Generative Adversarial Privacy
This work addresses privacy concerns for data holders by introducing a novel adversarial approach, though it appears incremental as it builds on existing GAN concepts.
The paper tackles the problem of data privacy by proposing a data-driven framework called generative adversarial privacy (GAP), which formulates privacy as a constrained minimax game between a privatizer and an adversary, and it shows that GAP provides privacy guarantees against strong information-theoretic adversaries, with performance evaluated on the GENKI face database.
We present a data-driven framework called generative adversarial privacy (GAP). Inspired by recent advancements in generative adversarial networks (GANs), GAP allows the data holder to learn the privatization mechanism directly from the data. Under GAP, finding the optimal privacy mechanism is formulated as a constrained minimax game between a privatizer and an adversary. We show that for appropriately chosen adversarial loss functions, GAP provides privacy guarantees against strong information-theoretic adversaries. We also evaluate GAP's performance on the GENKI face database.