Retention-Replacement Probability Generation for Differential Privacy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data protection techniques apply uniform retention probabilities, potentially leading to unnecessary protection processing depending on the source database.
Innovation Solution
A retention-replacement probability generation device that includes a global optimal solution determining unit, a region generating unit, and an in-region optimal solution generating unit, which determines and generates optimal retention-replacement probabilities to achieve suitable perturbation while ensuring ε-differential privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If uniform retention probability is applied to all attribute values, then privacy protection is simplified and easier to implement, but unnecessary protection processing may be applied reducing data utility
Solution Approach 1:
The patent applies different retention probabilities to different attribute values based on their specific characteristics and sensitivity. Instead of using a uniform retention probability for all attribute values, the system calculates and applies customized retention probabilities tailored to each attribute value's privacy risk profile and importance, thereby avoiding unnecessary protection processing while maintaining adequate privacy safeguards.
2Reliability
If retention-replacement perturbation is applied to protect privacy, then individual data privacy is preserved, but analysis precision may be degraded due to data distortion
Solution Approach 1:
The patent dynamically adjusts retention probabilities as key parameters to optimize the balance between privacy protection and analysis precision. By calculating retention probabilities based on attribute value characteristics, sensitivity levels, and query workloads, the system adapts the perturbation intensity to minimize data distortion while maintaining privacy guarantees through mathematical bounds on information loss.
3Loss of information
If high retention probability is used to preserve data utility, then analysis precision is improved, but privacy protection effectiveness is reduced
Solution Approach 1:
The patent applies differentiated retention probabilities where high-risk attribute values receive higher replacement probabilities (lower retention) and low-risk attribute values receive lower replacement probabilities (higher retention). This localized approach to probability assignment ensures that privacy protection is strengthened where needed while preserving data utility where the risk is minimal, thereby resolving the contradiction between privacy protection and data utility.
Data Source
AI summary
Provided is a retention-replacement probability generation device that is capable of generating retention-replacement probability that realizes retention-replacement perturbation of a suitable level. Included are: a global optimal solution determining unit that, outputs a global optimal solution in a case where a global optimal solution exists that is a replacement probability of the attribute values in which the transition matrix P and histogram vector expression v of the attribute values yield ∥Pv−v∥=0; a region generating unit that, in a case where the global optimal solution does not exist, generates a region that is defined by an inequality equivalent to conditions for both replacement probabilities corresponding to i'th and j'th attribute values satisfying ε-differential privacy, and an inequality equivalent to conditions for the replacement probability of one and the retention probability of the other corresponding to the i'th and the j'th attribute values satisfying ε-differential privacy.