Private Learning with Public Feature Conditioning

Shuli Jiang, Walid Krichene, Nicolas Mayoraz

We study differentially private (DP) regression in settings where each data sample includes public, non-sensitive features -- common in applications such as recommendation and adve…