1 paper · 1 filter
Arun Ganesh, Ryan McKenna, Brendan McMahan +2
We initiate a study of algorithms for model training with user-level differential privacy (DP), where each example may be attributed to multiple users, which we call the multi-attr…