Less is More: Revisiting the Gaussian Mechanism for Differential Privacy

arXiv:2306.02256v3 Announce Type: replace
Abstract: Differential privacy via output perturbation has been a de facto standard for releasing query or computation results on sensitive data. However, we identify that all existing Gaussian mechanisms suffer from the curse of full-rank covariance matrices. To lift this curse, we design a Rank-1 Singular Multivariate Gaussian (R1SMG) mechanism. It achieves DP on high dimension query results by perturbing the results with noise following a singular multivariate Gaussian distribution, whose covariance matrix is a randomly …

arxiv computation cs.cr data design differential privacy identify mechanism privacy query results sensitive sensitive data standard