面向国土空间分区规划的空间异构差分隐私计算方法和系统
By employing a spatially heterogeneous approximate differential privacy computation method, the contradiction between privacy protection and data autocorrelation in territorial spatial zoning planning is resolved. This method achieves efficient computation within the (ε,δ)-differential privacy framework, ensuring both data privacy and application availability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING PLANNING & NATURAL RESOURCES INFORMATION CENT
- Filing Date
- 2026-05-09
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies for land spatial zoning planning, traditional differential privacy computing methods cannot effectively protect data privacy and disrupt the spatial autocorrelation structure of data, leading to privacy leaks and data application failures.
A spatially heterogeneous approximate differential privacy computation method is adopted. Through a spatial extension sensitivity model, a privacy leakage compensation mechanism, and a spatially correlated Gaussian noise mechanism, all defined in a (ε,δ)-differential privacy framework, efficient privacy computation of spatially correlated data is achieved.
With strict mathematical guarantees, data privacy is protected while maintaining the spatial autocorrelation structure of the data, supporting subsequent spatial analysis applications and solving the privacy-utility imbalance problem.
Smart Images
Figure CN122174276B_ABST