Abstract:
In view of the memory limitation and irreversibility of data stream, a differential private histogram publication algorithm for data stream (ASDP-HPA) is proposed in the ...Show MoreMetadata
Abstract:
In view of the memory limitation and irreversibility of data stream, a differential private histogram publication algorithm for data stream (ASDP-HPA) is proposed in the paper, which accounts of the differential privacy model and adaptive sampling method, combined with the autoregressive integrated moving average model and dynamical sliding window technique. Compared with traditional method of noise injection for input dataset, the algorithm effectively saves the overall privacy budget, reduces the publication error and improves the usability of the streaming histogram. The results of experiment show that adoption of dynamical sliding window gains higher memory utilization. And various selection of privacy budgets validates that the algorithm is effective and feasible with higher accuracy.
Published in: 2018 5th IEEE International Conference on Cloud Computing and Intelligence Systems (CCIS)
Date of Conference: 23-25 November 2018
Date Added to IEEE Xplore: 14 April 2019
ISBN Information: