Text this: Detecting Generative Model Inversion Attacks for Protecting Intellectual Property of Deep Neural Networks