Provable Lipschitz Certification for Generative Models
This work addresses the need for provable robustness guarantees in generative models, offering an incremental improvement over prior methods for domain-specific applications in machine learning security.
The paper tackles the problem of certifying Lipschitz constants for generative models by introducing a scalable technique that uses layerwise convex approximations with zonotopes, achieving efficient and tight bounds on small networks and scaling to VAE and DCGAN architectures.
We present a scalable technique for upper bounding the Lipschitz constant of generative models. We relate this quantity to the maximal norm over the set of attainable vector-Jacobian products of a given generative model. We approximate this set by layerwise convex approximations using zonotopes. Our approach generalizes and improves upon prior work using zonotope transformers and we extend to Lipschitz estimation of neural networks with large output dimension. This provides efficient and tight bounds on small networks and can scale to generative models on VAE and DCGAN architectures.