Under differential privacy, a GAN's generalization gap is bounded, and experimentally, Lipschitz regularization reduces both train-test gap and membership attack success.
Tenenbaum, William T
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Generalization in Generative Adversarial Networks: A Novel Perspective from Privacy Protection
Under differential privacy, a GAN's generalization gap is bounded, and experimentally, Lipschitz regularization reduces both train-test gap and membership attack success.