Google AI Blog: Deep Learning With Label Differential Privacy - U.S. Census  ## Metadata - Author: **U.S. Census** - Full Title: Google AI Blog: Deep Learning With Label Differential Privacy - Category: #articles - URL: https://ai.googleblog.com/2022/05/deep-learning-with-label-differential.html ## Highlights - The underlying assumption of DP is that changing a single user’s contribution to an algorithm should not significantly change its output distribution. - Tags: #ai #computing - DP algorithms include a privacy budget, ε, which quantifies the worst-case privacy loss for each user. Specifically, ε reflects how much the probability of any particular output of a DP algorithm can change if one replaces any example of the training set with an arbitrarily different one. So, a smaller ε corresponds to better privacy, as the algorithm is more indifferent to changes of a single example