Research On Correlation Optimization Of Differential Privacy Regression Analysis Based On Laplace Mechanism

With the development of Internet,the amount of data has increased significantly in recent years.As a result,our daily life is filled with more and more valuable information.However,everycoin has two sides.With today's advanced network,some of our behaviors may result in the leakage of our personal information whichcontains great commercial value and has a wealth of return.But theinformation is also facing the danger of being stolen at any time.Nowadays,it has become our top priority to secure our personal information,and many studies about privacy protection are in progress.Privacy protection is a very effective way to protect personal information,which is to add noise into data so as to protect the real data from being obtained by attacker.In this way,we could greatly improve the security of our privacy.But the utility of the data will be reduced when the noise is too much.This paper is based on the linear regression analysis of predecessors who researched the differential privacy Laplace mechanism,solving the problem of excessive time complexity and excessive noise.As for these two problems,the method using space to save time is adopted to lower the time complexity.When noise needs to be added to the data,the data must be filteredfirstly,and then adding noise to reduce the amount of noise.Two characteristics of this method are described as follows: first,when noise needs to be added to the data,the data must be filtered and optimized,in this way,the amount of noise could be reduced.Second,to improve the security of data,the additional constant noise is no longer used.Sorting random noise and adding it to the sorted data so that the usability of the data can be improved.Since the adding data directly to the regression analysis will cause privacy disclosure,the method based on the differential privacy Laplacian mechanism is proposed to add noise and put the processed data into the regression analysis,which can greatly increases the security of data.The proposed work is as follows:(1)the works about privacy protection have been investigatedto learn about the progress in differential privacy.The existing deficiencies are analyzed to make improvements.(2)In this paper,the related theories provide a foundation for the proposed method.(3)The mathematical model is set up with the improvements that are verified by experiments.