Robustness Certification of Generative Models
This addresses the robustness verification problem for generative models, which is important for ensuring reliable AI systems in applications like image generation, but appears incremental as it builds on existing certification approaches.
The paper tackles the problem of certifying that all images generated along latent-space interpolation paths satisfy given properties, which is challenging due to the non-convex nature of these sets. The authors present ApproxLine, a scalable certification method that provides both deterministic and probabilistic guarantees for verifying specifications involving generative models and classifiers.
Generative neural networks can be used to specify continuous transformations between images via latent-space interpolation. However, certifying that all images captured by the resulting path in the image manifold satisfy a given property can be very challenging. This is because this set is highly non-convex, thwarting existing scalable robustness analysis methods, which are often based on convex relaxations. We present ApproxLine, a scalable certification method that successfully verifies non-trivial specifications involving generative models and classifiers. ApproxLine can provide both sound deterministic and probabilistic guarantees, by capturing either infinite non-convex sets of neural network activation vectors or distributions over such sets. We show that ApproxLine is practically useful and can verify interesting interpolations in the networks latent space.