- Original Pdf: pdf
- Keywords: robustness certification, formal verification, robustness analysis, latent space interpolations
- TL;DR: We verify deterministic and probabilistic properties of neural networks using non-convex relaxations over visible transformations specified by generative models
- Abstract: Generative networks are promising models for specifying visual transformations. Unfortunately, certification of generative models is challenging as one needs to capture sufficient non-convexity so to produce precise bounds on the output. Existing verification methods either fail to scale to generative networks or do not capture enough non-convexity. In this work, we present a new verifier, called ApproxLine, that can certify non-trivial properties of generative networks. ApproxLine performs both deterministic and probabilistic abstract interpretation and captures infinite sets of outputs of generative networks. We show that ApproxLine can verify interesting interpolations in the network's latent space.
- Code: https://www.dropbox.com/s/np89rh2q8hzr1pj/approxline_submit.tar.gz?dl=0