Privacy-Utility via GAN
Generating shareable representation of data via adversarial nets
The urge for offering personalized products and services, and the need for acquiring new users fuel the practice of data sharing among companies. At the same time, data sharing faces the headwind of new laws emphasizing users’ privacy in data. Under the premise that sharing of data occurs from a giver to a recipient, we propose an approach to generation of representational-data for sharing, which achieves value-addition for the recipient’s tasks, while preserving privacy of users.