Greedy Algorithm Based on Bucket Partitioning for Differentially Private Histogram Publication
Sun La · Journal of Xiamen University · 2013
Existed methods of differential privacy histogram publication cannot efficiently protect the privacy of data sets which have a large amount of small counts for different values of certain attribute.We proposed an efficient greedy algorithm to handle the differential privacy histogram publish problem of this particular data sets.The greedy algorithm adopted the merging strategies of closest neighbor barrel,and optimized the merging procedure using the red-black tree.Many simulation experiments were launched to compare the proposed algorithm with similar algorithms in terms of the availability of published data and the efficiency of algorithm.The simulation results verify that the greedy algorithm is efficient.