Scaling Polyhedral Neural Network Verification on GPUs

Christoph Müller, François Serre, Gagandeep Singh, Markus Püschel, Martin Vechev

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Certifying the robustness of neural networks against adversarial attacks is critical to their reliable adoption in real-world systems including autonomous driving and medical diagnosis. Unfortunately, state-of-the-art verifiers either do not scale to larger networks or are too imprecise to prove robustness, which limits their practical adoption. In this work, we introduce GPUPoly, a scalable verifier that can prove the robustness of significantly larger deep neural networks than possible with prior work. The key insight behind GPUPoly is the design of custom, sound polyhedra algorithms for neural network verification on a GPU. Our algorithms leverage the available GPU parallelism and the inherent sparsity of the underlying verification task. GPUPoly scales to very large networks: for example, it can prove the robustness of a 1M neuron, 34-layer deep residual network in ≈ 22 seconds. We believe GPUPoly is a promising step towards the practical verification of large real-world networks.
Original languageEnglish (US)
Title of host publicationProceedings of 4th MLSys Conference
EditorsA. Smola, A. Dimakis, I. Stoica
Number of pages14
StatePublished - 2021


Dive into the research topics of 'Scaling Polyhedral Neural Network Verification on GPUs'. Together they form a unique fingerprint.

Cite this