2023/04/21 by Tang, Qiaoyue, Lécuyer, Mathias · 1 citation
#Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Machine Learning (cs.LG)
paper · doi:10.48550/arxiv.2304.11208
We observe that the traditional use of DP with the Adam optimizer introduces a bias in the second moment estimation, due to the addition of independent noise in the gradient computation. This bias leads to a different scaling for low variance parameter updates, that is inconsistent with the behavior of non-private Adam, and Adam's sign descent interpretation. Empirically, correcting the bias introduced by DP noise significantly improves the optimization performance of DP-Adam.