A Symmetry Histogram Publishing Method Based on Differential Privacy

The differential privacy histogram publishing method based on grouping cannot balance the reconstruction error and Laplace noise error, resulting in insufficient accuracy. To address this problem, we propose a symmetric DPHR (differential released). Firstly, algorithm uses exponential mechanism to sort counting of original bucket globally improve accuracy; secondly, an optimal dynamic programming global minimum which as evaluation function ordered histogram. This way, can achieve while balancing errors. Experiments show that effectively reduces cumulative between published under long-range queries satisfying ε-differential improves usability data.