Differential Privacy Query Processing Noise Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing differential privacy solutions face challenges in balancing data privacy security and query result accuracy, as improving privacy security often leads to reduced accuracy.
Innovation Solution
A query processing method and apparatus that allocate suitable noise power to each query in a batch, determining query types, sensitivity, and noise power based on privacy budget parameters to balance privacy protection and accuracy, using a Gaussian mechanism to add noise to query results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If noise power is increased to improve privacy security, then privacy protection is improved, but query result accuracy deteriorates
Solution Approach 1:
The patent applies local quality by differentiating noise power allocation based on query types and sensitivities. Different query types (counting, maximum, minimum, mean, variance) receive different noise power levels according to their specific sensitivity characteristics, rather than applying uniform noise power to all queries. This allows optimal balance between privacy protection and accuracy for each query type.
Solution Approach 2:
The patent changes the parameter of noise power dynamically based on query sensitivity and privacy budget. The noise power is calculated as a function of query sensitivity and allocated privacy budget, allowing the system to adjust noise levels according to the specific requirements of each query while maintaining overall privacy guarantees across the query batch.
2Device complexity
If uniform noise power is applied to all queries, then implementation is simplified, but query result accuracy deteriorates
Solution Approach 1:
The patent segments the query processing into distinct types (counting, maximum, minimum, mean, variance) and allocates noise power separately for each type. This segmentation allows the system to optimize noise power for each query type based on its sensitivity characteristics, improving overall accuracy while maintaining manageable complexity through systematic categorization.
Solution Approach 2:
The patent implements local quality by assigning different noise power levels to different query types based on their specific sensitivity requirements. Counting queries, maximum/minimum queries, mean queries, and variance queries each receive tailored noise power allocation, ensuring optimal accuracy for each query type while maintaining privacy protection.
3Measurement precision
If noise power is reduced to improve query result accuracy, then accuracy is improved, but privacy security deteriorates
Solution Approach 1:
The patent dynamically adjusts noise power parameters based on query sensitivity and privacy budget constraints. The noise power is calculated using a formula that incorporates both the sensitivity of the specific query and the allocated portion of the privacy budget, ensuring that accuracy is maximized while maintaining provable privacy guarantees for the entire batch of queries.
Solution Approach 2:
The patent introduces dynamics by making noise power allocation adaptive rather than static. The system calculates and assigns different noise power levels for different queries based on their sensitivity characteristics and the privacy budget distribution, allowing the system to optimize the balance between privacy and accuracy dynamically for each query type.
Data Source
AI summary
Implementations of the present specification provide a query processing method and apparatus. The method includes: first, determining query types of L queries to be performed on a target dataset, where the target dataset includes data of an object; next, determining query sensitivity of each query type of the query types for the target dataset; and then, determining, based on the query sensitivity corresponding to each query and a privacy budget parameter predetermined for a total set of the L queries, a noise power allocated to each query. Based on this, for a target query in the L queries, an actually returned result of the target query can be determined as an original query result of the target query added with a target noise sampled from target noise distribution of differential privacy, where the target noise distribution is determined based on the noise power allocated to the target query. As such, privacy of the target dataset can be protected.