Differential Privacy in Privacy-Preserving Big Data and Learning: Challenge and Opportunity

Differential privacy (DP) has become the de facto standard of preservation due to its strong protection and sound mathematical foundation, which is widely adopted in different applications such as big data analysis, graph process, machine learning, deep federated learning. Although DP an active influential area, it not best remedy for all problems scenarios. Moreover, there are also some misunderstanding, misuse, great challenges specific applications. In this paper, we point out a series limits open corresponding research areas. Besides, offer potentially new insights avenues on combining differential with other effective dimension reduction techniques secure multiparty computing clearly define various models.