Differential Privacy Algorithm for Multi-Query in Interactive Framework
In the differential privacy interactive framework,data sets usually need to answer multiple queries.With the gradu-al consumption of privacy budget,the risk of privacy disclosure is increased.Therefore,it is very important to save and track the consumption of privacy budget,which should not exceed the limit given by privacy budget.To solve the above problems,this paper designs a Multi-Query-Differential Privacy Mechanism(MQDPM).The idea of reusing noise is adopted.The same type of queries can reuse noise,save the cost of privacy budget,and improve the number of queries that can be supported by the dataset.Secondly,due to the contradiction between the availability and security of differential privacy,MQDPM uses the better availability Analysis Gaussian Mechanism(AGM)as the noise disturbance mechanism,and uses the Newton downhill method to replace the dichotomy iteration,which reduces the time complexity of AGM iteration.Finally,the blockchain is used to record the privacy budget,which is convenient to track the use of the privacy budget and ensure that it does not exceed the given limit.Experiments on the public IPUMS data set show that compared with the existing algorithms,MQDPM proposed in this paper not only reduces the query re-sponse time,but also effectively saves the privacy budget overhead under the same privacy budget limit,and has higher data avail-ability.
万方数据
维普
