy protection measures are required for various data release technologies in order to ensure the privacy and safety of information, while differential privacy as a reliable model for privacy protection is extensively researched and applied. This paper presents the histogram data publishing solutions under differential privacy model, namely adding noise on the optimized histogram structure and then carrying out isotonic regression algorithms on the histogram privacy sequence. In this case, differentcial privacy model keeps all the statistical properties of the histogram unchanged and the concealment of privacy information, and in addition, histogram reconstruction and isotonic regression algorithm are effective in improving the accuracy of data release via histogram. This paper provides a solution about isotonic regression to decrease the error on histogram reconstruction based on previous research.

Index Terms—Data release, differential privacy, histogram, isotonic regression.

[PDF]

Cite: Xiujin Shi, Ling Zhou, "Difference Privacy Histogram Release Based on Isotonic Regression," Journal of Software vol. 11, no. 1, pp. 1-9, 2016.