What can Discriminator do? Towards Box-free Ownership Verification of Generative Adversarial Network
This addresses the need for robust intellectual property protection in generative AI models, particularly for image synthesis tasks, offering a novel verification method that is more practical than prior black-box approaches.
The paper tackles the problem of illegal theft or leakage of trained Generative Adversarial Networks (GANs) by proposing a box-free ownership verification scheme that uses the discriminator to learn a hypersphere capturing the generator's unique distribution, enabling verification by checking outputs only without specific inputs, and demonstrates effectiveness across over 10 GAN architectures with immunity to input-based removal attacks.
In recent decades, Generative Adversarial Network (GAN) and its variants have achieved unprecedented success in image synthesis. However, well-trained GANs are under the threat of illegal steal or leakage. The prior studies on remote ownership verification assume a black-box setting where the defender can query the suspicious model with specific inputs, which we identify is not enough for generation tasks. To this end, in this paper, we propose a novel IP protection scheme for GANs where ownership verification can be done by checking outputs only, without choosing the inputs (i.e., box-free setting). Specifically, we make use of the unexploited potential of the discriminator to learn a hypersphere that captures the unique distribution learned by the paired generator. Extensive evaluations on two popular GAN tasks and more than 10 GAN architectures demonstrate our proposed scheme to effectively verify the ownership. Our proposed scheme shown to be immune to popular input-based removal attacks and robust against other existing attacks. The source code and models are available at https://github.com/AbstractTeen/gan_ownership_verification