Private Learning with Public Features

Walid Krichene, Nicolas Mayoraz, Steffen Rendle +3

We study a class of private learning problems in which the data is a join of private and public features. This is often the case in private personalization tasks such as recommenda…